Baltic Classifieds Stock

Good morning, everyone. Despite a challenging market environment, BCG once again demonstrated the resilience of its business model. We delivered a solid financial results, with both revenue and EBITDA growing by 7% while maintaining our industry-leading EBITDA margin of 78%. The significant headwind came from Auto24 business line, where the introduction of the new car tax in Estonia reduced listing volumes.

Excluding this one-off impact, the rest of the group delivered double-digit growth. Looking across our verticals, real estate was our strongest performer, growing by impressive 17% Jobs and Services also had an excellent year, accelerating from 7% growth in the first half of the year to 11% in the second half of the year, resulting 9% growth for the full year. Automotive remained stable despite significant volume headwinds. Meanwhile, our generalist marketplace continued to deliver steady growth of 3%. Given our exceptionally strong balance sheet and attractive market valuations, we significantly accelerated our share buyback program. By mid-June 2026, the company has repurchased over 10% of its issued share capital. Finally, I am pleased to announce that the Board has proposed a final dividend of EUR 0.031 per share, representing a 19% increase compared to a year ago, subject to shareholder approval at AGM.

One of the key indicators we closely monitor is our leadership position related to our nearest competitors. I'm pleased to report that our leadership remains exceptionally strong across all major marketplaces with traffic levels ranging from 5 to more than 60x those of our closest competitors. Overall, our traffic mix has remained very stable. The majority of the visitors continue to access our marketplace directly, which reflects the strength of our brands and loyalty of our users. The traffic from Gen AI platforms remain neglectable. While this channel is growing, it has primarily shifted traffic away from traditional search engines while the direct traffic continues to grow, I think this is a very important thing to highlight.

Our people continue to be one of the BCG's greatest strengths. In our latest employee engagement survey, more than 95% of employees said they are proud to be part of BCG. This level of engagement is reflected in our exceptional average employee tenure of 8 years, which is remarkable for a technology company. We also remain committed to diversity and inclusion. Our workforce is well balanced with almost equal gender split, and women hold 50% of leadership positions. This places BCG among top performers within FTSE 250 of our gender diversity.

Finally, we continue to make a good progress on sustainability. Since 2022, we have reduced our CO2 emissions by 75%, supported by our continued transition to renewable energy, which now accounts for 88% of our total energy consumption.

Now I will hand over to Lina to speak about finance in more detail.

Thank you, Simkus. Good morning, everyone. I'll now take you through the financial results in detail, starting with the revenue performance. On the right side, you see total revenue information over the past 2 years by business line. And on the left, the classifieds revenue accounting for 91% of group's revenue and split by business line as well. The B2C revenue, that's business plan subscriptions and C2C customers using self-service, mainly individuals.

The group delivered revenue of EUR 88.5 million for the year, a 7% increase on the prior year. Growth was driven by continued monetization progress across our 4 classified revenue streams. And last spring, we implemented C2C pricing changes across all our major platforms, and these have contributed to performance throughout the entire year. And as in previous years, we introduced B2C pricing and packaging changes from September and October.

Arturas will cover the key drivers of growth in more detail later in the presentation. But in summary, real estate representing is 29% of group revenue was again our strongest performing business line with revenue growing 17% to EUR 26 million. B2C grew 20% of real estate and C2C grew 12%. Autos representing 36% of group revenue and was flat at EUR 31.5 million. Auto B2C grew 11%, but this was offset by a 9% decline in C2C revenue. Jobs and services generating 1/5 of the group revenue grew 9% to EUR 17.4 million, with both B2C and C2C each growing 9%. Generalist, being 15% from group revenue and predominantly C2C, grew 3% to EUR 13.6 million. In total, B2C revenue representing 54% of group revenue grew 13% and C2C representing 37% of group revenue grew 1%. The remaining 9% comprises advertising and ancillary revenues, which together were broadly flat at EUR 8.1 million.

I'm now turning to cost and profitability. People costs remain our largest operating expense, representing approximately 14% of group revenue and almost 65% of operating costs before depreciation and amortization at EUR 12.8 million. Programming development costs are within people costs and handled in-house. The 2% increase reflects 3 main factors: a growth in headcount. We ended the year with 163 full-time employees. That's 7 full-time employees more than a year ago. Annual salary reviews in line with Baltic wage inflation, approximately 10%. And these were significantly offset by performance share plan PSP costs decrease 2026 has EUR 0.3 million cost in relation to PSP, down from EUR 1.9 million in the prior year, thus reflecting performance below the PSP targets.

Marketing costs represent 1.5% of revenue, and the majority of group's traffic is direct. The search traffic is minimal. And this year, we had some targeting marketing expenditure, particularly across social media channels in the younger audiences. IT costs, which is third-party services costs continue to be 1% of revenue and other costs, predominantly administrative and data acquisition costs, 5% of revenue. Our total operating costs, excluding depreciation and amortization, were EUR 19.9 million, an increase of 8% on the prior year.

Maybe back to the previous slide, EBITDA grew in line with revenue, 7% to EUR 68.6 million. The EBITDA margin was maintained at 78%, unchanged from the prior year. Below EBITDA, depreciation and amortization decreased 24% to EUR 8.3 million. And the principal driver was a 26% reduction in amortization of acquired intangibles, reflecting the full amortization of customer relationship assets recognized on the 2019 and 2020 acquisitions. This reduction is the reason why operating profit of EUR 60.4 million grew faster than EBITDA at 13%. Operating profit adjusted for the acquired intangibles amortization grew in line with EBITDA.

Now moving to our cash generation, debt and leverage. Cash generated from operating activities grew by 5% to EUR 69.9 million and maintaining our cash conversion ratio at 99%, consistent with recent years and demonstrating the quality of our earnings. After income tax and net interest payments, net cash inflow from operating activities was EUR 60.4 million. Looking at the net debt bridge, we started the year in a near net cash position with net debt of EUR 4.4 million. And over the course of the year, we drew on new debt facilities to fund accelerated share buyback program. I will expand on the capital allocation more on a later slide. In January 2026, we refinanced our existing debt facilities -- the new arrangement with the bank comprises a EUR 125 million term loan facility, which may be drawn in tranches and a EUR 20 million revolving credit facility. At the same time, we repaid in full the EUR 15 million remaining under the previous facility. And by the end of financial year, EUR 73 million had been drawn under the new term loan.

Our operating cash flows, combined with partial drawings under the new debt facility funded share repurchases for cancellation totaling almost EUR 77 million. That's a payment amount and purchases of company shares to EBT for EUR 3.1 million alongside dividend payments of EUR 18.7 million during the year. We closed the year with net debt of EUR 46.2 million, representing leverage of 0.7x EBITDA, up from 0.1x at the prior year-end. And since April 2026, a further EUR 45 million has been drawn to continue the share buyback program, bringing the total dividend drawings under the new facility to 118 million as of the date of this announcement. The remaining term loan capacity is EUR 7 million with the full EUR 20 million revolving credit facility remaining undrawn.

In this slide, you see the consolidated profit and loss summary. The revenue, EBITDA, operating profit and adjusted operating profit have been explained earlier. And before I go to the rest of the lines, the only adjustment to our financial performance metrics is amortization of acquired intangibles with the deferred tax impact. Starting from net finance costs, it accounted to EUR 1.8 million, a reduction from 2.4 million in prior year. And although interest expense increased in the second half of 2026 following drawings under the new debt facilities, this was more than offset by a lower average debt balance during the first half of the year and also interest income earned on the cash balances. Profit before tax grew by 15% to EUR 58.6 million, and the effective tax rate increased from 12% to 13%, primarily as a result of corporate income tax rate in Lithuania rising from 15% to 16%. Income tax expense was EUR 7.7 million.

From 2026, the Lithuanian corporate income tax rate increases from 16% to 17% and following the repayment of historical intercompany funding, the group Estonian operations are now generating distributable profits. Now under the Estonian and Latvian tax regimes, profits are taxed only when distributed. The group continues to assess capital allocation opportunities, including reinvestment and M&A. And no decision has been made to distribute profit from Estonia or Latvian subsidiaries. But if we were to decide to do so in the foreseeable future, we would recognize an immediate one-off tax charge of around EUR 6 million on accumulated profits. And thereafter, profits generated in Estonia and Latvia would give rise to an annual deferred tax charge at the applicable rates of 22% and 20%, respectively, to the extent they are expected to be distributed.

Accounting profit for the year grew 14% and adjusted net income, the reference metric used to our capital allocation policy grew 7% to EUR 58.1 million. It adds back the post-tax impact of acquired intangible amortization and the associated deferred tax. And on a per share basis, adjusted basic EPS grew 9% to EUR 0.123 and basic EPS grew 16% to EUR 0.108. Both EPS measures grew faster than the net income growth, reflecting the reduction in the weighted average share count resulting from the share buyback and cancellation program.

I will now turn to our capital allocation policy. Now since IPO in 2021, our capital allocation policy has been to return materially all adjusted net income to shareholders, historically through dividends of around 1/3 and the balance through share buybacks and debt repayment. During the first half of 2026, we became net cash positive. In 2026, the Board concluded that the company's share price didn't reflect the underlying fundamentals or long-term prospects, and we considered market concerns regarding the long-term impact of AI to be materially more cautious than our own assessment and viewed recent trading headwinds as temporary. Accordingly, we introduced leverage to fund an accelerated share buyback program. The EUR 145 million debt facility secured in January provided the capacity to execute the strategy. By year-end, we had repurchased and canceled 7.6% of company's issued share capital, increasing to 10% by mid-June.

Now following shareholder approval in May, we intend to continue repurchasing shares subject to market conditions, available authority and the group's capital position. At the September AGM, we expect to seek authority to repurchase up to 15% of the company issued share capital. As always, these authorities represent maximum authority rather than intention to utilize it in full. Continuation of the accelerated share buyback program beyond the group's existing financing capacity would require additional debt financing. And the Board has not established fixed thresholds for either leverage or share price. Capital allocation decisions will continue to be based on information available at the time.

