Match Group (MTCH) 2026 年第三季財報電話會議：Tinder DAU 成長與 2027 年展望

重點摘要

• Tinder 全球日活躍使用者 (DAU) 在 9 月恢復年增率成長，為 3.5 年來首次。美國地區的 DAU 也在 10 月底轉為正成長。
• Tinder 第三季月活躍使用者 (MAU) 年減 6%，相較第二季的 7% 降幅有所改善。付費人數下滑 4%，而第二季為下滑 5%；排除匯率影響的直接營收則減少 2%。
• 管理層預期 Tinder 2026 全年直接營收將呈低個位數衰退，優於 2 月時的預測。公司的目標是 2027 全年直接營收恢復成長、2027 年第四季前付費人數實現成長，並在 2027 年第四季底前實現 MAU 同期持平。
• Hinge 第三季營收成長約 19%，全球 MAU 成長 12%。Match Group 維持 Hinge 於 2027 年達到 10 億美元營收的目標。
• Match Group 今年迄今產生了 8.32 億美元的自由現金流，並投入 5.91 億美元用於股票回購、股利發放及員工股權獎勵的淨額結算。
• 管理層預計應用程式內購費用將對 2027 年調整後 EBITDA 獲利率造成 1 個百分點的逆風，其中包括額外 2,000 萬美元與 Apple 相關的成本。

關鍵財務數據

業務與營運表現

Tinder 轉型轉機擴大

Tinder 報告用戶成果已連續多個季度有所改善。第三季在「Sparks」（定義為雙方各發送三次訊息的六向對話）與「Sparks 覆蓋率」（收到 Spark 的用戶百分比）兩項指標上，均創下開始追蹤以來最強勁的年增表現。

此改善廣泛涵蓋各種性別、年齡和地區。女性 MAU 的年降幅收窄了 8 個百分點，季降幅收窄 2 個百分點。30 歲以下 MAU 的改善趨勢比 30 歲以上用戶更為明顯。

10 月份，佔 Tinder DAU 近 70% 以及營收 50% 的市場，其 DAU 錄得同期持平或年增長。然而，由於產品改善主要增強了現有用戶的參與度與留存率，MAU 的表現仍落後於 DAU。

管理層表示，下一階段取決於能否讓舊用戶與潛在用戶重新考慮使用 Tinder。其「Better with Friends」路線圖包括 DoubleDate、Groups 和 Events 功能。DoubleDate 已創造 3,000 萬次配對。在挪威，18 至 22 歲的 Tinder 用戶中有 15% 加入了社群群組，且女性按讚群組個人資料的機率是單一男性個人資料的兩倍。

在 2026 年優先提升用戶體驗後，Tinder 正將注意力轉向變現管道。管理層表示正在評估新的營收方案，但未提供具體的產品細節。

Hinge 目標 2027 年營收達 10 億美元

Hinge 持續實現雙位數的營收成長。管理層預期第四季成長率約為 19% 至 19.5%，並表示若在 2027 年維持類似的成長率，將有助於實現 10 億美元的營收目標。

成長機會包含 30 至 40 歲的用戶群以及沿海市場以外的美國中小型城市。Match Group 估計，透過在這些領域的擴展，Hinge 在美國可增加至少 100 萬名月活躍使用者。公司亦計劃於 2027 年進軍日本市場，但管理層預計日本市場在未來數年內不會帶來實質的用戶或營收貢獻。

Hinge 正在測試一種較低價格的訂閱方案，主要針對女性及價格敏感型用戶。Hinge 的每位付費用戶平均收入約為 30 美元，幾乎是 Tinder 的兩倍，而其付費滲透率僅略高於 Tinder 的一半。早期測試提升了兩種性別的轉換率，其中女性的增幅更為明顯。

價格最佳化以及較低營收之國際市場的較快成長，對 Hinge 第三季的每位付費用戶平均收入帶來壓力。管理層預計，隨著定價模式運行一整季，第四季該壓力將會上升，但仍會產生淨正向營收。

E&E 業務持續承壓

長青與新興品牌 (Evergreen and Emerging, E&E) 持續面臨逆風，原因為 Azar 在遭到 App Store 下架後進行 App 重構，以及其他品牌表現疲軟。Match Group 正優先發展較少數的核心產品，包括 Match、BLK、Upward 和 HER，同時利用共享技術支援更廣泛的產品組合。

管理層預計 E&E 營收衰退在 2027 年僅會有適度改善，並表示該部門恢復成長仍需要時間。

AI 與共享技術

隨著在全體員工中擴大使用 AI 工具，Match Group 將其年度 AI 支出從最初約 500 萬美元的預算增加至略高於 1,000 萬美元。公司透過放緩招聘速度而非裁員來抵銷部分增加的支出。

共享技術專案包括集中化的信任與安全工具、Tinder 的 Events 基礎設施、基於 AI 的產品原型設計以及專有的客戶關係管理 (CRM) 平台。管理層預計 2027 年 AI 成本將再次適度上升，但計劃透過模型選擇、使用量限制與開放權重模型來控制支出。

管理層財測指引

• Match Group 預期 2026 全年營收將略高於財測指引區間的中點。
• 公司預期將大幅超越 2 月設定的全年調整後 EBITDA 獲利率目標。
• 全年自由現金流預計將超越 2 月財測指引區間的高標。
• Tinder 2026 全年直接營收預計將呈低個位數的年衰退。
• 管理層預計 2026 年第四季 Tinder 的 MAU、付費人數及排除匯率影響的直接營收趨勢將有所改善。
• 針對 2027 年，管理層目標 Tinder 全年直接營收實現成長、第四季前付費人數成長，並在第四季底前實現 MAU 同期持平。
• Hinge 於 2027 年達到 10 億美元營收的目標保持不變。
• 管理層預計 2027 年 E&E 營收將繼續呈現負成長，僅較 2026 年的衰退程度微幅改善。
• Match Group 計劃隨時間推移，持續將 100% 的自由現金流用於股票回購、股利發放及員工股權獎勵淨額結算，不過每季的推進節奏可能會有所變化。

