DoubleVerify Holdings, Inc. (DV) 2026 Q1 法說會逐字稿

Thank you for standing by. My name is Rebecca, and I will be your conference operator today.

感謝您耐心等候。我叫 Rebecca，今天將擔任本次電話會議的接線員。

At this time, I would like to welcome everyone to the DoubleVerify first quarter 2026 earnings conference call.

此刻，我謹代表公司歡迎各位參加 DoubleVerify 2026 年第一季財報電話會議。

(Operator Instructions)

（接線員指示）

I will now turn the call over to Bridley Johnson, Investor Relations. Please go ahead.

現在我將把電話轉交給投資人關係的 Bridley Johnson。請開始。

Good afternoon, and welcome to DoubleVerify's first quarter 2026 earnings conference call.

各位下午好，歡迎參加 DoubleVerify 2026 年第一季財報電話會議。

With us today are Mark Zagorski, CEO and Nicola Allais, CFO.

今天與會的有執行長 Mark Zagorski 與財務長 Nicola Allais。

Today's press release with this call may contain forward-looking statements that are subject to inherent risks, uncertainties, and changes and reflect our current expectations and information currently available to us, and our actual results could differ materially. For more information, please refer to the risk factors in our recent SEC filings, including our Form 10-Q and Form 10-K. In addition, our discussion today will include references to certain supplemental non-GAAP financial measures and should be considered in addition to and not as a substitute for our GAAP results. Reconciliations to the most comparable GAAP measures are available in today's earnings press release, which is available on our Investor Relations website at ir.doubleverify.com.

今天的新聞稿與本次電話會議可能包含前瞻性陳述，該等陳述受固有風險、不確定性與變動影響，並反映我們目前的預期及現階段可得資訊；實際結果可能與之有重大差異。更多資訊請參閱我們近期向美國證券交易委員會（SEC）提交的文件中的風險因素，包括 Form 10-Q 與 Form 10-K。此外，我們今天的討論將提及若干補充性的非 GAAP 財務衡量指標，應與我們的 GAAP 結果一併考量，而非取代 GAAP 結果。與最可比 GAAP 指標的調節表載於今日的財報新聞稿中，該新聞稿可於我們的投資人關係網站 ir.doubleverify.com 查閱。

Also, during the call today, we'll be referring to the slide deck posted on our website.

另外，今天的電話會議中，我們將引用已張貼於我們網站上的簡報投影片。

With that, I'll turn it over to Mark.

接下來我把時間交給 Mark。

Thanks, Brinlea, and good afternoon, everyone.

謝謝你，Brinlea，各位下午好。

We delivered strong Q1 results as we continued our solid execution on our product innovation, strategic, and financial roadmaps. In Q1, we achieved 10% year-over-year revenue growth, led by accelerating growth of our social verification and optimization solutions, and we delivered a 31% EBITDA margin, which exceeded expectations, largely due to AI-fueled operational efficiencies. Advertiser growth was positive across all key industry verticals in the quarter as we continue to benefit from our focus on further diversification of customer engagements and ad spend across various client types.

我們交出強勁的第一季成果，持續在產品創新、策略與財務路線圖上穩健執行。第一季營收年增 10%，主要由社群驗證與最佳化解決方案的加速成長帶動；同時我們達成 31% 的 EBITDA 利潤率，優於預期，主要受惠於 AI 驅動的營運效率提升。本季各主要產業垂直領域的廣告主成長皆為正向，我們持續受惠於聚焦進一步分散客戶合作與各類客戶型態的廣告支出。

We also repurchased $100 million worth of shares year-to-date, reflecting confidence in our business and our commitment to returning capital to shareholders as a core element of our long-term value creation strategy. We expect to deliver a strong 2026 as we successfully execute on our strategic plan to verify the quality, optimize the investment, and prove the impact of digital ad impressions across any platform, media, or market or advertiser spend.

我們今年迄今亦回購了 1 億美元的股票，反映我們對業務的信心，以及將資本回饋股東作為長期價值創造策略核心要素的承諾。隨著我們成功執行策略計畫，在任何平台、媒體或市場、以及任何廣告主支出下，驗證數位廣告曝光品質、最佳化投資並證明成效，我們預期 2026 年將有強勁表現。

The solid results this quarter were fueled by our core growth catalysts. Social activation and measurement products, streaming TV verification, and our dynamic suite of solutions that empower advertisers to better navigate the evolving ecosystem of AI advertising platforms and GenAI content. Across all of these sectors, our incredibly durable value proposition remains tantamount. DV is the independent, essential trust layer that marketers rely on to ensure their ad spend is protected from fraud in unsuitable context, and most importantly, delivers the highest possible return on investment. And this essential role in the ecosystem continues to expand as new product innovations power our growth flywheel.