Turning to dividends. The Board has also adopted a progressive ordinary dividend policy. And under this policy, the ordinary dividend will go broadly in line with adjusted net income while preserving flexibility in the group's broader capital allocation framework. Accordingly, we are recommending a final ordinary dividend of EUR 0.028 per share, together with a special dividend of EUR 0.03 per share. Together, this maintains our distributions for 2026 at approximately 1/3 of adjusted net income, consistent with our previous policy during this transition year. And in total, the dividends in respect of financial year 2026 would amount to EUR 0.044 per share, which represents a 16% increase on the total dividend paid versus last year. And finally, we will continue to evaluate value-creating investment opportunities, including M&A and share buybacks while maintaining flexibility in how those opportunities are financed. As announced today, this includes the acquisition of Cenubanka.lv, strengthening our data capabilities in the Latvian real estate market.

This concludes the financial section of our presentation, and I will now hand over to Arturas.

Thank you, Lina. I will take over and will review our strategic progress across core business lines, dive into the KPIs and provide an overview of key product development. Real estate was our clear growth champion this year. Revenue delivered a 17% increase to reach EUR 26 million. In the B2C segment, the monthly number of brokers grew by 3% and the number of clients reached a record high of 5,300. It was primarily driven by small brokers transitioning from C2C customers to become B2C customers. At the time, B2C ARPU increased by 16% to now EUR 252. The improvement was supported by pricing and packaging changes implemented in autumn. Besides ARPU growth, these updates were designed to encourage customers to use a wider scope and try out a wider scope of our services. They were also underpinned by a data product update from the previously acquired Untu platform.

In the C2C segment, we achieved significant yield improvements. Revenue per listed ad grew by 26% to now EUR 80. These increases reflect the continued strategic shift to our premium, longer duration packages. They are now chosen by more than half of our customers. The shift partly -- this shift partly affected an 11% decline in the number of listed ads. The market is hot and properties do sell faster. Concurrently, active ads declined by 6%. Consequentially, transactions required fewer listing extensions that are part of this listed ads metric.

From a market perspective, activities strengthened across the region. This momentum was supported by lower interest rates and improving macro environment. Total transaction volumes increased by 5% over the past 12 months. Average apartment prices in both the capital cities also rose by 5%. Lithuania was the main driver of this regional activity. Residential transactions here surged by 12%, and this surge was partly due to anticipatory spending ahead of changes to the national pension system as in April, individuals were allowed to redeem part of their pension savings freely. Our market leadership remains as strong as ever. KV and City24 combined had a 16x lead against the #2 in Estonia, while in Lithuania, Aruodas leads the next competitor by a record 62x.

Our automotive business delivered a resilient performance. Revenue remained flat at EUR 31.5 million. While total growth was muted, this headline figure marks a clear divergence between B2C and C2C segments. In the B2C segment, the average number of dealers declined slightly by 2% from record levels to 3,600 dealers. The decrease was primarily driven by weakened market conditions in Estonia, which accounts for now slightly more than 1/4 of our auto business line. Conversely, B2C ARPU increased by 13%. This growth was driven by pricing and packaging changes implementing in Lithuania during the autumn of '24 and '25. We strategically postponed B2C pricing adjustment in Estonia to support our customers during a challenging period. However, the package update was introduced there in May 2026 and is already in place.

The C2C segment faced volume headwinds during the period. Listed ads declined by 25%, active ads similarly by 26%, despite this inventory pressure, yields improved substantially. Revenue per listed ad rose to 22% to EUR 41. These gains were driven by our April '25 and March '26 price changes. Yield growth was supported by an increased consumer preference for premium longer duration listing packages that we intend to upsell that also include our car history check service, adding up to an overall marketplace transparency.

Overall, performance was impacted by 2 primary external factors. First, the Estonian car tax created a tough year-on-year comparable due to the transaction surge prior to this introduction. And the transaction in the Estonian market dropped by 43% year-on-year. Second, the region experienced its coldest and longest winter in 30 years. The severe weather disrupted typical C2C activities during January and February. We provide more detailed monthly charts to illustrate both of these effects in the appendices of this presentation. As a result of these factors, combined car transactions across both markets declined by 11%. Meanwhile, the average car price continues to grow at -- moderately at 2%. Despite these challenges, we firmly maintain our leading market positions. Autoplius holds a 5x lead over its nearest competitor, while Auto24 holds 28x lead in their respective markets. Looking forward, it's encouraging to note that trading trends have improved since March 2026.

Our jobs and services business line delivered strong growth. Revenue increased by 9% to reach EUR 17.4 million. Growth was closely mirrored in both B2C segment jobs and C2C segment services revenue streams. In the B2C segment, jobs ARPU increased by 8% to EUR 496. This expansion was supported by targeted price changes. The total number of active employers grew by 1%. This increase reflects our continued ability to successfully penetrate the long-tail customer segment. The C2C or Services segment achieved a 12% increase in active listing. The momentum was driven by an expanding client base and robust user engagement. While overall volume was healthy, the yield per active ad in services declined slightly by 3% to EUR 26. This compression was primarily due to a shift in the mix of service providers on all platforms. Specifically, a high-yield building service provider segment experienced a very high demand for them reducing the need to advertise in this hot market supported by a boom in real estate.

The C2C service segment saw a 12% increase in active ads, driven by a growing client base and strong engagement. While volume was healthy, the yield per active ad and services declined slightly -- oh, sorry, I repeated myself. Underlying labor demand remains well supported by a resilient economy. The stability is reflected by strong average salary growth of 8%. Finally, CVbankas firmly maintains a 5x leadership position over its nearest competitor. This established market share ensures we remain the primary destination for job seekers and recruiters in Lithuania.

Our generalist business line, which serves as a defensive component of our portfolio, delivered revenue growth of 3% to reach EUR 13.6 million. We achieved a significant 23% yield improvement on Skelbiu, our largest generalist platform. Consequentially, revenue per listed ad increased to EUR 10. These gains were driven by strategic pricing changes of both value-added services and listing fees and increased consumer uptake of premium packages also supported it. These adjustments offset a 13% decline in volume of paid listed ads. The largest share of revenue on Skelbiu originates from vertical categories, such as autos, property services and jobs. Because of this structure, our own specialized vertical platforms act as its main competitors. Strategically, we are entirely comfortable with users shifting from general listings to our dedicated verticals.

Our specialized platforms to provide a superior user experience and unlock a higher monetization opportunity. Total inventory, including paid and free remained highly resilient through the period. Active ads declined by minus 2%, remaining very close to last year record levels. Crucially, our active ad counts encompass both paid and free advertisements from customers. This blended approach ensures our platforms remain the primary destination for organic traffic. It also serves as a powerful competitive moat built on unique content. Our generalist portals firmly maintain market leadership across their respective regions. Skelbiu stands as the fifth most visited website in Lithuania. It currently holds a commanding 24x lead over its nearest competitor. In the Estonian market, Osta maintained a clear 2x.

Moving on KPIs. To our product development, we continue to execute on our strategic aim of investing in fit-for-purpose technology. Our approach to AI focuses on practical tools that reduce user friction and enhance platform efficiency, ensuring our marketplaces remained a definitive starting point for the Baltic population. At the CVbankas, the job seeker onboarding experience was significantly streamlined through the integration of AI-powered CV creation tools. Candidates can now upload existing documents to the platform. The system automatically parses and populates the profile, removing the friction of starting a job search. This feature has seen a rapid adoption across the user base. 51% of all new CVs are now generated using this way.

As a reminder, CVbankas operates strictly as a closed ecosystem. Candidates maintain internal profiles on the platform rather than relying on external documents, CV documents. This structure results in a highly organized database covering a significant portion of the labor market. Consequentially, this asset provides us with a highly defensible and future-proof data mode. We also meaningfully improved job search functionality on CVbankas through AI-powered synonym matching. So job seekers no longer need to know the exact wording of a specific role title they're seeking. The optimized search engine now actually identifies -- automatically identifies and displays jobs with similar meaning. On Skelbiu, we launched AI-powered image moderation to enhance platform safety. The service automatically checks user uploaded images of prohibited content. Furthermore, it enabled our moderators to review and improve moderation parameters and patterns directly.

On the automotive segment, we introduced AI-driven automation to the listing process on both Autoplius and Auto24. The system analyzes vehicle images, external technical data and user descriptions to automatically generate listing details. It also automatically populates key technical attributes of the vehicle. The automation reduces manual input for sellers, increase the data accuracy and improve search relevance. In real estate, KV.ee, we introduced new service packages specifically for real estate developers. This initiative marks a shift away from shared broker plans, improving monetization while offering more targeted marketing and analytical tools for them. The update also groups related listings under their prospective real estate development.

The structural change follows the path of Autoplius, where we developed segment -- where developer segment led revenue growth in the past couple of years. On the product, we launched a new lead generation feature called Request a Viewing. The tool allows potential buyers to submit contact details and prefer viewing times directly through the platform. By removing the traditional barrier of a phone call, this feature increases total lead volume. Furthermore, it provides the marketplace with deeper insights into the user intent.

Strategically, we are building a comprehensive data layer across our online marketplace. This goal was furthered by our June 2026 acquisition of Cenubanka Business in Latvia. It's a leading Latvian real estate data and market analysis platform. Cenubanka aggregates property transaction data from the registries, listing information and market reports. It serves as a key business tool for brokers, appraisers, developers and financial institutions to assess property values in Latvia. Following our acquisition of Untu in Lithuania last year, Cenubanka strengthens our proprietary data set. It provides the technical foundation for advanced market intelligence features across our footprint. And finally, it provides a structured transactional data required to develop agent-based interfaces in the future.

And thank you, and I'm handing back to Simkus to guide you through the outlook.

Thank you, Arturas. The Baltic economies have experienced remarkable growth over the past 3 decades, driven by a strong export, healthy labor markets, increasing productivity and a vibrant technology sector. The region also has benefited from a strong public finance, solid credit profile, steadily rising purchasing power. These fundamentals continue to create attractive opportunities for both our customers and BCG. Looking ahead, we remain optimistic about the economic outlook. In particular, Lithuania, our largest market, where over 70% of the revenue is generated, continues to be one of the strongest performing economies in the European Union, providing a solid foundation for our future growth.

The group expects revenue growth of around 10% in 2027 with growth anticipated to be slower in the first half and faster in second half of the year. Real estate, auto and jobs are expected to be primary growth contributors, while generalist is expected to remain broadly flat. Revenue growth outlook reflects confidence in our product pipeline and pricing and packaging changes, but cautious on the inventory trend. We expect the full year margin to be in line with previous medium-term guidance of mid-70s. So thank you for listening, and now we are open for the questions.