風險與關注事項

• 應用程式商店的經濟機制預計將降低 2027 年的獲利能力。自 10 月 1 日起，Apple 將 Match Group 需支付的歐盟應用程式內購費用從 20% 出頭調升至 26%，而與 Google 合作抵銷部分 Play 商店費用的協議將於 2027 年第一季到期。
• 在假設未來無其他費用變更的情況下，Apple 與 Google 的支付政策變更合計預計將對 2027 年調整後 EBITDA 獲利率造成 1 個百分點的逆風。
• Tinder 的 DAU 復甦尚未轉化為 MAU 成長。管理層表示，進一步進展取決於改變舊用戶與潛在用戶（尤其是更年輕的 Z 世代用戶）對品牌的看法。
• Tinder 計劃採用的即時推薦系統及全新變現舉措的成效仍有待驗證。
• E&E 持續面臨產品與平台相關的壓力（尤其是 Azar），管理層預計復甦過程將較為漫長。
• 2027 年行銷支出的規模與配置仍處於評估階段，可能會影響短期獲利率。

分析師問答亮點

分析師關注 Tinder 正向的 DAU 趨勢與下滑的 MAU 之間的差距。管理層將此差異歸因於現有用戶的參與度與留存率增強，同時指出 DoubleDate、Groups 與 Events 旨在吸引舊用戶與新用戶重新考慮使用。

針對 Tinder 改善的營收前景，管理層表示預期的用戶回饋金從 6,000 萬美元降至 2,500 萬美元。相較於 2 月的展望，這帶來了約 2 個百分點的改善，而更強勁的底層趨勢則額外貢獻了 1 個百分點。

Match Group 將 Tinder 的行銷策略轉向漏斗下層 (lower-funnel) 的用戶獲取及特定功能推廣活動。管理層表示，推廣當地 Events 的活動在註冊與重新啟用效率上顯著優於廣泛的品牌宣傳。受活動時程安排與投資回報紀律影響，約 800 萬美元的行銷支出已從第三季推遲至第四季。

在推薦技術方面，管理層將 Tinder 的系統描述為仍處於開發的早期至中期階段。即時個人化推薦預計將於年底前推出，但管理層坦承其效果仍有待證明。

針對 Hinge，管理層表示隨著市場走向成熟，擴展市場的成長速度自然會有所放緩。核心市場成長有所改善，歐洲市場保持強勁但成長放緩。拉丁美洲與印度仍屬於早期階段的機會，而日本則被定位為長期成長市場。

完整法說會逐字稿

---

完整財報電話會議逐字稿

管理層陳述

Operator

Welcome to the Match Group Third Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Tanny Shelburne, Senior Vice President of Investor Relations. Please go ahead.

Tanny Shelburne

Thank you, operator, and good afternoon, everyone. Today's call will be led by CEO, Spencer Rascoff; and CFO, Steve Bailey. They'll make a few brief remarks, and then we'll open it up for Q&A. Before we start, I need to remind everyone that during this call, we may discuss our outlook and future performance. These forward-looking statements may be preceded by words such as we expect, we believe, we anticipate or similar statements.

These statements are subject to risks and uncertainties, and our actual results could differ materially from the views expressed today. Some of these risks have been set forth in our earnings release and our periodic reports with the SEC. Also during this call, we will discuss certain non-GAAP financial measures. Reconciliations to the most directly comparable GAAP financial measures are provided in the published materials on our IR website.

These non-GAAP measures are not intended to be substitutes for our GAAP results.

With that, I'd like to turn the call over to Spencer.

Spencer Rascoff

Thanks, Tanny. Good afternoon. Thanks, everyone, for joining us. We're trying something new this quarter. We've posted our full prepared remarks to the Investor Relations website at ir.mtch.com for you to read through. But our plan for today's call is for me and Steve to just spend a few minutes on the key highlights and then devote the majority of our time to answer your questions.

A year ago, the big question you asked us was whether we could turn Tinder around. We knew we had to start by improving user outcomes, meaning we had to build better experiences that resonate with Gen Z, deliver more relevant recommendations and improve trust and safety. And that if we got those foundations right, then better user trends and ultimately, revenue would follow.

Tinder has had multiple consecutive quarters of improvement in the metrics that matter. This quarter, we had our strongest quarterly year-over-year performance in both Sparks, that's 6-way conversations, and Sparks Coverage, the percentage of users getting a Spark since we started tracking those metrics. In September, global daily active users returned to year-over-year growth for the first time in 3.5 years, and we've cut year-over-year declines in monthly active users nearly in half over the last year.

The improvement is broad-based. It spans women and men, age groups and geographies. We're seeing growth again in parts of the business where trends were moving in the wrong direction not that long ago. In October, markets representing nearly 70% of Tinder's DAU and 50% of its revenue had growing or flat DAU year-over-year, with the U.S. crossing into positive territory exiting the month.

So we now have evidence that those changes are working. There's still a lot of work ahead, particularly with the youngest Gen Z users. While we've made substantial progress improving the experience for existing users, the next step is reconsideration. That means giving people a reason to come back or to try Tinder for the first time. And that's what our Better with Friends Roadmap is designed to do. We're giving people more social and lower pressure ways to meet often with friends and in real life because that's what users want. Events is the feature I'm most enthusiastic about, but they're all geared towards creating more opportunities for people to form meaningful connections in real life.

We're building with our mission always in focus to make Tinder the most fun way to spark something new with someone new. We live and breathe this daily. Tinder employees can recite this clearly, and it's what keeps us working hard. Turning to Hinge. The business continues to deliver strong user growth and double-digit revenue growth year-over-year. We see clear opportunities ahead, too. Even in core markets where we've operated the longest, Hinge has room to reach more people beyond urban metro markets and grow with older daters.