本季的穩健成果由我們核心成長催化劑所驅動。包括社群啟動與衡量產品、串流電視驗證，以及我們的動態解決方案套件，協助廣告主更好地因應 AI 廣告平台與生成式 AI（GenAI）內容所帶來的生態系演變。在所有這些領域中，我們極其持久的價值主張仍然至關重要。DV 是獨立且不可或缺的信任層，行銷人員仰賴我們確保其廣告支出免於詐欺與不適當情境的影響，更重要的是，帶來最高可能的投資報酬率。而我們在生態系中的關鍵角色也持續擴大，因為新的產品創新正推動我們的成長飛輪。

Let me share a few recent stats that underscore the impact of these investments.

我分享幾項近期數據，以凸顯這些投資所帶來的影響。

Driven by continued success on Meta, social measurement grew 23% year-over-year, a significant acceleration from Q4. Social activation, our fastest-growing solution set, grew 92% year-over-year in Q1, up from 62% in the fourth quarter. Authentic AdVantage on YouTube, which combines side-bid AI optimization with pre-bid filtering and post-bid measurement, launched in Q3 last year and is also expanding rapidly. It is now on track to deliver $10 million of expected ACV in 2026.

受惠於在 Meta 的持續成功，社群衡量年增 23%，相較第四季明顯加速。社群啟動是我們成長最快的解決方案組合，第一季年增 92%，高於第四季的 62%。YouTube 上的 Authentic AdVantage 結合了旁路出價（side-bid）的 AI 最佳化、投放前（pre-bid）過濾與投放後（post-bid）衡量，於去年第三季推出，目前也在快速擴張。目前預計在 2026 年可帶來 1,000 萬美元的預期 ACV。

CTV measurement impression volumes also grew, up 28% in the quarter, and our ABS-enabled streaming TV pre-bid do not air lists entered general availability in January, with three top 15 customers representing hundreds of millions in CTV spend implementing these DV-only streaming TV controls. DV continues to break new ground in the drive towards greater transparency in streaming TV. AI measurement tools like SlopStopper, which is now available on YouTube, an AI Agent ID are showing meaningful engagement rates. Our AI SlopStopper measurement solution for mobile and online video and display is already applied to over 40% of measured impressions. And the pre-bid tool is being tested by six of our largest advertisers.

CTV 衡量的曝光量在本季也成長 28%；我們由 ABS 支援的串流電視投放前「不播出清單」（pre-bid do not air lists）於 1 月正式全面上市，其中有三家前 15 大客戶、代表數億美元的 CTV 支出，正在導入這些僅 DV 提供的串流電視控管機制。DV 持續在推動串流電視更高透明度方面開創新局。像 SlopStopper 這類 AI 衡量工具（目前已在 YouTube 上提供）以及 AI Agent ID，皆展現出具意義的互動率。我們針對行動端與線上影音、展示廣告的 AI SlopStopper 衡量解決方案，已套用於超過 40% 的已衡量曝光。而投放前工具正由我們六家最大廣告主進行測試。

Our mid-term goal remains to increase the contribution of social, streaming TV, and AI-driven solutions from under 30% of total revenue today to approximately 50%. As we drive this evolution, our mobile and online video and display business remained stable in Q1 with approximately two-thirds of impressions that we engaged with delivered on mobile in-app and mobile web environments. We remain focused on creating a revenue mix that closely aligns with the fastest-growing global digital ad sectors.

我們的中期目標仍是將社群、串流電視與 AI 驅動解決方案的營收貢獻，從目前占總營收不到 30%，提升至約 50%。在推動此一轉型的同時，我們的行動端與線上影音、展示廣告業務在第一季維持穩定；我們所觸及的曝光中約有三分之二來自行動 App 內與行動網頁環境。我們仍專注於打造與全球成長最快的數位廣告領域高度一致的營收組合。

DV continues to drive new revenue opportunities. Distance ourselves from competition and create meaningful margin expansion through AI efficiencies and product innovation. AI solutions, social activation tools, and streaming TV quality solutions are positively impacting our customers' ad performance and building a foundation for TAM and market share expansion for DoubleVerify.

DV 持續開拓新的營收機會。透過 AI 效率與產品創新，我們與競爭對手拉開距離，並創造具意義的利潤率擴張。AI 解決方案、社群啟動工具與串流電視品質解決方案，正面提升客戶的廣告成效，並為 DoubleVerify 的 TAM 與市占率擴張奠定基礎。

Shifting focus to the role that AI is playing in the ongoing expansion of our product-led growth cycle, we continue to lean into AI to operate more efficiently, launch products faster, and improve margins. And as the emerging AI advertising universe evolves, it is creating new revenue opportunities that expand our TAM as we extend our essential role in this burgeoning environment.