Alastair Reid from Investec. Three for me. Firstly, could you just sort of talk a little bit more about some of your assumptions for the guidance? I mean, particularly in the auto segment, obviously, some sort of easier comps given the Estonian tax situation and also the weather sort of how much that kind of that comes sort of super normal growth that you might see there sort of contributing to the guidance for this year?

Secondly, can you just touch a bit more on sort of competition in both, I guess, Lithuanian autos and also in sort of generalist with Vinted and the like. How do you think about sort of marketing spends potentially going forward?

And then lastly, just on sort of data products. How are you sort of thinking in terms of your latest launches and rollout about how you manage the pace of that in the context of any sort of dilutive effect on margins?

So I will speak about the guidance and competition and Arturas can cover the data question. So on the guidance, we feel confident in what's within our own control. So this is a pricing and packaging. We already implemented the C2C pricing in spring, and we scheduled to implement a B2C pricing in autumn. We feel that underlying markets, especially in real estate, but also increasingly in automotive and also in jobs is well supportive for the pricing events. And we will target the yield expansion there in line with our previous practices. So -- and we are kind of -- we are -- we have a high expectations on the pricing events. That's why we are planning the second half growth to be higher than the first half because most of the pricing -- B2C pricing event will contribute to the second half. Where we feel more cautious are -- on the inventory.

And on the inventory, we need to speak separately automotive real estate. In automotive, we have 2 different directions. In Estonia, the recovery continues. And there, we have a positive inventory growth. So far, automotive business in Estonia recovered probably 70% to 80% of the expected recovery level. If we compare the number of transactions a year ago, so this number is around 40% month-over-month, but still around 30% below 2 years ago. We don't expect that to recover fully to the 100% because we were -- we think that part of this market will not recover, especially cheap cars, but our expectation is that it should reach around 85% to 90% of the previous market. So in the Estonia market, we still have around 15 percentage points to go in terms of the recovery.

In Lithuania, the dynamics is different. Lithuania and automotive market is performing very well. Last year, in terms of the transactions, it grew 8%. It's a big increase annually. So in Lithuania, we have a headwind -- inventory headwinds because Lithuanian economy is doing well, purchasing power is increasing. The time to sell a car is decreasing. So that's resulting in the inventory headwinds. But as explained earlier, this is a good timing, good moment for the pricing events, which are scheduled in auto.

In real estate, real estate market is doing very well, especially Lithuania, but also Latvia, Estonia. Lithuania -- in Lithuania, we had a record number of transactions last year. And this year, we expect even have a higher number. In such a hot market, naturally, there is a headwind in terms of inventory because it's -- the transactions happens very quickly. But also, we implemented in C2C pricing changes, and we significantly increased the penetration of the most -- of the premium most expensive package from roughly 20%, 30% to half of the all choices. And this also leads to a lower number of expansions. So that's why we are expecting the headwind in terms of the inventory, but still a very positive environment for the B2C pricing event coming in autumn.

And the jobs market has continued to do well. So this year, we expect in Lithuania have a salary growth around 8% average salary growth. So it's a good environment for the labor market. And the pricing changes are happening also in autumn here, but it's being implemented gradually month over the month over the next 12 months. So we are feeling optimistic and confident in all the 3 verticals, especially taking into account the pricing events coming in autumn. But where we have a cautiousness is in inventory. So that explains our guidance.

On the competition part, Arturas, would you like to cover the Autoplius, Autogidas dynamics? Or I can start, maybe you can give. So Autoplius has currently a 5x lead compared to Autogidas. Historically, that's one of the highest lead we ever had. When we IPO-ed, our lead was 3x. And 10 years ago, our lead was less than 2x. So the highest lead we had ever was 6x a year ago. During the last year, Autogidas was much more active in marketing, including the TV advertising. And well, arithmetically, it reduced our lead from 6x to 5x. But we are not too concerned on that because it does not impact our business, our fundamentals. And considering that Autogidas is now owned by the private equity -- local private equity, we also think that it will not -- this increased marketing expenditure will not continue forever.

Arturas, would you like to add?

We're talking more about the generalist platform competition issue was back there. So we're still probably repeating ourselves that the segments that are competitive, so that home goods and oriented to consumers makes up a small percentage of Skelbiu's overall revenue. What we're happy with is that we are maintaining the content on the platform that we have, which is the strategic aim and actually positioned us well. The traffic numbers are healthy. We're not losing anything and actually gaining in that sense. But there is some natural, let's say, limitations in the consumer segment on these smaller segments in terms of pricing. However, it never were our revenue driver.

On data products?

Yes. So on data and AI-related products, we're following the pathway, which is client needs based primarily. AI may be a means to get there rather than profit goal itself. That's how we view it. We're strengthening this data layer which happened before in Autos with the car history and finally our plan data pool. We're progressing the same way in properties with Untu in Lithuania and Cenubanka in Latvia and overall viewing slate. We're happy with the tech stack we have. We don't see it as a limitation. There's incremental improvements that are required, but it's always been the case. And also the team setup seems good in terms of the know-how and the qualifications they have. So no major changes or no -- not really much changes in the future that we foresee apart from incremental improvements. We aim to make this knowledge our core competence, so most likely we will not rely on a lot on third-party providers. I'm not talking here about the LLMs, but other service providers to fill the gaps.

Now on the cost side, majority -- we see that in the very near term, these costs will reach probably about EUR 1 million per year. And majority of that are the people costs, of which the majority are already baked in, meaning that we already have that team in place that makes up the cost. There is naturally some probably token costs involved into that, but we are designing everything this way that it doesn't blow our token budget or anything like that in any meaningful way unless there's direct and very clear ROI problem.

I wanted to just a few words also on data spend. So we feel that the data layer is really creating a strategic advantage of our platforms. To give you an example, the car history report we developed 2 years ago, 1.5 years ago. Now 30% of all the listings in Autoplius have the car history report. That's double the amount compared to a year ago. And it's really creating our competitive edge and competitive advantage and it's very hard to copy to replicate. And that's really a defensive mode for us.

Same happening and same strategy we have for the real estate, where we invested in Untu last year. This year, we acquired the Cenubanka, but we feel that this is kind of a right strategy to go and it will improve our market position. On the cost side about the data products, so we also build those to be profitable, probably not as high profitability as the marketplace products, but still probably 50% plus margin on the data products we sell.

Maybe before Kevin, that most recent Cenubanka acquisition, it's a small business, but it's not that money burning, it's a profitability business.

Will, your were at second.

It's Will Packer from BNP Paribas. A couple for me. Coming back on the outlook for the year. One thing that stands out is that you push C2C yield monetization by over 20% in the year gone perhaps understandable in the context of some of the inventory headwinds. Should we start to worry about the sustainability of that kind of increase? There's some sort of cautionary tales from across your European peers about pushing yield too hard and having ramifications. So just how you think about that question?

Secondly, could you help us think through what a realistic number for the buyback is this year? Is it a similar quantum to last year? Is it double? This is quite a long statement, but I have, kind of, no idea what buyback number to assume. So just some color there. And then lastly, the guidance on generalists sort of stands out for being quite weak, no growth. Could you just remind us what factors have impacted growth for the year ahead? And should they abate from FY '28?

Maybe I'll start about yields then we'll answer the share buybacks and then probably Arturas can cover the generalist. On the yields, yes, the yields for C2C grew around 20% last year. But even though kind of percentage-wise, it might look high. But basically, it was like adding some, I don't know, 5 years or 10 years to the listing. And when you are transacting such an expensive item as automotive or real estate, it doesn't really matters in the end.

So I think that whenever we think about the yields in C2C and the possible drop-off rates or possible -- we think that we are here kind of looking more or working with the time because the price sensitivity on the digital products are declining over the time, and people are more and more eager to spend for and pay for the digital products. And it's not the price itself, what's the issue, but more like people's expectations or people's mentality. So I think that every year, it's positive for us because people are more and more comfortable paying for digital products. And the fees they pay are really marginal to the asset values.

I could add maybe a couple of comments. The yield growth, well, it's probably the average of averages. So it masks a lot of details behind it. It doesn't mean that in every segment -- in every micro segment, the change was like that. We apply value-based pricing. So we adjust for price sensitivity always in the, for example, the cheaper segment. We monitor the drop-off rates, which are intact after the changes. And also, this average yield growth is very much affected by the package mix. So it doesn't mean that all prices grew by 26%. It's that we optimized the pricing structure so that clients themselves chose to pay more. That's proving to their sustainability of the company.

That's definitely important to flag that the penetration of our premium most expensive package increase almost doubled from 20% to 30% to half. And this actually also impact the yield growth. Lina, would you like to cover buybacks?

Yes. So the statement about capital allocation in the results statement is basically, let's say, 2 major messages. One is accelerated buyback. And the other is moving to dividends growing in line with adjusted net income to have the flexibility to use -- to allocate capital based on how we see most value -- and to answer simply, we don't have a target level of debt target level of leverage. We don't set a mechanical level as such. The pace and the scale of buybacks will depend on the market situation, also additional funding, whether -- and how much we can get alternative uses of capital as well.

So -- and currently, in the end of May, we asked shareholders for 10% authority to buy back PCG's issued share capital. And at the AGM, we intend to ask for 15% for the next year. So the Board is very much supportive of the share buybacks in an accelerated way. Based on the market conditions and all the rest that I already listed, we'll see where we get.

And the last one?

But maybe just to add that also one of -- very important, of course, is to keep the flexibility and also prudent leverage. So we'll be watching that as well.

And maybe just help us accelerating the buyback versus what base H2 full year, 3 months, 9 months, 1 month, just...

Not accelerating, just to clarify, not accelerating from this standpoint, but we accelerated already in 2026. So just to continue accelerated share buybacks because currently, we're in the market buying back based on the safe harbor limits. So this is already [ executed ].

And the last one was on the generalist competition.

So talking about the competition between our own ecosystem. So the setup is, as mentioned, is the way that generalist is sort of a moat and a feeder of customers to the vertical platforms, which are better monetized and the synergies works well in services, jobs, autos and real estate. So we're comfortable generally with clients moving to the more expensive platforms [ probably ] will develop.