We're targeting these opportunities deliberately with more relevant marketing. The early results are strong as registration trends across all genders and age groups improved as we exited Q3, and we think Q4 U.S. new registrations will be up year-over-year. Hinge's 2027 product roadmap is the strongest we've had in the last several years.

The depth of thinking about intention daters from what they need to what helps them get out on great dates is evident and gives me so much confidence in where Hinge is headed. At E&E, we put a more focused strategy in place, prioritizing a smaller set of brands and taking learnings from Tinder and Hinge to improve user outcomes. We are still early in this work, and it will take time to see the impact. At Match Group, we're approaching 2027 with a stronger foundation built through a lot of hard work and a clear path to durable growth.

With that, I'll turn it to Steve.

Steven Bailey

Thanks, Spencer. We're happy with the Q3 results and with the momentum in the business as we close out the year and turn our focus to 2027. We expect to finish the year slightly above the midpoint of our full year revenue guidance range and meaningfully exceed the full year adjusted EBITDA margin target we set back in February. We also expect to exceed the high end of our free cash flow guidance set back in February, and we remain committed to deploying 100% of free cash flow towards buybacks, dividends and net settling of employee equity awards over time, recognizing that the pace of capital return may vary from quarter to quarter.

Year-to-date, we've generated $832 million in free cash flow and deployed $591 million towards buybacks, dividends and net settling of employee equity awards, and we've reduced our diluted shares outstanding by 4% year-over-year through October and by 10% since January of 2025. Here's how we're thinking about the financial performance and expectations at each of our business units.

At Tinder, we expect to achieve our goal of improving year-over-year MAU and payer trends throughout the year, and we expect further improvement in both of these metrics as well as an improvement in direct revenue trends in Q4. That means that's meaningful progress towards returning Tinder to sustainable growth. As we said last quarter, we continue to expect Tinder's full year 2026 direct revenue performance to be better than we expected in February, declining at low single-digit rates year-over-year as a result of both smaller-than-expected impact from user experience tests and stronger underlying trends.

As we begin to shift our focus to 2027, while there's still more work to do, we're feeling great about the momentum, and we're confident we can deliver full year 2027 Tinder direct revenue year-over-year growth, along with year-over-year payer growth by Q4 and flat MAU year-over-year by the end of Q4. At Hinge, performance continues to be strong, and we expect to be within the 2026 full year guidance range we provided in February.

Our expectation for $1 billion in full year revenue in '27 hasn't changed. E&E continues to face headwinds from Azar's app redesign necessitated by the App Store ban earlier this year and weakness at other E&E brands, but we expect the strategy changes Spencer just outlined to slow revenue declines over time with the goal of returning to growth. But this will take time, and we expect E&E year-over-year revenue declines to only modestly improve in '27 compared to 2026.

Turning to Match Group profitability. We've done a good job this year executing on the Tinder turnaround while staying financially disciplined. We intend to take that same approach in 2027 as we balance making the right investments in our business to maximize long-term shareholder value with delivering near-term adjusted EBITDA and free cash flow. which we're giving back to shareholders through buybacks and the dividend.

We'll give detailed adjusted EBITDA guidance and a full year margin target on the call in February, but here's how we're thinking about 2027 profitability sitting here today. First, we expect to remain very disciplined on headcount and OpEx, including AI costs. Second, we now expect IP fees to be a 1 point margin headwind in 2027 based on current App Store policies and expiration of our Google partnership agreement, which I'll talk more about in a minute.

And third, we're still evaluating the appropriate level and allocation of marketing spend across the portfolio using our PRISM framework. As a reminder, PRISM is our standardized marketing measurement framework that we recently introduced at Match Group. It gives us a consistent, rigorous quantitative assessment of marketing effectiveness that we use to inform investment decisions across the company.

2027 will be the first time we're incorporating it into our annual planning process. Now let me give you a little more color on IP fees. First, an update on Apple. In the U.S., we continue to pay 0% in IP fees on alternative payments on iOS, and we don't expect changes to this fee structure until there's a decision in the Epic versus Apple case. In the EU, Apple actually raised fees on in-app purchases effective October 1, and the fees we pay increased from the low 20% to 26%.

At the same time, Apple will begin allowing alternative payments, which we plan to test, but there are a number of restrictions and the economics are far less favorable than they are in the U.S. We expect the increased in-app purchase fees to cost us an additional $20 million in 2027. Now on Google.

In 2024, we entered into a partnership arrangement with Google that provided value exchange across our broad relationship that has effectively offset the additional fees we incurred when we adopted Google's billing options at that time. As we previously disclosed at the announcement of that deal and in subsequent filings, that partnership is set to expire in the first quarter of 2027, and we expect PlayStore fees to meaningfully increase beginning in Q2 of next year.

Our expectation when we entered into the agreement was that legal and regulatory pressure would reduce Google's in-app purchase fees by the time our agreement ended. However, Google's new global fee structure announced in March provides a little benefit to us. Taking into account both the Apple changes and the Google partnership agreement expiration and making no assumptions on future fee changes, we expect IP fees to result in a 1 point headwind to adjusted EBITDA margin in 2027.

While we're looking for ways to mitigate the impact, we also need to ensure we're making the right investments in the business to drive long-term shareholder value. We'll make that determination as part of our annual planning process, and we'll give you more details on 2027 revenue and margin expectations next quarter.

With that, let's open it up to Q&A.

Operator

[Operator Instructions] The first question comes from Robert Coolbrith with Evercore ISI.

分析師問答

Robert Coolbrith

Just wondering if you can maybe talk us through the path to positive DAU that you had in the quarter or had in September. Maybe if you could sort of disaggregate that across retention improvement or if there's just been just a case of stacking those retention gains that you've had in successful quarters across more cohorts? And then any early indications of improving levels of consideration or reconsideration?