將焦點轉向 AI 在我們產品驅動成長循環持續擴張中的角色，我們持續加大 AI 的運用，以更高效率營運、更快推出產品並改善利潤率。同時，隨著新興的 AI 廣告宇宙演進，它正在創造新的營收機會，擴大我們的 TAM，並讓我們在這個快速成長的環境中延伸不可或缺的角色。

Regarding this new environment, we've identified three main areas where DV has the largest AI growth opportunities and which we are already seeing traction with the customers. First, the agentic buying and selling of media, where we are building new products, connecting with and leading the development of the numerous protocols that will help advertisers lean into AI-based buying.

針對這個新環境，我們已辨識出 DV 在 AI 方面最大的三個成長機會，且我們已看到客戶端的採用動能。第一，是媒體的代理式（agentic）買賣，我們正在打造新產品，並參與且引領多項協定的發展，協助廣告主採用以 AI 為基礎的採購方式。

Second, we are empowering advertisers to navigate the dynamic AI-impacted advertising landscape as AI cyber fraud and AI content swap becomes prolific.

第二，當 AI 網路詐欺與 AI 內容置換日益猖獗之際，我們正協助廣告主在受 AI 影響、快速變動的廣告環境中順利應對。

And third, we are digging into the massive potential ad market on LLM chatbots, where many of our current advertisers are beginning to deploy their marketing dollars [if] had little in the way of transparency and independent measurement.

第三，我們正深入探索 LLM 聊天機器人上的龐大潛在廣告市場；許多現有廣告主已開始投入行銷預算，但在透明度與獨立衡量方面[如果]仍相當有限。

Let me talk briefly about each one of these opportunities.

我簡要談談這三項機會。

First, we are focused on establishing security and trust in the agentic advertising ecosystem. Trust has always been essential in our industry, and we recently joined the Ad Context Protocol, AdCP, a coalition of ad-tech companies established by agentic advertising organizations to define standards for ad buying and selling by AI agents. According to eMarketer, about two-thirds of ad buyers plan to focus more time on agentic ad buying this year. While in our early days, we are actively engaged to make sure DV is at the forefront of establishing standards that will continue to preserve trust and transparency for its advertisers wherever they choose to deploy their advertising investments.

首先，我們聚焦於在代理式廣告生態系中建立安全與信任。信任一直是本產業的關鍵要素；我們近期加入了廣告情境協定（Ad Context Protocol，AdCP），這是一個由代理式廣告組織所建立、由多家廣告科技公司組成的聯盟，旨在制定 AI 代理進行廣告買賣的標準。根據 eMarketer，約三分之二的廣告採購方計畫今年將更多時間投入代理式廣告採購。雖然仍處於早期階段，我們正積極參與，確保 DV 站在制定標準的最前線，持續為廣告主在其選擇投放廣告投資的任何地方維持信任與透明度。

As with all of our engagements, we remain independent and agnostic and the way we operate in the agentic advertising world will be the same, with the ability to plug into any agentic protocol from the IAB framework to platform-specific systems that are important to our customers.

如同我們所有合作一樣，我們將維持獨立與中立（agnostic）；我們在代理式廣告世界中的運作方式也將一致，具備可接入任何代理式協定的能力，從 IAB 架構到對客戶重要的平台特定系統皆可。

Second, we are expanding tools to protect ad investments from AI-fueled challenges. We continue to enhance our market-leading suite of AI tools that combat the increasing challenges of navigating AI slop and avoiding AI cyber fraud. With the launch of DV AI SlopStopper for Social, we've expanded our capability for advertisers to avoid low-quality, AI-generated content on YouTube and will broaden our coverage to other walled gardens in the coming quarters.

第二，我們正在擴展工具，以保護廣告投資免受 AI 驅動的挑戰影響。我們持續強化我們市場領先的一套 AI 工具，以對抗在辨識 AI 垃圾內容（AI slop）與避免 AI 網路詐騙方面日益增加的挑戰。隨著推出 DV AI SlopStopper for Social，我們已擴大廣告主在 YouTube 上避開低品質、AI 生成內容的能力，並將在未來幾季把涵蓋範圍擴展至其他封閉式生態（walled gardens）。

Fueled by malicious AI, cyber fraud continues to become more sophisticated, threatening to challenge the ROI and efficiency gains driven by the positive use of AI. In Q1 2026, DV's Fraud Lab continued to harness AI to fight fraud as AI-powered fraud schemes proliferated at a record pace and became even more sophisticated.

在惡意 AI 的助長下，網路詐騙持續變得更加精密，可能威脅到 AI 正向應用所帶來的 ROI 與效率提升。在 2026 年第一季，隨著 AI 驅動的詐騙手法以創紀錄的速度擴散並變得更為複雜，DV 的 Fraud Lab 持續運用 AI 來打擊詐騙。

AI-powered bot schemes continue to evolve faster than ever with 140% more bot scheme variants emerging in Q1 '26 compared to Q1 '25. In parallel, app-based fraud continues to accelerate dramatically, especially across mobile and CTV, where we have classified over 1300 apps as fraudulent since the beginning of 2026.