Probably the last thing on generalist to mention that also -- generalists also compete in categories, let's say, growth categories with Vinted. And we are also addressing this issue. Currently, we are in the final stage of developing buy now functionality and Skelbiu. So that should strengthen the position.

Jess, I saw that you were...

Jessica from Peel Hunt. I just got a couple, please. The first one on -- you've given lots of color on the different verticals. But when we think about the overall numbers outside of generalists, which segments or verticals can we think about in terms of higher growth than the group guidance versus lower?

Second one is just on costs. If you look at the cost last year, one of the biggest drivers is added headcount. So is there anything baked in for this year that we should think about? What kind of growth rates we should think about for the overall cost? And is there anything we should factor in?

And just the final one, just in terms of what you've just said about generalists. When you look at generalists, you compete also with the likes of Vinted or some of the others. And to my understanding, you only charge on the high-value items. So are there any initiatives to grow in terms of -- to support the growth? And actually, could you actually charge for some of the items which you don't charge for right now?

So I will answer about the verticals, Lina will cover the costs and about the generalists Arturas can cover. So on the verticals, we think that real estate will be a growth champion again this year because of very supportive underlying market and then the pricing actions we are taking -- we took and we are going to take in autumn. Also, we are optimistic on jobs because jobs are -- have accelerated last year from 7% in H1 to 11% in H2. But the market is also very supportive because this year, the salaries predicted to grow 8% -- average salary predicted to grow 8% again, economy is doing well. So the labor market looks also very supportive for the job portal to grow. And automotive is also in a good position to grow, especially taking into account the lower comparables and the recovery of the automotive market in Estonia. So kind of all the 3 verticals, but probably real estate should stand out. On the costs?

On the costs, reported operating expense fluctuate in the recent years because of PSP mainly, which is performance driven. So it's also prudent to look at the cost before this performance share cost line. Like this year, the cost was EUR 0.3 million, but the cost is expected to increase next year to normalize in line with normalizing results. So if excluding the PSP costs, the operating cost line is expected to grow in line with historic numbers and roughly in line with revenue.

Regarding the generalist, maybe a bit more flavor of how the pricing works on Skelbiu. So we monetize the vertical categories for the 3 that services jobs, auto and property, that's one thing. For the remaining categories, we monetize business customers, including the semi-professional business customers who are selling something as a means to do business rather than the own things. So it doesn't matter on the value, it's more this type of structure. In order to reinitiate the growth in this generalist segment, however, we are well progressing on the transactional buy now functionality that should go live in the next year, and it would serve as a nice addition on the current listing revenue we have rather than replacing it.

Just this year...

It is already signed the deal, yes.

Andrew?

It's Andrew here from Barclays. I've got 3, if that's okay. First one is coming back on AI, sorry, and to, I guess, get a bit more context in terms of how you're thinking about new products into next year, and I guess, particularly areas and things like conversational search, tools for your dealers or agents, some of the things that your peers are working on would be helpful to understand where you guys are at on those.

Second one is to clarify where you're at on B2C pricing in Estonia autos. Did you say in the prepared remarks that you put something through in May? Apologies if I misunderstood that. It would be helpful to understand what went through and then how it feeds into your thinking in the autumn and how much tolerance there is for pricing in the dealers in Estonia.

And then third question is on M&A. I guess kind of curious if you've chosen to do a small deal in Latvia, which maybe hasn't been such a focus in the past. So how are you kind of thinking about M&A more broadly and also kind of views on the Latvian market within that?

So I can start with the AI-related initiatives. So repeating, let's say, we approach the question from the client needs-based way. We're just seeking for the places where there is friction and how AI can help solve it. So bits there and there are already introduced. On one of the big projects we're working on and also set to go in this year already started is there's many more for conversational AI assisted AI search, you name it in many different ways. But it's going to be a focus to get launched in the next year, and we are progressing there well.

Sort of we're set up the data processing pipelines so we can basically search not only by what's provided by the seller, but also rely what's visible in the photos, what's visible from the third-party data sources we also have from our data platforms and combine that to deliver the better results for the customer. So -- but the seeking is to firstly make this available within our systems. And if the situation changes somehow, we don't see right now from the Gen AI platforms themselves. So let's say, then the absolute majority of work is having these products in place rather than connecting them.

Actually Arturas [ Saur ] is the key guy in AI strategist. So he's a very, very good person to talk about it. But mainly also like a general thinking our, let's say, about the user experience and about the conversational search. So currently, we are in the testing mode in real estate. But we are not rushing. We are not kind of pushing it to launch as quickly as possible. For us, it's important that it functions well and the users are not disappointed with it. So we probably -- we will make the final testing only then we will launch that. But also the thinking is that the current solutions like filter-based search is just does a job very, very well. And we don't have very high expectations that once after launching the conversational search, but suddenly everyone will start using it massively.

It's a spectrum. Basically, you have filter-based search on one end and full chat interface on the other. We're pretty sure the evidence from the initiatives here shows that fully conversational within chat experience is just not what works for customers, either on the platform or either in the LLM tool. However, the filter-based search naturally has some improvements to be made to solve the blind spots. And so we're setting to take best of both worlds to try to improve what's working rather than recreate it.

On the B2C in Estonia, so we delayed the pricing event a year ago, not a year ago in autumn because it was like a peak of the really the most difficult period for the private dealers. And we did implement it in May this year. The scope of the pricing that was quite minor in this time because we still see that the dealers are price sensitive, and they are emerging from this crisis, but they are still thinking how not to increase their expenditure too much. So it will have maybe less impact on the revenue growth for Estonia, but Estonia overall is recovering well. Since January, it's growing double digit, and we are happy with the progress.

On the M&A part, so I think that there might be some other add-ons, let's say, like the one we had in Latvia. And because we think that actually with especially data products, we are strengthening our competitive moat. And when we do these acquisitions, we always think can we build internally, how long it will take and how long it will -- how much it would cost? Or is there an existing tool which we can plug in easily. So, so far, the last few acquisitions, which we did Untu and Cenubanka, which translates into Price Bank properties. So in our thinking, where it was a cheaper and definitely so much quicker to acquire a business and to plug in rather than to build from scratch. Just a reminder, in car history report business, we build it from scratch. In fact Arturas did it, led the project.

Justinas, can I just clarify on B2C autos in Estonia? Are you still planning to do an increase this autumn as well -- or was that in May, and now we should not be expecting anything till autumn '27?

That would be to close the pricing events if we were to implement in autumn. So no plans for additional changes.

Okay. So nothing September '25, small May '26, nothing September, October, September through to next year.

Yes. In autos Estonia. In Lithuania, autos is scheduled pricing event.

Marcus?

Marcus, JPMorgan. Can we actually follow up on Andrew's question on the AI topic. Obviously, we had now 9 months of discussions around this topic. It seems also what you said today that the focus is more to be better in-house, but to really bet on in-house solutions on your own tech team. You highlighted sort of like the traditional way works pretty well. Obviously, it's going to get improved, but in a nutshell, doesn't need much. I sit here and I see others buying off-the-shelf solutions from Anthropic, from OpenAI. And these are obviously big words. And I'm just sitting here and really struggling to understand what is the difference? Is the difference that you say okay, we are in a really niche markets. Our market shares are proportionately much higher, and therefore, we can afford plus we have a very strong tech team. What is sort of the argument really to play it a bit differently? That would be maybe the first question.

The second question is then on costs again. And you said on the cost base broadly flat. Obviously, we have the guidance. But what should we think about personnel costs, including inflation? You heard for the market, it's what, around 8% or so. I bet it's maybe at that level for you as well in terms of inflation and you're hiring as well. So what is sort of like a realistic number in terms of personnel costs going up?

Arturas?

Yes, I can start. Regarding the probably a bit of clarification needed on buying the third-party solutions here. We're not building our own LLMs or anything like that and replacing what they can offer with the tools. We're using quite substantially their solutions in our process. It's just that there's no external specialist or external third-party company doing it. For example, where we to develop the agent, we want to have the know-how. This is a for upcoming in the upcoming years to have it within the team rather than relying on the service provider. I wouldn't say there's -- in terms of partnership with the LLMs, I wouldn't see that we're doing something differently in this sense.

Is it if, for example, the question then comes because what we hear you seem to be pretty relaxed about token costs because your infrastructure allows it, but you have to like a very sophisticated search, better ways to make it faster by just accessing your data very, very quickly. And therefore, you don't have this token cost problem. By the sounds of it, that's sort of like where I'm trying to get to.

So in the context of the marketplace, we see that the core token usage is sort of once are listing to process it and then to prepare it to be accessed later. So it's more or less fixed thing that we can control. And then there's just different ways how you develop -- how you use the tokens during the search. And we see pathways that are actually token conscious if we talk about the search because it's the most adopted thing and has probably the most potential, let's say, risk to -- but at the same time, there's the technology, well, it's hard to predict which way it's going to go, but we see, let's say, opportunities of these relevant capabilities actually coming to the user devices as well as introduced in Google and Apple events in the past couple of months where I think if some tokens you can take data across yourself, you can use some of the, let's say, user devices. It's just one of the possibilities there, but there are numerous methods to put.

Also, I would add, I think there is a back-end solutions, let's say -- and I think that we have progressed quite well here. Just an example, let's say, CV creation with AI help. We just uploaded the PDF and we created CV moderation, many other areas. So I think that we are quite advanced and we are not lagging behind. On the other side, the front-end facing solutions, let's say, a conversational search. I think it's -- at the moment, it still feels overstated or in fact, when we speak to the peers who have already launched those, they are not working so well, but the usage of those are very tiny.

And we kind of initially expect that suddenly people will start looking for apartment with bright living room. But in fact, they just type I want an apartment in center, which is perfectly searched through the filters. So probably in this context, maybe we are slightly behind, but there are no good working solutions yet. And we see it as an advantage because we might see what others are doing and to learn from those. By the way, once we will launch this conversational search, it will be quite advanced because we will already have ability to look into the pictures, to understand those and it will not -- will be already quite enhanced. On the cost base and personnel costs, Lina, would you like to cover?