Spencer Rascoff

Yes. Thanks, Robert, and thanks for your recent research note. Look, it was a big milestone this quarter. Turning DAU positive in September year-over-year for the first time in 3.5 years felt great. There was -- there were a lot of high 5s and just excitement around achieving that milestone. And then just a couple of weeks later, at the end of October, U.S. DAU turning positive year-over-year was also just felt great.

Those are the DAU trends. MAU improving from negative 10% to negative 5.5% over the last few quarters also speaks to the quality of the improvements in the product and also improvements in MAU -- similar improvements in MAU across our top 5 revenue markets. The reason why I'm confident about the sustainability of the turnaround is how broad-based the improvements have been.

So really any way you cut it by age, by gender, by geography, we're seeing improvements across really every which way you look at it. So for example, looking at it by gender, our female MAU is improving at an even faster rate than male MAU in every region and every age group. Female MAU declines narrowed by 8 points year-over-year and 2 points quarter-over-quarter.

So this is not a turnaround that's just focused on male users. Quite the contrary, it's actually, we're improving among female users even more quickly. Looking at it by age, again, this is not a turnaround focused on just adding easier-to-acquire older users. Our under 30 MAU trends have improved even more than our over 30 MAU trends, and we're improving across every single age group.

So that's the exciting news about the MAU turnaround. I think kind of the question behind your question, perhaps, is why is DAU growing faster than MAU. And the answer to that is we have made so many product improvements that help the existing experience basically improve our users' current experience, but we still have more work to do to drive reconsideration.

And that's where the Better with Friends social RoaDmap comes into play. We had a hit in DoubleDate. We layered on another hit with events. We're in the process of building a third hit with groups. And when you take those 3 things together and you do brand studies and ask users about their perception of Tinder, you start to see really good promising results about people changing their perception of what Tinder is. So we still have work to do. It's not -- it's definitely too soon to declare victory, but I really am proud of and impressed by where we're at today. Steve, anything to add or...?

Steven Bailey

You've covered it.

Operator

The next question is from Shweta Khajuria with Wolfe Research.

Shweta Khajuria

Spencer, could I please ask a follow-up to what you were just talking about in terms of the delta between DAU growth and MAU growth. So there is this idea of perception and bringing people back into the ecosystem, and that's why the delta. So how long do you think that lag -- how long do you think it takes for that lag to narrow?

In other words, for that perception to change, bring those people back and turning around MAU growth? And then the second question I have is around next year, Steve, you spoke about Tinder expectations. To the degree that you can speak to other segments, Hinge and E&E, that would be very helpful or just at a platform level, the puts and takes for both revenue and EBITDA. Understood. You're not guiding just yet in terms of EBITDA, where would the levers be coming from to offset the headwind? And how should we be thinking about overall top line growth?

Steven Bailey

Let me take -- we'll take this in a couple of pieces. Let me talk a little bit more about the Tinder sort of metrics improvements and Spencer jump in as you wish. The way to think about DAU outpacing MAU, here's kind of where things stand. Of course, we're very excited that DAU crossed over to positive growth. we're just as excited about MAU improving. So we got about 1 point improvement on MAU from Q2 to Q3 with MAU down 6%, which is better than the 7% it was down the quarter before.

We also saw an improvement in payers, which we're very excited about. So the payers were down negative 5% in Q2. They were down negative 4% in Q3. So progress there as well. Revenue, we like to look at FX neutral, was down 2%. That was the same as Q2. And I'll just stress that these metrics aren't going to move completely in lockstep every quarter. But if you take a step back and look at the revenue FX-neutral trends, you've got a diagram in the prepared release, you can see progress over the last few quarters from down 5% in Q4 last year to down 2% now.

And then what I think is also encouraging is we expect all those metrics to improve in Q4. So we expect payers to get better in Q4. We expect MAU to get better in Q4, and we expect that FX-neutral direct revenue to get better as well. So we are starting to see it flow through more than just DAU, but to these other core metrics as well.

Did you want to add anything to that?

Spencer Rascoff

No, that's good.

Steven Bailey

And then when we -- let's talk about 2027 a bit. We're not going to give, as you sort of alluded to, full year revenue guidance or EBITDA guidance until February, which is sort of our normal course. I tried to give some color on how we were thinking about things. The only thing I would add is, yes, like if we break it down, you look at Tinder, we're feeling really good, right? We've got a -- we're making progress. I just told you we expect further progress in Q4, and we're really setting ourselves up well to meet the goals I laid out in terms of payers, revenue and MAU improvement next year.

Could it be sooner? Maybe. It's hard to pinpoint exactly, but I think we're on a really good trajectory at Tinder. When you look at Hinge, that achieving the $1 billion in revenue, which we still feel very good about, I think about it this way. Revenue growth was about 19% in Q3. We expect it to be about the same, maybe a little bit better in Q4. So think of 19%, 19.5%.

And then we've got to keep the growth rate at about that same level next year. And if we do that, we'll hit the $1 billion. So we think we've got the right road map and strategy in place to do it. That's going to come from improving core trends and expansion. We talked about expanding into Japan in the prepared remarks, too. And so I'm feeling good about the trajectory there, too. And then E&E, what we said was modest improvement from the declines we're seeing this year. That's really in recognition of -- Spencer talked about putting the right strategy in place, but it's going to take some time to get that business back to growth.

And we don't -- so we think only modest improvement, but still negative at E&E next year. In terms of profitability, the way I would think about it is we've got the 1 point IP fee headwind. We said we'd be really controlled in terms of headcount and AI costs. The really big consideration is really around marketing spend. What is the right level of marketing spend, particularly at Tinder and Hinge in 2027 to maximize long-term returns and shareholder value. And so we need to balance that against hitting short-term margin targets. And that's what we'll do. We'll think very hard about. We'll do it in a very quantitative way through the PRISM framework, and we'll give you a lot more details on all that in February.

Operator

The next question is from Jason Helfstein with Oppenheimer.