AI 驅動的機器人（bot）詐騙手法持續以前所未有的速度演進；相較於 2025 年第一季，2026 年第一季出現的 bot 手法變體多了 140%。同時，App 型詐騙也持續大幅加速，尤其在行動端與 CTV 上；自 2026 年初以來，我們已將超過 1300 個 App 分類為詐騙。

Finally, we are focused on capitalizing on the massive potential ad market that AI chatbot marketing will represent. According to eMarketer, ad spend on LLMs is forecasted to grow by over $25 billion by 2029, with ad spend expected to cannibalize over 14% of search spend, a $400 billion market that DV has historically not been able to access.

最後，我們聚焦於把握 AI 聊天機器人行銷所代表的龐大廣告市場潛力。根據 eMarketer，LLM 的廣告支出預計到 2029 年將成長超過 250 億美元，且預期將侵蝕超過 14% 的搜尋廣告支出；搜尋是一個 4,000 億美元的市場，而 DV 過去一直無法觸及。

OpenAI recently shared that they expect to generate $100 billion in advertising revenue by 2030, underscoring just how the market may be moving even more rapidly than analysts are predicting. As has been the case for the open web, mobile, streaming, and social environments, unbiased, independent measurement will play a key role in engendering the advertiser trust needed for this new ecosystem to thrive.

OpenAI 近期分享，他們預期到 2030 年將創造 1,000 億美元的廣告營收，凸顯市場的推進速度可能比分析師預測的更快。如同開放網路、行動、串流與社群環境的發展經驗，公正、獨立的衡量將在建立廣告主所需的信任方面扮演關鍵角色，讓這個新生態得以蓬勃發展。

While AI platform ad models continue to evolve, advertiser demands remain the same, ensuring ad transactions are trusted and transparent, and ads are viewable, brand suitable, and delivered to legitimate traffic within authentic content environments. Our enterprise customers and agency platforms have made it clear to us that expanding beyond test budgets in AI environments will require even greater transparency and trust than is present today.

儘管 AI 平台的廣告模式持續演進，廣告主的需求仍然相同：確保廣告交易可信且透明，且廣告可視、符合品牌適配性，並在真實內容環境中投放至合法流量。我們的企業客戶與代理商平台已明確表示，要在 AI 環境中把投放從測試預算擴大，將需要比目前更高程度的透明度與信任。

We are confident that, as we have shown on social and streaming platforms, our role as an essential trust layer will extend to this new ecosystem, and we are engaged in discussions with several LLMs who are leaning into ad-supported models. As AI drives digital advertising to become more automated, agentic, and opaque, and as AI slop becomes a must-avoid content category for advertisers; the need for independent verification, protection, and performance measurement has never been greater. Regardless of platform, buying mode, or message, DV will be an integral, trusted part of the ad equation.

我們有信心，正如我們在社群與串流平台所展現的，我們作為關鍵信任層（trust layer）的角色也將延伸至這個新生態；我們也正與數家傾向採用廣告支持模式的 LLM 進行討論。隨著 AI 推動數位廣告變得更自動化、更具代理性（agentic）且更不透明，以及 AI 垃圾內容成為廣告主必須避開的內容類別；對獨立驗證、防護與成效衡量的需求從未如此迫切。無論平台、購買模式或訊息為何，DV 都將成為廣告方程式中不可或缺且值得信賴的一環。

Moving to social verification, the social sector remains our fastest-growing business segment and is a core driver of our next phase of growth. No other verification or measurement provider has more innovative solutions for advertisers seeking to protect their spend on social platforms and ensure it performs. Social activation accelerated meaningfully to over 90% year-over-year growth in the first quarter, up from around 60% growth in Q4.

談到社群驗證，社群領域仍是我們成長最快的業務區塊，也是推動我們下一階段成長的核心動能。沒有任何其他驗證或衡量供應商，能為希望在社群平台上保護支出並確保成效的廣告主提供更多創新解決方案。第一季社群啟用（activation）顯著加速，年增超過 90%，高於第四季約 60% 的成長。

This acceleration was driven by continued scaling of our social pre-bid solutions, elevated by enhanced product capabilities on Meta as well as expanded capabilities across TikTok and YouTube. 87 advertisers have now utilized Meta activation since launch, up from 68 in the fourth quarter, with 31 of these customers coming from our TOP100 clients.