So maybe slightly repetitive. But in general, the people cost is the key part of our costs. All the developments are done in-house by the team. And also, again, it's important to look at the costs also separately. So PSP cost separately and then salaries separately. So again, if PSP costs are expected to normalize, which is going to be in terms of percentage growth higher, the people cost, the salary costs, the total other part of the people cost is expected to grow in line with historic trends. Just slightly the growth is slowing down a little bit because of the wage inflation growth moderating a little bit as close to 8% is expected for this ongoing year in terms of wage inflation in the Baltics. We see that happening. This is what brings the cost a little bit down, but it's -- to reiterate, it's a key part of the cost and expect it to continue growing in teens looking without the PSP costs.

Maybe just also to add quite a big part of our team is IT team. And historically, the IT salaries used to grow much quicker even than the market average. So it was not unusual to see the IT salaries to grow 15% annually. I think now the pressure for IT personnel or developers has been reduced.

Giles Thorne from Jefferies. Justinas, what are you looking for from Arturas as COO that you weren't getting from Simonas? Arturas, we're going to talk about you like you're not here. And with Simonas going, who's going to be the GM of Skelbiu now?

So we have a GM for Skelbiu already for 2 years. It's a different person. He's not with us. Well, it's Simonas was one of the first employees in the company and definitely contribute a lot to the success and it looks very sad that he's leaving. But Arturas himself, he's an IT entrepreneur 15 years ago, we acquired the business from him, which he developed, also programmed, sold, expanded, so multi skilled. And since then, Arturas -- he was managing Skelbiu for a year, then Autoplius for 10 years, Aruodas for a few years and then working as a Development Director, especially leading IT topics.

So I think that his new responsibility definitely will enhance and support our AI organizational transformation. It will definitely become a much higher priority. And I think that because he was already leading both for more than a year or almost 2 years. And being this -- having this entrepreneurship skills, I think he will be just an excellent person -- the best person to replace Simonas. And looking historically, he went through all the biggest, most important business divisions. He knows those intimately well. And that was part of our succession planning because we already kind of identified that Arturas 5 years ago, probably that in case of Simonas leaving Arturas is taking over.

Just to follow up on Skelbiu, my mistake around Simonas. Transactional, why is it taking until next year?

Not next year, this year.

So okay, next financial year, but this calendar year.

Yes. Yes. No.

No. No.

When are you going to launch Transactional Skelbiu?

We are working on it already for half a year. We would love to -- well, it's on the final stages, probably to be launched within a month.

For the rest of the summer.

David from Morgan Stanley. Just a quick one on the yield expansion. I want to kind of elaborate on how much of that is from pricing? How much is it from product, how much is it from maybe prominence maybe in FY '26? And also just a general guide going forward, how should we view the mix between pricing and products, etc.

So for the B2C customers, the biggest part would be pricing, but also added a very strong product update last autumn. So far we added car history reports and we added valuation tools for real estate companies. So while it's difficult to quantify exactly, but we can say both, but maybe a bigger part is price changes because last year already underlying market was doing very well. In C2C segment, probably, again, 2/3 was a price, but 1/3 also came from the upsell of the premium package, which grew from 20%, 30% to 50%.

Just adding a bit of flavor, the structure of how we're selling the products to our business customers are absolutely mainly bundles. It's not that you can -- it's possible -- it's not possible to buy a particular service off the shelf. You choose either good, better, best type of package. So that's why it's a bit more difficult to pinpoint was it the price or was it actually the product in a sense. On the separately bought value-added services, a prominent product, it's just worth noting that it's in single digits of our revenue. So in B2C, it's not what we rely on. So there's not much, let's say, should the reordering happen in the AI age. It's not affecting a lot.

Sean Kealy from Panmure Liberum. I've got a couple, if that's okay. I guess, first of all, you've seen us -- I think it's fair to say that a lot of public market investors are very focused on AI as risk reward, et cetera. But I'm interested in your take on how private market investors are thinking about the issue. I know you yourselves will look for opportunities. And I'm sure you're more on the pulse of how some of the private market operators are thinking than maybe some of the rest of us. So interested to get your take on that.

Arturas, a couple for you. Can we -- is it possible to help us disaggregate the impact of selling longer duration premium products within real estate from the ARPU increases? I just want to get a sense for how much of that uplift in yields has been driven by shift in product mix versus price? And then you referenced that new build has been a bit weaker in Lithuanian real estate than the broker market just in terms of advertising less. Are you able to give us any color on if there's any mix shifts there?

And then -- and apologies for this, another couple, if that's okay. Just on tax and the Estonian Automotive. You've got a EUR 6 million additional tax charge this year should you distribute profits. Should we -- how should we think about tax rate going forward? Is it fair to assume that the Estonian business' margin is equal with the group rate? Are there any offsets, et cetera? Just thinking about not necessarily this year, but outer years?

And then finally, just given the profiling on Estonian B2C in automotive, can you just remind us how big you're expecting Estonian Automotive to be as a percentage of automotive this year? Apologies for quite so many.

We are only 3 people and you have 5 questions. But probably I can start with private market perceptions about the marketplaces. I think that -- well, it's also quite obvious from our capital allocation policy. We think that public markets overstate the AI risk and AI impact and overreact. So this is why we're kind of accelerating our share buybacks. And this is not happening in private markets. I think private markets are much more confident. And usually, especially when the private equity firms invest in the marketplace, they do so much more research and probably I would dare to say that private equity firms are even more knowledgeable when we are investments and less their opinion -- less dependent on or less -- they are less overreacting. On the impact from the longer duration in real estate and pricing, Arturas explained, it's difficult to say exactly, but probably a 2/3 of pricing, 1/3 duration. That would be.

Just to understanding pretty much the absolute majority of the direct price increases were less than the yield growth. So what basically happened is shift from the middle package to the highest one, although the highest price didn't change that much. It depends on platform by platform. But it's a strategy that in the previous years, probably 4, 5 years ago, successfully worked north of -- and in property, we saw quite equal distribution between short and long packages. We're doing the same playbook, moving to the longest ones, making them very attractive sort of popcorn pricing type of thing. And also it's a better user experience -- extension it's not something you are willing to burn on. It's sort of under delivery of what you promised.

And there was a question regarding builders and brokers, probably a bit more correction to the understanding in this case. Actually, over the past year, the builder segment in Lithuania was performing very, very well. There's lots of inventory, at the same time, with lots of demand, so the off market, and it's actually considering the pension reform, the window it's not that long. It's not going to be forever. So everybody is trying to sell at the same time. So we actually did have revenue increase from builders in both terms of advertising and value-added services in terms of packages. So that's actually worked pretty well. And it's been for 3 years already a growth driver in the whole Lithuanian B2C segment. And the builder packages are what we're improving in Estonia right now.

And probably that would continue to be next year because the builders are also having very deep pockets. And when the new -- there are new developments happening, so we need to invest more into marketing also. We have high expectations for that segment. Lina, would you like to cover this question?

Yes, of course. So actually, answering to your question, I think the good reference is the revenue split by country. And currently, we generate 73% of revenue from Lithuania, 25% from Estonia and 2% from Latvia. And given the margins are quite similar, broadly similar, the revenue split could be used as the reference point.

And could you please repeat the question about Estonia Automotive B2C, what was the question there?

Yes. Just on -- so I think you said that you did a small B2C pricing round in May. There won't be one this autumn. Just when we're thinking about trying to reflect that through the Automotive division, can you remind us how big the current Estonian automotive business is as a percentage of the mix?

25% from the total automotive. Any other questions?

And we might have some conference call questions. So Adam, over to you, please.

[Operator Instructions] Nothing my side, so back to you.

So thank you once again for coming. We feel really very privileged every time when we come to London in July to have such a good weather, also Wimbledon and now it's a soccer tournament. So thank you.

Hello, everyone, and thank you for joining the Baltic Classifieds Group's 2026 Half Year Results Announcement. My name is Lucy, and I'll be coordinating your call today. [Operator Instructions]

It is now my pleasure to hand over to your host, Justinas Simkus, CEO, to begin. Please go ahead.

Good morning. Our performance in the first half of the year is close to updated guidance shared after AGM in September. Revenue grew by 7%, reaching nearly EUR 45 million. EBITDA also increased by 7%, exceeding EUR 35 million and delivering an industry-leading margin of 78% Cash conversion was close to 100%, and we moved into net cash positive position with a surplus of EUR 5 million. We remain committed to our capital allocation policy. Almost EUR 20 million was returned to shareholders through the dividend and share buybacks, and we voluntarily repaid EUR 10 million of debt. The Board has declared an interim dividend of EUR 0.013 per share, an 8% increase from the last year to be paid in January.

Aside from the tax affected Estonian Auto segment, we delivered double-digit revenue growth. Our core revenue streams performed well with B2C up 15% and C2C up 8%. This strong performance was supported in particularly by another outstanding year in our real estate business, making its second consecutive year of exceptional results. Our lead over close competitor, which we internally consider the most important KPI remains strong across all major portals ranging from 5 to over 4 to 5x depending on the portal. Our websites attracted an average of 58 million visits per month, the equivalent of entire Baltic population visiting our sites 10x each month. We introduced price changes for both B2C and C2C customers at the level similar to previous years with the exception of car dealers in Estonia. These changes supported yield growth across our business and positions BCG well for the continued progress as the full impact of the recent B2C price changes will only be seen in the next reporting period.

In the first half of the year, ARPU increased across all business lines, up 16% in real estate, 13% in auto and 5% in jobs. Yields per C2C listed ad also rose sharply, 27% in Real Estate, 29% in Auto, 26% in Generalist, while the yields for services C2C active ads remained broadly stable year-over-year. This strength of the Baltic's economies has led to faster selling times, which has reduced advertising inventory, especially compared to record levels seen a year ago. As previously communicated, auto transactions in Estonia have dropped by half due to tax changes adding further pressure on inventory. This trend is particularly visible in lower number of C2C listings, while B2C subscription numbers remain at record highs.

Now I will hand over to Lina to talk about financials in more details.

Thank you, Justinas. Good morning. Our revenue grew by 7% to EUR 44.8 million and the 91% of revenue comes from the core, which is listing fees from both contracted clients, B2C and individual self-service users, C2C. B2C revenue accounts for 52% of group revenue and grew by 13% and C2C representing 39% of group revenue grew by 3%. The remaining 9% of revenue comes from noncore revenue streams. Ancillary accounting for 5% of the revenue and primarily being derived from auto financial intermediation declined by 8%. And this half year, it was directly impacted by the decreased number of auto market transactions in Estonia. Display advertising revenue contributing 4% of group revenue remained broadly in line with last year.