Jason Helfstein

Just some more color on the givebacks. I think it sounds like the full year givebacks for Tinder are going to come in below expectations. Why have they not been necessary this year relative to your plan? And any more color there?

Spencer Rascoff

Yes. Jason, thanks for the question. We've been fortunate to discover that when we launch improvements to the user experience, the accompanying MAU retention mitigates revenue loss. For example, we changed our recommendation algorithms to maximize user outcomes, and we nervously wondered if it would hurt revenue. And what we found was that the benefits of improving user retention outweighed the detriment of prioritizing user outcomes over revenue. That might not always be the case, but that's what we mostly saw in 2026.

Steven Bailey

Yes. Let me just add a finer point to that, too. That's all exactly right. And what it means financially is that we thought at the beginning of the year, we need -- we would need $60 million in user givebacks in '26. What we're now saying is that we only ended up using $25 million. If you do the math, that's about 2 points of revenue at Tinder, and that's what's driving about 2 points of that improvement versus expectations we had set back in February.

There's about another point of improvement in there in terms of what we think we'll actually finish the year at for Tinder full year revenue relative to what we said in February. That other point is improving trends. So I want to be really clear about that. Really, what we're seeing at Tinder is 2 things: the need for less user givebacks and also improving trends. So that sets us up really well for continued improvement in '27.

Operator

The next question is from Bryan Smilek with JPMorgan.

Bryan Smilek

Spencer, as you enter Tinder's resurgence phase, can you just help us think about the allocation of marketing spend by channel to reignite the consideration set? I think in the prepared remarks, you had mentioned Gen Z marketing around creator and social content and event partners. So just curious, anything that you can share on the LTV curves of those channels.

Spencer Rascoff

Sure. So thanks, Bryan. We made a lot of changes to how we allocate and how we invest our marketing in 2026. Firstly, we shifted a lot of our spend towards down funnel user acquisition instead of up-funnel brand building. Secondly, we pointed much more of our marketing spend towards actual features rather than sort of high-level positioning about Tinder or the category.

And we saw both of those changes drive really improved results in '26. The second one, I think, is particularly important and reminds us how product and marketing work hand in hand. So for example, when we market on Instagram that Tinder has events in your city, and we look at the cost per registration or cost per reactivation for those types of marketing creatives and compare it with a year or 2 ago when we were buying the same channel, but promoting use Tinder because Tinder is a dating app, it's really night and day in terms of the lift that we get and in most cases, the efficacy of how far the dollar goes when we're focusing on product innovation rather than not.

As Steve alluded to, we also now have a framework where for really the first time ever going into an annual planning process, we have a cross-brand rubric and analytic framework to decide what the right marketing spend and right marketing allocation should be between business units and brands within business units. And so we're deep in the midst of that right now as we finalize our 2027 plans.

So I don't have any more detail to report, but we're just in a much smarter place here as we speak to you in fall of 2026 than we were in fall 2025, coming off the heels of a really effective '26 for marketing and also the development of these new frameworks to assess marketing efficacy.

Operator

The next question is from Nathan Feather with Morgan Stanley.

Nathaniel Feather

I guess as you think about DAU turning positive, the focus on MAUs, what are kind of the 1 to 2 key gating factors to returning to flat MAU growth from here? How do you think about the main levers you can pull to really drive that reconsideration of the brand as you continue to push the products out?

Spencer Rascoff

Yes. Thanks, Nathan. The biggest lever is convincing people who are not already on Tinder or who maybe used Tinder 2, 4, 5, 10 years ago to give it another shot. And this Better -- with Friends road map is the best idea that we have to execute on that.

So as I said, DoubleDate is a winner. We've delivered 30 million matches on DoubleDate so far, which just speaks to the scale and product market fit that DoubleDate has. We just launched Groups in the U.S. and a couple of weeks earlier in a couple of other countries, just to give you some insight into groups. Women are twice as likely to like a group profile as they are to like a single male profile.

That's -- again, that speaks to the product market fit that we think Groups will have within our product. 15% of our users in Norway, age 18 to 22 are already in a group. The reason I mentioned Norway is that was the first country that we launched in. So it's been up for a couple of extra weeks. So we think Groups has the potential to start -- to be another hit and to change the way people perceive Tinder. And then finally, events, which we've been very vocal about the potential that we think events has, events has absolutely started to change perception of Tinder.

When we look at our measures of brand health, questions that we ask users like does Tinder have people that you want to meet on it? Is Tinder an app that with your kind of people? Is Tinder just for hookups or for things more than that? All those measures of brand health have improved over the summer and into the fall as we've rolled out events into more and more markets. So I'm happy to talk more about events if there are other questions about it. But those are the types of initiatives that we're focused on in order to drive reconsideration, which is what will take us these last couple of points from negative 5.5% MAU to eventual MAU growth.

Operator

The next question is from James Heaney with Jefferies.

James Heaney

You mentioned in the letter an increased focus on Tinder revenue initiatives in 2027. Curious what kinds of options you'd be looking to explore and how you think you can improve monetization without disrupting the user engagement momentum that you have?

Spencer Rascoff

Yes. Thanks. Good question. When I took over Tinder, I started by focusing on the user experience. We had this theory, which has proven to be true that if you improve the user experience, then the audience metrics will improve and then other good things will flow from that. We moved very quickly. We changed a lot of the user experience. And today, Tinder works better than it did before. We're now turning our attention towards revenue, and we're really looking at everything.

We understand what we have currently and what we can learn from other brands, and we're approaching it from pretty much a blank slate in terms of the starting point. It's a bit early to say where that will go, but I can assure you we're bringing the same level of focus, energy, vigor, urgency, innovation to monetization that in '27 that we brought to the user experience in '26. And it's -- we have a lot of ideas that we're not ready to talk about yet, but I'm optimistic that '27 will be a year of a lot of monetization and innovation.

Operator

The next question is from Youssef Scully with Truist Securities.