這一加速主要來自我們社群 pre-bid 解決方案的持續擴張，並受惠於 Meta 上更強的產品能力，以及在 TikTok 與 YouTube 上擴大的功能覆蓋。自推出以來，已有 87 位廣告主使用 Meta 啟用方案，高於第四季的 68 位，其中有 31 位客戶來自我們的 TOP100 客戶。

As of the end of the first quarter, our Meta activation product was already at a $12 million annualized run rate. On YouTube, DV Authentic AdVantage has seen strong customer adoption. Some of our largest CPG customers have started scaling on the solution, driven by the significant ROI improvements that it delivers. Through the combination of Cybit's AI optimization with social pre-bid filtering and post-bid measurement, DV Authentic AdVantage customers have seen their media CPMs decline by as much as 36%, while reach has expanded by 64% and brand suitability integrity remains strong.

截至第一季末，我們的 Meta 啟用產品年化營收運行率（annualized run rate）已達 1,200 萬美元。在 YouTube 上，DV Authentic AdVantage 已獲得強勁的客戶採用。我們部分最大的 CPG 客戶已開始擴大使用該解決方案，主要受其帶來的顯著 ROI 改善所驅動。透過 Cybit 的 AI 最佳化、社群 pre-bid 篩選與 post-bid 衡量的組合，DV Authentic AdVantage 客戶的媒體 CPM 最多下降 36%，同時觸及（reach）擴大 64%，且品牌適配性完整性仍維持強勁。

As with DV's Meta pre-bid solution, we are just starting to scratch the surface with the impact that Authentic AdVantage can have on our customers' business and our growth profile, and we're excited about the significant opportunities ahead for both products.

如同 DV 的 Meta pre-bid 解決方案，我們才剛開始觸及 Authentic AdVantage 對客戶業務與我們成長輪廓所能帶來影響的表面；我們對兩項產品未來的重大機會感到振奮。

As mentioned previously, our social suite of tools are ramping, and we recently announced the expansion of DV AI Verification to include DV's AI SlopStopper for Social. This new industry-leading offering is designed to help advertisers navigate the growing challenges posed by low-quality AI-generated content and safeguard brand reputation across social and video-centric environments, starting with YouTube. DV's AI SlopStopper for Social is another DV tool that empowers advertisers to ensure their brand investment is protected wherever they spend while driving stronger media outcomes.

如先前所述，我們的社群工具套件正在加速爬坡；我們近期也宣布擴展 DV AI Verification，納入 DV 的 DV AI SlopStopper for Social。這項全新、業界領先的產品旨在協助廣告主因應低品質 AI 生成內容帶來的日益嚴峻挑戰，並在以社群與影音為核心的環境中（從 YouTube 開始）守護品牌聲譽。DV 的 DV AI SlopStopper for Social 是另一項 DV 工具，讓廣告主無論在哪裡投放，都能確保品牌投資受到保護，同時帶來更強的媒體成效。

Additionally, in the quarter, we expanded brand suitability coverage across Snapchat's Discover Feed format, enabling our advertisers to have complete coverage across Snap Discover tiles placements. And we recently announced that we achieved Media Rating Counsel, or MRC, accreditation for TikTok video viewability becoming the first measurement vendor to receive the accreditation. As advertising investment continues to grow across video-centric social platforms like TikTok, independent verification plays a critical role in ensuring transparency and accountability.

此外，本季我們擴大了在 Snapchat Discover Feed 格式上的品牌適配性覆蓋，讓廣告主能在 Snap Discover 圖塊版位上獲得完整覆蓋。我們也近期宣布，我們取得了 Media Rating Counsel（MRC）對 TikTok 影片可視率（viewability）的認證，成為首家獲得該認證的衡量供應商。隨著廣告投資持續在 TikTok 等以影音為核心的社群平台成長，獨立驗證在確保透明度與問責性方面扮演關鍵角色。

And with accredited measurement informed by tens of trillions of historical ad transactions, advertisers can now evaluate campaign effectiveness with greater confidence and ensure their media investments deliver real value. This milestone underscores our commitment to delivering the highest standards of measurement accuracy and transparency and further demonstrates the company's alignment with the MRC accreditation process as a critical layer of accountability in digital advertising.

而在經認證的衡量、並由數十兆筆歷史廣告交易所提供的資訊支持下，廣告主如今能以更高信心評估活動成效，並確保媒體投資帶來真實價值。這一里程碑凸顯我們致力於提供最高標準的衡量準確性與透明度，也進一步展現公司與 MRC 認證流程的一致性，將其視為數位廣告問責性的關鍵一層。

Turning to streaming TV, we continue to deliver product innovation to address advertiser demand for independent transparency and increasing fraud in streaming environments. Our continued product innovations helped grow CTV measurement volumes by 28% year over year this quarter.

轉向串流電視（streaming TV），我們持續推出產品創新，以回應廣告主對獨立透明度的需求，以及串流環境中日益增加的詐騙。我們持續的產品創新，協助本季 CTV 衡量量（measurement volumes）年增 28%。

We've already begun to see solid adoption of ABS Do Not Air list from eight of our largest advertisers as well as strong interest in our authentic streaming TV solution. And in this quarter, we announced that Spectrum Reach became the first partner to join DV's certified transparent streaming program, reinforcing its commitment to secure program-level transparency across streaming TV inventory.