The upper right, you see a donut chart showing revenue split by business line. And as you see, the verticals combined generate 85% of group revenue, with the remaining 15% coming from our Generalist, of which close to half is coming from vertical category services, real estate, jobs and auto. The real estate verticals combined continue to be the standout performer this half year, and the revenue grew 20%. Growth was mainly fueled by more B2C customers, a shift toward premium longer period packages and further yield improvement. Auto is our biggest business line was flat year-over-year because of the market headwinds that were explained by Justinas. We had less new listings, but the yield per listing has continued to improve. Jobs and Services grew 7%. And within the segment jobs, the B2C component grew 6% by attracting more companies and improving the yield and services, which is majority of the business line C2C and accounts for 1/5 of the business line grew 12% from more users, active ads and slightly improved yield, which is diluted by higher uptick of longer period packages.

Generalist revenue grew by 4% through mainly yield improvement. And Simonas will talk about each of the business line a bit later. But with regards to yields, this year, we followed a regular schedule of pricing events. At the start of the year, we implemented C2C pricing changes across all our major platforms, and these contributed to the performance throughout the entire half year. And as in previous years, we introduced B2C pricing and packaging changes in September and October. We did that across the real estate, jobs platforms and the auto platform in Lithuania. We have this time postponed the B2C price changes for auto in Estonia. And overall, pricing and packaging updates that were made will have a more pronounced impact in the second half of the year. And in the jobs business line because most of the contracts have 12 months duration. The impact of pricing changes are flowing through gradually over the course of the year.

Now in terms of the cost, over the half year, our team grew by 6% to on average 153 full-time employees. In addition to the team expansion, the personnel costs grew -- the growth was driven by annual salary reviews, reflecting the wage inflation trends what we have in the Baltics. Total investment in our people increased by 5%, reaching EUR 6.4 million and remains 14% from revenue approximately. But this growth was partially offset by the reduction in share-based payment expenses. As before, programming development costs are within people costs and handled in-house. That's in the salaries costs. The IT cost line reflects third-party services only, and these costs grew 13% and continue to account for 1% of the revenue.

Marketing costs this half year grew by 14%, driven by a few additional events and campaigns. It remains approximately 1% from revenue. And as a reminder, the majority of the group's traffic is organic with direct and paid search channels accounting for around 80% of total traffic. Paid search traffic is minimal. And as a portfolio of brands, we continue to leverage our own websites for advertising. We own Skelbiu.lt, which is ranked the sixth most visited website in Lithuania and is home to strong vertical categories, which drives high-quality traffic cost free. That's one of the reasons behind our strong EBITDA margin. And then other costs representing 5% of revenue grew by 17%, mainly due to growing data products-related data acquisition costs. In total operating costs before depreciation and amortization grew by 8%.

The bottom column chart shows depreciation and amortization split into 2. One part is from acquired intangibles and the other is from ongoing CapEx-related depreciation. In July 2024 and in January 2025, the intangible assets related to business client relationships acquired in 2020 with a 5-year useful life were fully amortized, and this represents the 40% decline in amortization from acquired intangibles. The ongoing CapEx-related depreciation is broadly in line with last year.

In terms of profitability, with a 7% increase in revenue and continued cost management, our EBITDA grew by 7% as well. There were no add-backs to our EBITDA, and we maintained our EBITDA margin of 78%. On the right-hand side, you see EBITDA to net cash bridge. We continue to be highly cash generative, maintaining a 99% cash conversion rate. Cash generated from operations grew 4% and the net cash inflow from operations increased by 7%, reaching EUR 30.6 million.

Adjustments to IFRS figures remain limited, that amortization of acquired intangibles and associated deferred tax impact. Adjusted operating profit continues to closely align with EBITDA and grew by 7% as well. Adjusted net income grew by 9% and adjusted basic EPS grew by 10%. As I mentioned before, we generated EUR 30.6 million in net cash from operations during the half year. We started the year with EUR 25 million gross debt, EUR 3.6 million in net debt and 0.1x leverage. During the half year, we paid the final 2025 dividend, that's EUR 12.5 million in total, spent an additional EUR 2.4 million to repurchase shares for future awards, repurchased and immediately canceled 1.8 million company shares for EUR 6.4 million, representing 0.4% of the issued share capital at the beginning of the year and voluntary repaid EUR 10 million of the outstanding gross debt. As a result, we ended the half year with a gross debt of EUR 15 million and the net cash position of EUR 5.1 million.

Our capital allocation priorities remain unchanged. We intend to continue returning 1/3 of adjusted net income each year by interim and final dividend split approximately 1/3 and 2/3, respectively. The interim dividend for the year '26 will be paid on the 23rd of January 2026 to members on the register on 12 December 2025. Dividends are paid -- declared and paid in euros. Shareholders can elect to have dividends paid in British pound sterling. We will continue considering value-creating M&A opportunities and all options for financing attractive acquisition remain open. We continue to be -- we could be debt-free by the end of the financial year. So shareholders can expect an update on capital allocation policy by the time of our full year results. And yes, we continue with our share buybacks on the market.

Thank you. I will now hand over to Simonas.

Hello, everyone. In the next 4 slides, I will walk you through our main KPIs for each business unit. Real estate market is very active. Number of transactions keeps growing for the last 1.5 years. It grew by 7% during the last 12 months and average price grew by 4%. Market sentiment among the agents and developers is very positive. Private sellers listed 7% less ads mainly because of the growing share of the longer duration packages, which translates into less extensions. Shorter selling time have led to 5% drop in the C2C inventory and the pricing adjustments introduced in the late spring and C2C customer base segmentation have resulted into 27% growth in yield per listed ad. The number of B2C customers increased by 4%. It is mainly migration from C2C, while the average revenue per broker grew by 16%. This growth was primarily driven by the annual pricing and packaging event.

We maintained very strong lead both in Lithuania and Estonia. Our platforms are respectively 48x and 16x bigger than competitors. As you already know, the automotive market in Estonia is experiencing difficulties. The number of transactions in Estonia dropped by 50%, while in Lithuania, it increased by 8%. Combined, the 2 markets resulted in a 14% decline overall. Despite the drop in transactions, the average car price increased by 3%, which suggests that dealers' margins did not change. The number of listed C2C ads dropped by 29%, again, primarily due to the headwind in Estonia and the higher adoption of longer duration packages. The number of active ads also declined as overall market inventory shrunk. As a result of our pricing actions in the spring, the average revenue per listed ad increased by 29%.

In the B2C segment, customer base is stable, while yield increased by 13%. The main reason for this growth is the higher adoption of premium packages driven by new products and pricing changes. And bottom left, you can see that our lead over closest competitor is very strong, 6x in Lithuania and 31x in Estonia.

Let's move to Jobs and Services. Jobs market stays active. The unemployment decreased by 0.5 percentage points to 7.1%. Average wage have grown significantly, increasing by 9% over the past year. The market remains supportive for our business with the companies continuing to invest in the recruitment. As shown in the bottom line chart in the bottom right chart, our customer base grew by 1% and the average revenue per customer increased by 5%. Yield growth was slightly diluted due to the bigger share of smaller customers in the customers' mix. C2C part of our jobs and services unit is represented by the Services segment. You can see the chart in the top right corner. The number of listings continues to grow, increasing by 11% over the past year. We are promoting longer duration packages that offer a discounted monthly rate, which has diluted the impact of our pricing actions. As a result, the yield has remained nearly the same as it was a year ago.

Our job board maintained strong leadership position with a lead of 5x over the closest competitor. And our biggest service vertical Paslaugos.lt facility is 2x bigger than the main competitor, which is service category of our own general Skelbiu.lt. As a reminder, I want to say that our biggest Generalist Skelbiu.lt is not a typical one. Approximately 70% of its revenue is derived from vertical categories, automotive, real estate, jobs and services. Therefore, Skelbiu competes with our own market-leading verticals. We strategically leverage Skelbiu.lt to strengthen our vertical platforms. We have cross-listing, which generates high-quality traffic to our verticals. We had 15% fewer paid listed ads compared to the last year. Please note that Skelbiu has both paid and free ads.

In the top right corner, you can see the number of active ads, which reflects the total amount of the content on the site, including both paid and free ads. Over the past year, the number of active ads remained flat. At the same time, we increased the yield for paid ads by 26%. This growth was driven by the price review and the price increase for the value-added service. Our lead of our closest competitor in Lithuania remains as strong as ever of 30x. And in Estonia, it's 2x. Traditionally, I have a couple of slides about product development at BCG. Over the past 6 months, we were focused on data products for business customers and AI-based tools. Starting with the real estate on the left-hand side of the slide.

At Aruodas, we introduced Property Price Compass. It is a tool for agents to assess the asking price of an apartment. We have integrated technology from the recently acquired Untu.lt platform and developed a product that extracts data on actual nearby transactions, connects it to the listing history and provides a competition overview with a typical selling times. The agent can then review the information and provide a pricing report for the vendor, which is backed with the real data. This update was the key one in the new agents packages.

At Untu.lt, agents now contact purchase leads via AI-assisted call tracking service. Service logs what has been spoken on the phone and suggest next actions. This improves the quality of the agent service and provides more visibility of what is actually happening post lead acquisition. At Autoplius, we have introduced Autopulsas, a market assessment tool for any car that allows user to monitor market dynamics for specific models as well as the broader categories such as the fuel type, year and more. This helps dealers to make informed decisions on their stock. The tool combines data collected from the users with information from the state registry. Also at Autoplius, we have introduced AI-assisted listing process. The system analyzes images and descriptions to automatically fill in key vehicle attributes. This speeds up listing creation, reduce errors, requires less effort from the seller and improves overall listing quality.

Let's move to jobs. We updated salary estimator. Over the years, we have accumulated a large database of job ads and CVs. We feed this data into AI model that estimates the most likely salary range for the given position and provides a forecast of the future salary trends. User can search across nearly 3,000 job positions. At Getapro, we launched AI Assistant. It's an AI chat tool that helps customers to define the most appropriate service for the job they describe. And at Skelbiu, we introduced AI-based buyer to seller message screening system to help prevent fraud. The system analyzes conversation patterns and user attributes and flags potentially suspicious users.