James Heaney

Maybe a quick question for Steve. Can you just quantify the marketing spend that was pushed into Q4 for Q3, the reasons for it? And would you have been able to grow a little faster had you actually met that marketing spend in intra-quarter?

Steven Bailey

Yes. In terms of quantity, we pushed about $8 million or so of marketing spend from Q3 to Q4. I think we came in about $12 million better than we expected, but $8 million of that was marketing spend. It's a couple of reasons. One is just timing of campaigns -- and two is we're going to stay financially disciplined, right? So in any given quarter, if we're not seeing the returns that we expect the marketing spend, we're not going to go spend those dollars just to fill the budget, right?

And so I like to see us underspend for those reasons because it means we can deploy them in a different quarter at higher ROI. So it's really a combination of those 2 things. How much revenue we would have generated had we not done that's a little bit of a hypothetical that I'm not going to do at the moment. But I mean, what I would say is, and I'll repeat it again, which I think is an important fact, we're going out and saying we expect MAU payers and revenue direct revenue. So while direct revenue didn't improve in Q3, we do expect it to improve in Q4 from a year-over-year growth perspective relative to Q3. So with the marketing spend in Q4 and everything else we have going, we expect to see improvement across all those metrics next quarter.

Operator

The next question is from Benjamin Black with Deutsche Bank.

Benjamin Black

So the recommendation system looks to be at least a big part of the foundation of Tinder's turnaround. So how much better can matching outcomes become as the system moves towards more real-time personalization? And what advantages does the Tinder scale really have and the data that you guys have, how much of an advantage is that for you guys? And then secondly, I'd be interested in hearing the learnings from misconnections. I think in the letter, you mentioned the 4x improvement in likes from women. And it seems a bit of a game changer to me. So what are some of the hurdles to sort of broadening out in misconnections?

Spencer Rascoff

Yes. Thanks, Benjamin. Q unification and our REX improvements have been really critical to our improving business results in 2026. We're -- we just rolled out C unification to other user types. I think we just rolled it out to women seeking men and women seeking anyone or other different LGBTQ audiences, but we still have more to go.

Real-time REX, we've said we intend to launch by the end of the year. We're very confident in the potential that, that will bring, but it's also unproven. So I think that event that we did, the CEO Connection event where I spoke with Vinay and Mark, our CTO and Head of Product at Tinder, about REX. I think Vinay said something like he thinks it's the second or third inning or maybe you said the third or fourth inning maybe in terms of Rs to try to size the amount of potential improvements that still lay ahead. I think that's probably about right.

So already huge improvements, but more to come. On mix connections, this is a feature that we're testing in Canada right now. If you have location services turned on, then once a week, we give you specific drops that say, here's somebody that you cross paths with in the week in a given neighborhood or in a given coffee shop, et cetera. And users love it.

The serendipity and kind of excitement is really meaningful, especially for Gen Z users, the nostalgia, the sense of a meat cute, which is what our users are really looking for. And the digital bridge between an app and an IRL experience is really powerful. We are now working to improve it. we'll be assessing when and whether to roll it out more broadly. There's still a lot of work to do to improve the product. But so far, so good in what we've launched so far.

Operator

The next question comes from Alec Brondolo with Wells Fargo.

Unknown Analyst

Maybe 2 for me. First, how do you choose the new focus brands for E&E? And does refocusing on a smaller number of brands create an incremental headwind to that business beyond Azarre in '27? And then secondly, on Hinge, the new subscriptions you're launching in the fourth quarter, giving users more choice and control, is there any incremental detail you could provide around the value proposition? What is the opportunity here on paid conversion?

Spencer Rascoff

Yes, I'll take the E&E question and then, Steve, you can answer the Hinge question. So in deciding which brands to prioritize with E&E, we looked at the white space that the different brands are going after and also the size and in order to prioritize. So for example, Upward and BLK and Her all are going after white space in their respective user segments, where we think there's a real right to win, a reason to exist, a category white space that is not fully won by a Hinge or a Match or Tinder or any stand-alone competitor.

And then brands like Match, which are kind of more general population, we think there's a reason for them to exist in the 50-plus age segment and just the brand awareness and brand power of the term match and just the size and scale of that really represents something that is worthy of prioritization. The other non focused brands like an OkCupid, like a Plenty of Fish, like Salons, these are brands that are still being worked on, but only synergistically across the portfolio.

So for example, we have a dedicated team that is improving profile creation and a dedicated team that is improving photo selection and a team that's building chat and a team that's building recommendations algorithm. When each of those teams launch a feature -- a set of features, it goes out across the whole portfolio. So something like OkCupid is not -- it's not like left to its own devices. It's still being improved upon by these shared teams that are building E&E-Y technologies. So that's how we prioritize the different brands. And Steve, do you want to take the Hinge question?

Steven Bailey

Yes. I think -- look, we're pretty excited about the lower-priced subscription tier in test right now. It's designed for women, in particular, but also for price-sensitive users. Remember that Hinge is one of the highest-priced products we have. The revenue per payer, for example, at Hinge is almost twice that of Tinder around $30. It's also has one of the lower payer penetrations. Its payer penetration is only slightly more than half that of Tinder.

So we think it makes sense to offer a lower-priced offering in what is otherwise a higher-priced product to get more payer conversion from more price-sensitive users. In terms of what it includes, think of it -- I'm not going to get into every specific feature, but it -- the broad brush strokes are additional filters and custom location boundaries, features that women find valuable. And what we've seen in test is it has improved conversion for both men and women, but women's conversion has been up more than men. And the other interesting thing here is that, remember, Hinge's gender balance between men and women is far better than most other apps.

So -- and women pay at far lower rates. That's true for Hinge, and that's true for everybody. And so if we can get this feature right and a feature set that really resonates with women, that will be really potentially powerful from a monetization perspective for Hinge in particular, with a large women user base and really good gender balance. So we're going to keep testing and tweaking, but the early reads are good. And then we expect to roll it out more in Q4. So stay tuned.