我們已開始看到八家最大廣告主對 ABS Do Not Air 名單的良好採用，同時也對我們的 authentic streaming TV 解決方案展現強烈興趣。此外，本季我們宣布 Spectrum Reach 成為首個加入 DV 認證透明串流（certified transparent streaming）計畫的合作夥伴，強化其在串流電視庫存上確保節目層級透明度的承諾。

Spectrum Reach will share key show-level data across their programming, including news and live sports, spanning both direct I/O and programmatic buying. These insights will be available directly within DV authentic streaming TV reporting, giving advertisers verified post-bid visibility into the specific programs their ads ran alongside.

Spectrum Reach 將分享其節目編排中的關鍵節目層級（show-level）資料，包括新聞與現場體育賽事，涵蓋直接 I/O 與程式化購買。這些洞察將可直接在 DV authentic streaming TV 報告中取得，讓廣告主在 post-bid 後獲得經驗證的可視性，了解其廣告具體與哪些節目同播。

By combining real, not implied or aggregated, show-level transparency in a privacy-focused way with DV's performance analytics and optimization capabilities, advertisers can now better understand how contextual relevance drives outcomes and makes smarter decisions to optimize future streaming investments.

透過以重視隱私的方式結合真實、非推估或彙總的節目層級透明度，以及 DV 的成效分析與最佳化能力，廣告主如今能更好理解情境相關性如何驅動成果，並做出更明智的決策以最佳化未來的串流投資。

This is just the start of our drive to deliver granular, unaggregated, show-level transparency across all streaming environments, and we are seeing momentum from additional platforms to join our certified transparent streaming program. The results of our innovation leadership are clear. We are growing client engagements and winning deals with new solutions where there aren't any competitors.

這只是我們推動在所有串流環境中提供細緻、未彙總的節目層級透明度的開始；我們也看到更多平台加入我們認證透明串流計畫的動能。我們在創新領導力上的成果十分明確。我們正在擴大客戶合作深度，並以沒有競爭者的新解決方案贏得新案。

We work with over 340 advertisers now, generating more than $200,000 annually. And our unique solution underscored a 77% greenfield win ratio in Q1, meaning that we're winning deals with solutions in new areas in which there are no competitive incumbents to displace. Investment in innovation continues to be DV's secret sauce to get stickier with our customers, win new deals, and gain market share.

我們目前與超過 340 家廣告主合作，每年創造超過 200,000 美元的收入。而我們獨特的解決方案在第一季凸顯了 77% 的綠地（greenfield）勝出比率，這意味著我們正在以解決方案贏得全新領域的交易，在那些領域中沒有需要被取代的既有競爭者。對創新的投資持續是 DV 的祕密武器，讓我們與客戶的黏著度更高、贏得新案子並提升市占率。

And AI is enabling us to innovate more efficiently than ever as we continue to expand margins while launching and expanding the tools that cement our role as the essential trust layer for buyers and sellers of digital media. With strong execution in the first quarter, we're leaning hard into AI-powered innovation that will continue to extend our leadership position. Looking ahead to the rest of the year, we remain focused on product development acceleration, partner expansion and market share growth and continued strong margins and cash flow.

而 AI 正在讓我們以前所未有的效率進行創新，因為我們在推出並擴展各項工具的同時，也持續擴大利潤率；這些工具鞏固了我們作為數位媒體買賣雙方不可或缺的信任層角色。憑藉第一季強勁的執行力，我們正大力投入 AI 驅動的創新，這將持續延伸我們的領導地位。展望今年剩餘時間，我們仍聚焦於加速產品開發、擴大合作夥伴版圖與市占成長，以及持續維持強勁的利潤率與現金流。

With that, let me turn the call over to Nicola.

接下來，我把電話交給 Nicola。

Thanks, Mark, and good afternoon, everyone.

謝謝你，Mark，各位下午好。

For the first quarter, we achieved 10% year-over-year revenue growth and 31% EBITDA margins. Off the strong start to the year, we're reiterating guidance for the full year. For the first quarter, total revenue was $181 million, representing 10% year-over-year growth. Total advertiser revenue, which includes activation and measurement, represented 90% of total revenue and grew 9% year-over-year, driven by 12% growth in volume, or MTM, partially offset by a 4% decline in fees or MTF.

第一季我們達成營收年增 10%，以及 31% 的 EBITDA 利潤率。在今年強勁開局之後，我們重申全年指引不變。第一季總營收為 1.81 億美元，年增 10%。廣告主總營收（包含啟用與衡量）占總營收的 90%，年增 9%；主要由量（或 MTM）成長 12% 所帶動，但部分被費率（或 MTF）下滑 4% 所抵銷。

Activation revenue grew 6%, with ABS representing 53% of activation revenue in the quarter. As of quarter end, over 75% of our top 500 clients were using ABS. Measurement revenue grew 16% year-over-year, with social measurement revenue increasing 23% and representing 49% of measurement revenue, and international revenue increasing 18% and representing 27% of measurement revenue.