And now I would like to hand back to Justinas to finish our presentation.

All right. Thank you, Simonas. The Baltic countries continue to show a strong economic fundamentals. Unemployment is broadly in line with euro area and is expected to fall across all Baltic states in 2025, showing that labor market is active. Wage growth remains high, reflecting a long-term trend of rising prosperity and catching up with Western Europe. Inflation is still higher than the euro area, a pattern we have seen for many years. This is driven by a strong demand and structural factors, and it remains a positive signal for investors in the region.

Public sector debt in Baltics is very low, which highlights the region's solid fiscal positions. Looking ahead, all Baltic countries are expected to grow in 2025. Lithuania, our key market, the biggest market is forecasted to grow by 2.5%, well above the euro area average. Over the long term, the Baltics has been one of the fastest-growing regions in Europe, which provides a strong fundamentals for our business. Despite record inventory comparables and challenges in the Estonian auto market, we expect revenue growth for the second half of the year will be higher than H1 and will accelerate into double digits for financial year 2027. Real estate and auto are expected to lead this growth, Jobs and Services and Generalists are expected to grow at a more moderate pace.

We remain cautious on inventory trends. We intend to implement product improvement and price changes for C2C in spring and B2C in autumn. With lower revenue growth and continued investment into our product, some EBITDA margin compression is inevitable. But even with the investment into data and AI, our EBITDA margin is expected to continue in line in the mid-70s. We intend to continue to return meaningfully all our excess cash to shareholders in a timely manner, of which at least 1/3 will be through dividends. We could be debt-free by the end of financial year, so shareholders can expect an update on capital policy by the time of our full year results.

Thank you for the attention. And now we are ready to answer your questions.

[Operator Instructions]

The first question comes from William Packer of BNP Paribas.

I've got 3, please. So firstly, by my calculation, this is the third cut to revenue guidance on Estonian Autos. Initially, it was new news around the tax changes. Then the second element was there was this political pushback leading to a further drag on inventory. Could you outline what's made you incrementally more cautious for H2 '26 and FY '27? That's the first question. Secondly, could you help us think through the range of outcomes for FY '27 EBITDA margin? You have a business with excellent visibility. mid-70s is a pretty wide range for a classified. How much OpEx growth are you planning for FY '27 as things stand today? And so if you were to deliver low double-digit revenue growth, where would margin land? Is 76% a reasonable number? I realize the midpoint of guidance is 75%.

And then thirdly, the share price after today's move is back towards the IPO price. In the last few years, you've more than doubled the EBITDA. Why not immediately buy back more shares if you have confidence that this share price dislocation is ultimately transitory?

Will, thank you for the questions. I guess I will start from the easiest one from the last one you mentioned about the company valuation and share prices. And yes, it's -- now it's close to IPO price. And the business is more than double the size. It's twice bigger. And we are back to the market. We were not buying shares in the last 30 days because we were in a close period at the time. But since today, we are back on the market and buying our shares back.

On the second question, on the car tax, I will answer and Lina will answer on the margin expectation and on our guidance. So on the car tax, such tax changes happens very rarely, once in 5, once in 10 years. So your visibility on the recovery is always very low. When we initially talked in July, we saw the recovery happening every single -- every next month since January till June. Then the recovery stalled in summer because of the politicians starting to debate about the car tax and maybe it should be removed. Then after the debate was over, in October, we again saw the increase in the transactions. But November was again declined month-over-month. We have one certain thing and one uncertain thing. So a certain thing is that we think that the car tax -- the size of the tax and overall, the car tax itself, it will not change the market fundamentals.

We think that the citizens will have just exactly the same amount of cars. The cars will age, the cars will be broken. New cars will come to the market. People will want to upgrade those. Also cars is something of a prestige status. So transactions will come back to the normal levels. So we are certain about that. What we are uncertain when it happens and on which month it will happen. We know one thing that actually in November and December that we have a very, very high comparables. Actually, in our -- this presentation in the appendix, you can see the -- you can find the current market dynamics. And the recovery, I would say it's slower than we initially expected. But we really -- from today's perspective, we know that this recovery will come sooner rather than later. We just don't have a crystal ball, and we don't know on which month exactly it will happen.

So here on the slide, you can see the market trends. The blue line -- dark blue line shows the transactions this year. So you see that the trend is positive. The yellow line shows transactions last year and this pulled forward demand, what we had in the beginning of last year. So November, December, it will be very, very tough comparables. But the trend is positive. And for us, it's also very clear that once we are already in January, the number of transactions will be so much higher than last January. So we will come back to the growth phase. So hopefully, that's explained our thinking.

And Lina will give a bit more color on our guidance.

Will, in terms of your second question, we continue to invest in our products and in services and our operating expenses this half year grew mid-teens, if not mentioned people cost growth offset by long-term performance share plan cost adjustment. So slower revenue growth as we navigate inventory trends and dynamics in Estonian autos will mean that if operating expenses continue to grow mid-teens, we mathematically expect some slight EBITDA margin compression as a result of reduced operating leverage. So uncertainty is in the inventory trends.

Did that answer?

Thanks for the color. Yes, it does. So the framing is you're committed to mid-teens OpEx growth. And so where the revenue growth lands will determine the level of margin dilution in FY '27, very helpful. Just to come back on your initial comments, Justinas, around the buyback. Would you consider taking on some leverage to take advantage of the share price dislocation? Is that something which is potentially on the agenda?

At this point of time, I can only say that the Board is actively discussing that question and all options are on the table. But no decision is taken. And I guess that's where we commend where it ends.

I can add, Will, that in line with the authorities obtained at the AGM, we are now on the market buying back BCG shares, and we wish it was possible to do it immediately. So if you know how, let us know.

Our next question comes from Andrew Ross of Barclays.

So to follow up on Will's. And I guess to start on cost growth. Why are you sure that mid-teens growth is definitely enough? It would be helpful just to get a sense of the thinking behind that budgeting process and why there's definitely enough investment embedded in mid-teens. The second question is an extension to that. Can you give us an update on some of the back-end IT projects in the group? So my understanding is it's not all on a single tech stack. Is that an impediment in a world of AI? And where are you on cloud migration? And then the third question, again, an extension. Can you give us a glimpse in terms of your AI product road map in the next couple of years? And I'm thinking things like conversational search that some of your peers have in the market, it would be really helpful to get comfort that is also in your pipeline.

Thank you, Andrew. So I guess Lina will answer the cost part and Simonas will cover the infrastructure and AI products.

Maybe expanding a little bit on the operating expenses question. Well, we are investing -- well, 70% of the -- close to 70% of the cost is people cost. So this -- the number of people is growing. We are organically growing the team. It's mid-teens growth in that cost line. IT costs, third-party services are also growing mid-teens, and that's including the investment into additional hosting and servers and et cetera. That's including the AI part, organic growth, other costs mentioned, everything else needed in the business. So based on that, we also did estimation for the future, how the future investments into certain IT developments might look like. We don't envisage any significant changes in that. So we consider mid-teens being the optimal and good investment level, if that answered.

Yes, that's helpful.

Maybe a few words about the infrastructure and as you asked about the cloud migration. Just as a reminder, actually, our main platforms are run on our on-prem, on our own infrastructure, own servers. But nearly all of the AI or data-related products, we actually running from the so-called public cloud. But don't be misunderstood about the public. Actually, it's configured as the private cloud, but we are using providers as the AWS or Google GCP, this kind of service providers. And we are not -- we don't have a plan to migrate our core applications to the cloud, to the public cloud. They will -- we will keep it as it is. We have our own private cloud, our own infrastructure that works all good. There are no huge seasonalities, which would require elasticity of the [indiscernible] it's quite stable.

For the AI, yes, we are growing this part of the infrastructure, and it grows gradually, partly it depends on the actual usage, and we pay as we go, basically more requests, more processing power, the higher the cost. So that's our approach so far. But given that the situation is changing, we might need some different kind of resources, maybe next year, more GPUs or something. So this might change. But we don't anticipate that something really sharp will happen in the next year or 2. So that's, as I mentioned, our approach. And about the road map connected to AI and maybe in general data, AI, machine learning and all the data products. So basically, we see from 2 sides, from buyer side. Of course, we do see and we do -- we already have implemented some tools from the buyer side to make their journey easier, more effective to discover the content they were not able or it was difficult to discover previously. So it's a semantic search that we would like to understand not the word by word, but actually what do they mean by telling like I'm looking for the recreational house, what does it mean, right?

So the AI, large language models, they do help in this field, and we are working on that. That's the main thing, the smart search, let's call it Smart Search. And from the seller perspective, of course, the first thing is to reduce the friction of the listing, the inventory. As I mentioned, we made the first steps in Autoplius already. Basically, we extract the information from the pictures. So the next -- there could be the next step to make it easier to price the object you are selling, especially if it's a property of a car or even a small item, then if you are selling used iPhone, it's not so easy to define the price. Is it EUR 100 or maybe EUR 300. So there is a really high -- the big gap between the pricing decisions. So that's the key components, which could be developed utilizing AI.

I would like maybe to add here a bit. I think that many reports focus too many or too much on how much AI would cost for the business like us, ours. And in our opinion, the costs are not so significant kind of it's something evolution, our costs. But I think what we are missing or what we are not kind of focusing enough, how much all these AI tools making our proposition, our products better. Just imagine the thing we this is what described, let's say, we record a phone call between the vendor and the broker. We make a summary. Out of the summary, we make an action plan. We can put in the calendar to send a proposal to meet tomorrow at 2:00 to do this thing. So it's kind of AI tools actually implemented are making our proposition so much better and so much more available. So I think that this is -- and that's kind of a lack of focus from the many reports what we see.

Our next question comes from Giles Thorne of Jefferies.

First question was back on whether you're investing enough in the business. You've often held up your take rate as being behind peers, and that's grounds for you to grow revenue, but the recent events at Rightmove confirmed that this is the wrong framing. It's more important to consider how the utility of your vertical platforms compares to peers. So some comments on where you think you are behind or better on overall utility of your platforms would be helpful. Second question is on the, obviously, Agentic AI risk and how that could disrupt top-of-the-funnel discovery.