Operator

The next question is from Jamesmichael with Citi.

Jamesmichael Sherman-Lewis

Two here, please. First, Tinder payer penetration increased sequentially and year-over-year and MAUs are still declining faster than payers. So how much additional payer conversion runway remains as we work to inflect back to MAU growth?

And how do you think about the balance of conversion gains versus audience growth into 2027? And then as a quick follow-up, Hinge's opportunity in core markets, I think you noted an uptick in U.S. marketing contributing to the improving registration trends. Could you talk a little bit about the opportunity to drive MAU growth and reconsideration in mature Hinge geographies?

Steven Bailey

Yes. On the Tinder payer penetration, you're right, Payer penetration was up again. That's something we've consistently seen is that the improvement in MAU is flowing through to improvement in payers, which is great. In terms of how much more room to go, while payer penetration at Tinder is higher than Hinge, which I just mentioned, it's still relatively low.

It's in the low 20%. So we're always looking to optimize not only for the payer experience, but the free experience, too. But in terms of the ability to drive further incremental payer conversion, I think there's lots of room. I think as we look ahead, what we're doing is we're not optimizing around short-term revenue growth. We're really taking a look -- Spencer got into this earlier, but finding ways to add value to the experience both for free and paid users. And if we can add value, then I see no reason why we can't increase monetization and payer conversion at Tinder going forward.

Spencer Rascoff

And I'll take the Hinge MAU and TAM question. So in the U.S., Hinge is most highly penetrated on the coast and in the 20 to 30 age group. If you look at smaller cities outside of the coast and age 30 to 40, there's ample room for Hinge to grow. We think at least another 1 million monthly active users in the U.S. as we focus more on those other user segments and geographies. We've already started implementing that playbook with different marketing tactics and messages in order to focus on those secondary cities and slightly older age users. And we think that, that playbook also is extendable to other core markets outside the U.S., like the U.K., Australia and Canada.

Operator

The next question is from Eric Sheridan with Goldman Sachs.

Unknown Analyst

I mean first one would be following up on, I think Benjamin asked earlier about the broader AI strategy. The flip side of that coin would be, how are you thinking about managing token costs or input costs within your broader cost structure as you continue to move from testing to scaling some of your AI initiatives? That would be one. And in the shareholder letter, you also talked about finding ways to drive tech platform efficiencies through elements of shared technology platforms or cross-pollinating.

So the second part would be, how should we be thinking about some of the efficiency dynamics that can be gained through some of those initiatives as we move out of '26 and into '27?

Steven Bailey

Yes. Let me take the AI token costs. I think we've done a really good job here. The costs are definitely going up. We've -- but let me give you some more details around how we're managing it. We started probably with a budget this year of about $5 million for AI cost, but we had pretty low usage. We basically had engineers using AI and nobody else, and it was spotty at that. We sort of early in the year, May time frame, we really wanted to push AI adoption and become a truly AI native company. We thought it was important for the overall long-term success of the business.

And so we gave all our employees tools like Claud, tools like OpenAI, enterprise search tools like Clean and gave them the training, gave them the tokens to go use these tools and our cost probably went from $5 million to about $10 million, a little more than $10 million. the way we do it was unplanned doubling of costs, but we think worthwhile, and we're getting really good efficiency gains out of it. The way we offset those costs was by controlling headcount. So not reducing the employee base, but hiring less, slowing headcount growth and saying, let's do more with less through the use of AI, and that's been really effective.

Right now, I would say we're maturing as a company in terms of AI token usage, and we're really in that sort of efficiency phase. We've got an AI gateway that controls what models we're defaulting to and what models engineers can use. We're doing all sorts of things like making the most expensive models only available by request. We have speed bumps or soft caps engineers get about $1,000 a month and non-engineers get about $250. Once you hit that cap, that doesn't mean you can stop -- you have to stop your work. It means you have to go to your manager and say, look at the work I'm producing.

And if it makes sense, you can keep going. And so we've given all the managers really great visualization into the spend within their departments, basically real-time token spending. And I think we've done an incredible job as a company overall in this regard. We do expect costs to go up a little again next year, but the hope is through these efficiency gains, we're now starting to look at open weight models, we'll be able to keep them under control. And then in terms of in the product and in the features themselves, and feel free to jump in, Spencer, there are more features that are using AI tokens and AI costs. And do you want to cover how we're thinking about that?

Spencer Rascoff

Yes. Well, let me answer it from a shared tech platform standpoint and efficiencies across the company. So I'll give you 4 quick examples. We've integrated our trust and safety tooling so that features like Facecheck are now built centrally across all brands. Another example of a shared service is Tinder's events product.

So the module tap 2 on Tinder, where you can see who's going to an event and you can see the different types of events and this whole product road map around events now powers not just Tinder, but also BLK and likely in the future, other Match Group brands. So those are 2 examples of things that we're building centrally across multiple brands. Then 2 other potential examples that we're working towards is Tinder has built an AI prototyping tool where any employee at Tinder can enter a prompt to visualize a new feature to Tinder, and it will instantly build a prototype of it and also build code for that prototype new feature. And that's something that has really changed the software development life cycle at Tinder, made us more design forward, just brought a higher pace of innovation, and that's something that we're looking to extend to some of our other brands, including E&E.

And then a fourth example is Tinder has a homegrown customer relationship management platform that manages all of Tinder's push notifications, text messages and e-mails. And it's really best-in-class better than what companies pay millions or tens of millions of dollars a year to SaaS providers for that type of technology, and we're considering extending that to some of our other brands. So those are just a couple of examples, Eric, of how we're building centralized technology and extending it across the whole company.

Operator

The next question is from John Blackledge with TD Cowen.

John Blackledge

For Hinge's newer markets outside of the U.S., is user growth and payer conversion following similar growth curves to Hinge's more mature markets? And then just another one on Hinge. On the RPP, should we expect that trend that we saw in 3Q to continue as you roll out the new lower-priced subtier?