啟用（Activation）營收成長 6%，其中 ABS 在本季占啟用營收的 53%。截至季末，我們前 500 大客戶中有超過 75% 正在使用 ABS。衡量（Measurement）營收年增 16%；其中社群衡量營收成長 23%，占衡量營收的 49%；國際營收成長 18%，占衡量營收的 27%。

Supply-side revenue represented 10% of total revenue in the quarter and grew 12% year-over-year. We're driving growth by adding new CTV and digital platform partnerships and by continuing to expand DV solutions on retail media networks.

供給端（Supply-side）營收在本季占總營收的 10%，年增 12%。我們透過新增 CTV 與數位平台合作夥伴，以及持續在零售媒體網路上擴展 DV 解決方案來推動成長。

Moving to expenses, in the first quarter, we delivered 82% revenue less cost of sales. Our continued investments and use of AI capabilities are allowing us to scale at a consistently efficient rate, even as we measure increasing levels of volume. We delivered $55 million of adjusted EBITDA, representing a 31% margin as compared to 27% margin in Q1 of 2025. Total expenses for product development, sales and marketing, and G&A increased 2% as compared to 10% revenue growth. We are showing early signs of the benefit of using AI capabilities to grow through improved productivity across the organization and increased software capitalization related to product development.

接著談費用。第一季我們達成「營收扣除銷貨成本」為 82%。我們持續的投資與 AI 能力的運用，使我們即使在衡量的量能持續提升之際，仍能以一貫高效率的速度擴張規模。我們交出 5,500 萬美元的調整後 EBITDA，利潤率為 31%，相較於 2025 年第一季的 27% 利潤率有所提升。產品開發、銷售與行銷，以及一般與行政（G&A）的總費用增加 2%，相較之下營收成長 10%。我們已看到初步跡象，顯示運用 AI 能力可透過全組織生產力提升，以及與產品開發相關的軟體資本化增加，來帶動成長。

We are scaling the business more efficiently, which results in increasing EBITDA margins. Stock-based compensation was $24 million in the first quarter, flat to prior year. For the second quarter, we expect stock-based compensation of approximately $25 million to $27 million and weighted-average fully designated shares outstanding of approximately 157 million shares. For the full year, we continue to expect stock-based compensation to range between $102 million to $107 million, a decline year-over-year, reflecting the impact of our updated equity incentive plan that reduced the annual value of equity grants in 2026 by over 40% as compared to 2025.

我們正以更高效率擴大業務規模，進而帶動 EBITDA 利潤率提升。第一季以股票為基礎的薪酬為 2,400 萬美元，與去年同期持平。第二季我們預期以股票為基礎的薪酬約為 2,500 萬至 2,700 萬美元，加權平均完全稀釋流通股數約為 1.57 億股。全年方面，我們仍預期以股票為基礎的薪酬將介於 1.02 億至 1.07 億美元，較去年下降；反映我們更新的股權激勵計畫之影響，該計畫使 2026 年股權授予的年度價值相較 2025 年降低超過 40%。

Turning to cash, year-to-date, we have repurchased 9.8 million shares for $100 million, of which 7.3 million shares were repurchased in the first quarter for approximately $75 million and 2.5 million shares were repurchased in April for approximately $25 million. Year-to-date, the 9.8 million shares were repurchased represent approximately 6% of fiscal year-end 2025 outstanding shares.

再來談現金。截至目前，我們已以 1 億美元回購 980 萬股，其中第一季回購 730 萬股、約 7,500 萬美元；4 月回購 250 萬股、約 2,500 萬美元。截至目前回購的 980 萬股，約占 2025 會計年度年末流通股數的 6%。

Net cash from operating activities in the first quarter was $4 million and was impacted by timing of collections and payments at the end of the quarter. For the full year, we expect free cash flow conversion of approximately 60%. We ended the first quarter with approximately $174 million in cash and no long-term debt.

第一季營運活動淨現金流入為 400 萬美元，主要受到季末收款與付款時點影響。全年我們預期自由現金流轉換率約為 60%。第一季結束時，我們持有約 1.74 億美元現金，且沒有長期負債。

Now turning to guidance. For the second quarter of 2026, we expect revenue to range between $199 million and $205 million, representing a year-over-year increase of approximately 7% at the midpoint. As a reminder, we're lapping our 21% growth rate in Q2 of 2025. And we expect adjusted EBITDA to range between $63 million to $67 million, representing a 32% adjusted EBITDA margin at the midpoint.