Can you talk about how important it is or how helpful it is in the face of this risk to have a large pool of private listings as you do and to operate both horizontal and vertical platforms. Some voice over to how that mitigates the risk would be very helpful here. And finally, back on -- well, not back on, we haven't spoken about it yet, but on Generalist and a long-running question, are you finally going to follow every single other generalist classified advertising platform in Europe into going fully transactional? Or are you going to stay firmly stuck in the past?

Thank you for the question. Could you a bit explain what do you mean by the utilities, so I understand your question correctly.

Understood. So across Europe, many platforms doing many things, and there are some platforms that are doing things, products and features that are brought to market sooner than others. And as outsiders, we sit here and we try and track it all and we listen to management teams talk about why they think their platforms are great places to do things. But truly, it's only really the insiders that truly have the understanding. So some commentary as to how you think maybe Autoplius is better than AutoTrader in the U.K. I think it's a much more sensible way of framing your growth outlook than just saying AutoTrader charges more money than Autoplius does. Hopefully, that's clearer.

Yes. That's clear. Well, I will answer that one, and I'll take the AI search and how -- why the C2C listings make it more defensible and then Simonas will answer a generalist. On the -- when we are -- I think it's a fair statement. When we are comparing monetization, we look how much revenue the agents or the dealers spend on our platform and then we compare it to international peers. Usually, we don't monetize the products which we provide extra. Let's say, it's a single bill usually, and that includes all the features we are offering. In my personal belief, I still think that 80%, 85% of the bill justification why the dealers or brokers pay so much is actually related to the leads, how much leads you are providing and how much you are better than your competitors and the alternative tools.

So I think that probably to put it in other words, if you even kind of strip out or discontinue some of the products probably still can maintain the same prices or -- and the products itself, I find it quite comparable in many different markets. I think that we, as players, we all speak the same language. We have workshops. We copy each other. And I don't see a very big difference between our propositions. And on our take rates, we are updating it, let's say, we are making these assessments once a year. The latest take rate analysis we did was spring last year. So in terms of the take rate, our take rates in automotive and real estate was at 3% and jobs plus at 4%.

On the C2C benefit and why actually C2C makes our business much more resilient. So to begin with, AI applications, first of all, cannot perform search on live listings. It's just -- it requires just too much computing. From a technical point of view, from today's perspective, it cannot do that. So in order for the AI application to work, so basically, the marketplace has to somehow plug in their database. It cannot just simply, for example, ChatGPT cannot just simply go and search on the web and give you a good result. It has to be plugged in. So I think in C2C, in this sense, is working and serving as a defensive layer because even, for example, C2C content usually is a single listed. It's a unique content.

So kind of -- in our case, we are -- we actually have control on half of the unique market, which does not exist anywhere. And if we would feel that AI application is somehow threatening our positions, we just simply would not integrate beer. Even number two, integrating beer will not help because the C2C listings is something what's unique. It's not available. And number two does not have it. So definitely, C2C market is something what is an extra defense layer in this context.

And on generalist transition to transaction, Simonas, maybe you want to talk.

Yes, I think the answer will be quite short. Of course, we are aware and monitoring all the trends in generalist segment overall that the move to our transactional model, which is quite common for -- especially post-COVID. It became very common to buy and to shop online. And we see this trend, and we believe that it is the right way to go into the right further development for our generalists as well. And as a reminder, we do have transactional generalist in Estonia for many, many years. It works. We know how to do it. And yes, the [indiscernible] development let's say.

It's maybe just kind of it's something what we think it's a good direction to move forward, and this is something we are working on.

And what type of time frame are we talking about Justinas? When will we see a fully transactional offering on Skelbiu?

Well, within half a year.

The next question comes from Sean Kealy of Panmure Liberum.

I've just got 2, if that's okay. First of all, listed paid ads on Skelbiu were down. I appreciate the total inventory is flat for the year. How much of that do you think is cannibalization by existing verticals by other players? Or do you think you potentially just push price too hard solely on this side? Secondly, you've also had a bit of a decline in C2C real estate ads. How much of that is movement into B2C? Or do you think there's something else moving there?

I guess, Simonas you can cover since you talked about it a bit.

Yes. We don't have like a 100% precise data. But definitely, part of the content was moved or is moved to the verticals. And we do it ourselves actually. We do intentionally promote the packages, including the verticals, right? Because we believe it's a better product at the end of the day, and it's -- for us, it's more lucrative. The percentage it's hard to name that I would say maybe around 50-50 could be the answer, but it's really hard to tell. We don't have the precise way to measure it.

I would maybe add, in most of the cases, our vertical is #1 in the market. And our generalist category is either #2 or #3. So usually, when the #1 is winning, so #2 inventory or listings are going to #1 rather than to #3. So it's kind of -- we know that majority of the decline is actually still remains in our ecosystem.

And the second question about the real estate C2C listings and how much of those are going to B2C.

I can cover it. Yes. Actually, it's partly on the slides because the B2C we have, of course, it's not -- it's in brokers, not in the ads, the listings. But 4% of the growth in the number of brokers, it's mostly from C2C. And average number of ads per broker that could be teens more or less.

Yes. Maybe, maybe just additional color. When we are looking at C2C minus 7% decline and B2C plus 4%, some of the listings move from here to here. Also, partly, that's decline of economy doing well and real estate selling quicker. So kind of in reality, the real decline is smaller and here, real growth is a bit also smaller. And we need to look kind of as a combination of both 2 segments.

And the third actually I mentioned on the longer duration packages, maybe need more color a bit because we count as a listed at either the newly listed property or something which was relisted or extended. So we count like transactions. And because we are selling longer duration packages, this means that there is a higher probability that it will be sold and not extended. So that's a bit shrinks the number of listed or/extended.

And that's very visible in the yield which trends in C2C.

I've got a follow-up if that's okay.

Yes, sure.

So if we take a step back and look at some of the headwinds in various parts of the business, you've had a couple of years of pretty exceptional growth. We've now got some headwinds in a couple of your different revenue lines and cost growth accelerating quite significantly year-on-year. If we look at change in free cash flow, do you think it's fair to say that this business in the next sort of, say, 3 to 5 years is going to struggle to break double-digit free cash flow growth?

Maybe Lina, you can take this question.

It should be viewed through the angle of EBITDA growth, and it very much depends on the inventory levels. You see BCG platforms, we are the market. And what's in our hands is the actions that we take in terms of the B2C and C2C pricing events. But in terms of the inventory and the volumes, this is the uncertain part. And as it reflects in the revenue consequently and mathematically, it goes through -- comes down to profitability, and that's the unknown, which cannot answer you straight away.

There are more unknowns. Now if we talk about the new products, which would bring extra revenue sources, we did one in -- okay, let's say, 2 years ago, 1.5 years ago with the car history product. Now it's a decent business, profitable, growing. And we see more opportunities and there might be even more opportunities occurring on the market, which we don't know. There are definitely some things to do in the property market. We made one step by acquiring Untu which we are selling leads and sharing the actual commission, brokers commission and quite a significant part of the broker commission is paid for us if the property is sold. And there are more kind of core product extensions, which could be quite lucrative. And we are thinking about that or chasing.

Our last question today comes from Marcus Diebel of JPMorgan.

I have also 3, if that's okay, maybe I'll take them one after one. The first one is on, again, the cost base. You're talking about mid-teens, mostly headcount. If you can just elaborate a bit more what exactly are those investments from here? I obviously seen the product slide, but what are the investments from here? And why is it a mid-teens number? Given obviously, AI also brings a lot of cost synergies. We understand software engineers are getting much cheaper. Maybe that's the first question.

Lina, do you want to take that?

Thanks, Marcus. This mid-teens comes from a combination of -- if you're talking about the people, investment into people, it's a combination of growth in the team. And we historically have single-digit growth in the number of -- in percentage terms in the number of people plus wage inflation in the Baltics, which is from high single digit to 10%. So a combination of those brings us to thinking about mid-teens.

But it's really the element that basically it's on new products, high cost inflation, which only gets partly offset by savings. Yes, that's the way to read it, I guess, because AI seems to be obviously a lot of cost savings as well, particularly on headcount because software engineers are more efficient. But yes, I hear you that's probably the right number then.

Yes. I mean...

And maybe -- yes.

Just to add, absolutely, many people are more efficient, including engineers and client support and moderators. Everyone is actually more engineer -- more efficient. But that does not reduce the headcount. Maybe that reduces the future new hires in these sectors. So basically, probably we will hire less of the client support and engineers when the business will continue to grow. But at the same time, we will hire more data scientists, infrastructure people. So kind of maybe the profile of the new hires are changing.

Yes. Perfect. That makes sense. Second question, again on AI. Just to be clear, sort of like how you want to address it. It seems that you still want to run the business as a walled garden, obviously, have an AI layer on top. You mentioned that the costs are not that high. Just to understand, there's no idea to really enter a partnership with one of the LLMs. I mean there's obviously the model that Zillow does, which is a stronger partnership rather than just using an AI layer. If you can just comment on this. And then maybe just related to this, I guess, is the question on M&A. I mean, clearly, asset prices and valuations in classifieds have changed a lot. What does it actually mean in terms of your appetite to do M&A, particularly when you're also sort of like highlighting potential for higher share buybacks? That's my last question.

So maybe I can take some of those. On AI, so they basically -- in our case, it wouldn't make any sense, let's say, making integration with the AI tools. It wouldn't make any sense. We think actually building our own semantic search, smart search, and we are working on that. Also, I think that we are building the kind of extra layers from the kind of scraping the data from our marketplaces, like registration walls, early registration walls. And kind of we think that probably in the future, there might be more competitors coming with AI approach. So we want to kind of be ready defending our core propositions, our core data. But yes, at this point, basically, it wouldn't make any sense kind of integrating with AI applications in our case.

And on the M&A, so we always -- whenever when we were considering M&A, we were weighting 2 options, either to buy a new company, new business or to buy our own stock. So from today's perspective, probably our appetite for new M&A is reducing because our own stock looks very attractive.

We have no further questions at this time. So I'd like to hand back to you, Justinas, for any closing remarks.

Thank you. Thank you for listening. Thank you for the questions, and talk to a majority of you in the coming few weeks.

Thank you.

This concludes today's call. Thank you all for joining. You may now disconnect your lines.