Steven Bailey

Yes. Let me start and feel free to jump in, Spencer. On the MAU growth, yes, it's an important point to make. Global MAU growth came in at around 12%. What we saw sort of under the covers is actually core MAU improved a bit, which is great to see. Spencer talked a little bit about that. And growth in Europe slowed a little bit, still phenomenally high growth. But to your point, there's a growth curve as these countries scale, we expect their growth to slow over time. And so that's really what you saw in the headline number was core getting a little bit better, Europe slowing down a bit and us netting out to about 12% growth. That's right around what we expected. puts us right in track not only for MAU, but also for revenue for the full year 2026. So we feel good about the way those curves are playing out. It's just what we expected. We're entering LatAm. That's growing faster but still small. And we've relaunched in India, where we have a pretty big MAU base today.

And then what we're doing is we're laying the foundation for expansion into Japan in '27. I wouldn't expect a lot of revenue from Japan in '27 or users, but it's an important -- first of all, it's one of our top -- it's probably the top 5 largest markets from a TAM perspective, from a revenue perspective in the category. So an important one for Hinge to enter. And I think we'll be -- we think there's a nice white space for it between Pairs, which is a little bit more marriage focused and Tinder, which is more casual. All our research says it should do very well there, and I think it will be a growth driver for 2028 and beyond. So that's -- those are some of the Hinge growth curves. What was the second part of the question?

Spencer Rascoff

I don't -- what was your second question, sorry?

John Blackledge

The RPP was down a little bit in 3Q. Yes, going forward.

Steven Bailey

Yes, that's -- yes, thanks for pointing that out. So yes, what you're seeing is we saw an acceleration in payers, and we saw RPP down a little bit year-over-year in Q3. That is price optimizations, a new model we rolled out at Hinge, that was playing the biggest effect. To a lesser effect is the new lower-priced tier that can have a bigger effect down the road.

And then you normally sort of underneath all that, we naturally get some RPP pressure because the growth in these expansion markets is still so much higher than the core markets and the RPP on average is lower there than in a market like the U.S. On the price optimization models, let me just give you a little bit more detail there. What that is, is a propensity to subscribe model. Most brands have it. It's something Tinder has been doing for years. I'd say most brands in the category do this. Hinge is just maturing in its monetization capabilities and so is now rolling this out where what it's doing is it's looking for those that haven't subscribed after a certain number of days and are unlikely to subscribe ever.

Again, Hinge is a high-priced product. We talked about that and giving a targeted lower price to those users and driving payer conversion. That's driving up -- that's accelerating payer growth and being offset by some RPP degradation. I'd expect that to actually increase a bit in Q4 just because this pricing model was rolled out sort of mid-Q3 and will be a full quarter in Q4. But it's generating net positive revenue -- it's another lever for offering a paid experience to more price-sensitive users. And third, and I think this is overlooked sometimes, payers help the ecosystem, too.

So there's indirect benefits of higher paid conversion. Payers retain longer, they respond to likes better. They are more likely to buy ALC features. So there's a lot of reasons why having a little bit higher payer penetration at Hinge still won't be even close to the levels of Tinder makes a lot of sense, but that's what's causing a little noise in those metrics. And I'd expect that to continue in Q4.

Operator

The last question will be from Andrew Marok with Raymond James.

Andrew Marok

And maybe one more on that Hinge and the international business. So you called out in the shareholder letter, and I think you answered a bit of this in the last question about the momentum in Europe and the newly launched LatAm markets. So I guess what kind of common threads are you seeing between these disparate geos? And on Japan, specifically, what is kind of giving you the confidence that there is a space in the market, which has kind of been a tricky market for everyone in the past?

Spencer Rascoff

Yes. Thanks, Andrew. What we're seeing as the common thread in the different markets is that Hinge's slow and steady approach of customizing the product, customizing the marketing for the individual geography is working. So for example, in India, having different -- slightly different set of features like screenshot banning, waitlist for men and a couple of other unique trust and safety features like ID verification and others are particularly useful.

So we're taking a deliberate approach to Japan, learning from, gosh, probably 10 years of pairs of success in the matrimony space. We've done a huge amount of original research in Japan to identify that there is white space between the Matrimony apps and the more casual apps like Tinder. So we think there's an opportunity for Hinge to succeed without cannibalizing pairs in Tinder. And we're going to customize the Hinge product and marketing in order to achieve product market fit there just as we have in Brazil and Mexico and India over the last year or so. I think, operator, are there any other questions? Or is that the last question?

Operator

That was the last question, sir.

Spencer Rascoff

Okay. All right. So I'll just wrap up here. As a reminder for anyone who joined late, we posted the full prepared remarks to the Investor Relations website for you to read through, so you can find the full script there. Hopefully, you like this new more efficient approach. I just want to close by taking stock of my first 7 quarters as CEO here. When I joined 21 months ago, Tinder users and revenue were declining and there was skepticism that Tinder could be saved at all.

Today, we put points on the board and mostly put that question to rest. Several quarters of improving user trends at Tinder top to bottom from Sparks to Spark coverage to DAU to MAU an improvement in payers and revenue when adjusting for FX. So we've demonstrated that this turnaround is real and sustainable. We also took Hinge from $150-ish million in quarterly revenue back in Q1 2025 when I started to over $220 million in quarterly revenue in Q2 of '26.

We purchased her. We purchased about half of SNFs. We significantly improved employee engagement. We drove massive productivity improvements from AI. So we got a lot done in a pretty short period of time. Meanwhile, we bought back $1.2 billion of stock and paid $369 million of dividends and reduced diluted share count by 10% year-over-year as of yesterday. I'm very proud of what we've accomplished in these last 21 months, and I'm even more optimistic about where we'll be 21 months from now.

We'll talk to you in early February at Q4 earnings, and I hope to see many of you at our second Tinder Sparks event on February 23 in New York. Until then, thanks, everybody. Bye-bye.

Operator

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.