接著是財測指引。對於 2026 年第二季，我們預期營收介於 1.99 億至 2.05 億美元，中位數約代表年增 7%。提醒一下，我們正在與 2025 年第二季 21% 的成長率相比較（lap）。我們預期調整後 EBITDA 介於 6,300 萬至 6,700 萬美元，中位數的調整後 EBITDA 利潤率為 32%。

For the full year 2026, we are reiterating our prior guidance. We expect revenue to range between $810 million and $826 million, representing an 8% to 10% year-over-year increase and expect adjusted EBITDA margins of approximately 34%.

對於 2026 全年，我們重申先前的指引。我們預期營收介於 8.10 億至 8.26 億美元，年增 8% 至 10%，並預期調整後 EBITDA 利潤率約為 34%。

As discussed on our prior call, incremental growth in 2026 will be driven by three product-led growth engines: first, continued adoption of our solutions across social and streaming TV; second, growth from existing enterprise clients scaling our product offering; and third, continued new customer acquisition driven by DV's differentiated products. Our first quarter results demonstrate progress on each growth driver with increasing social activation revenue growth, increased adoption and scaling of new products and a consistently high win rate. Our first quarter results show solid execution. With a clear focus on durable growth and expanding profitability, we're well positioned to continue to deliver long-term shareholder value.

如同我們在上一通電話會議中所討論，2026 年的增量成長將由三個產品驅動的成長引擎帶動：第一，持續在社群與串流電視上採用我們的解決方案；第二，既有企業客戶擴大採用我們的產品組合所帶來的成長；第三，DV 具差異化的產品持續帶動新客戶取得。我們第一季的結果顯示在各項成長動能上都有進展，包括社群啟用營收成長加速、新產品採用與擴張提升，以及持續維持高勝率。第一季結果展現了穩健的執行力。在明確聚焦於可持續的成長與獲利能力擴張之下，我們已具備良好條件，持續為股東創造長期價值。

And with that, we will open up the line for questions. Operator, please go ahead.

接下來，我們開放提問。接線員，請開始。

(Operator Instructions)

（接線員指示）

Matt Swanson, RBC Capital Markets.

RBC Capital Markets 的 Matt Swanson。

Great, thank you so much, and congrats on a solid start to the year.

很好，非常感謝，也恭喜你們今年有個穩健的開局。

I think I'll pick up right where Nicola left off there. So I mean, it was a great quarter for social measurement, but I mean, really focusing on that growth opportunity and social activation. Any time you have something with over a 90% growth rate, we should probably start there. Can you just kind of give us an update on how things are trending and kind of how the rollout's been going relative to your expectations with your customers, especially on Meta?

我想我就從 Nicola 剛剛提到的地方接著問。社群衡量這一季表現非常好，但我想更聚焦在那個成長機會以及社群啟用。任何一個成長率超過 90% 的項目，我們大概都應該先從那裡開始。你能否更新一下目前的趨勢，以及相較於你們對客戶的預期，推廣（rollout）的進展如何，特別是在 Meta 上？

Yeah, thanks for the question, Matt.

好的，謝謝你的問題，Matt。

So obviously, we're really pleased with the scale and -- scaling and speed of scaling on social activation, which is really being driven by all three prongs. The first, as you mentioned, Meta activation and the new Meta pre-bid tools we have there. The second, growth in YouTube through our Authentic AdVantage solution. And third, through TikTok as well which continues to grow at a really nice pace on the pre-bid side. So our social activation business is really running on all cylinders.

很明顯地，我們對社群啟用的規模——以及規模擴張的速度——感到非常滿意，而這主要是由三個面向共同驅動。第一個如你所提到的，是 Meta 啟用以及我們在那裡推出的新 Meta 競價前（pre-bid）工具。第二個，是透過我們的 Authentic AdVantage 解決方案在 YouTube 的成長。第三個，也包括 TikTok；其競價前（pre-bid）端的成長仍以非常不錯的速度持續推進。因此，我們的社群啟用業務可說是火力全開、全面運轉。

And one of the things I think this really underscores is the fact that our solutions are really playing an essential role in the walled gardens. I think there were some questions about whether or not, hey, is DV as powerful or as needed in the walled gardens that is in the open web and the growth of this is proving that out. I think it also underscores the power of the pre-bid kind of post-bid engine where we can launch pre-bid solutions where we already have measurement in place, we see a nice catalyst for growth.

而我認為這真正凸顯的一點是，我們的解決方案在封閉生態（walled gardens）中確實扮演著不可或缺的角色。我想之前有人質疑，嘿，DV 在封閉生態裡是否像在開放網路一樣強大或同樣必要，而這項業務的成長正在證明答案。我也認為這凸顯了 pre-bid 與 post-bid 引擎的力量：我們可以在既有量測已就緒的情況下推出 pre-bid 解決方案，並看到一個很好的成長催化劑。

So it's scaling well. Meta in particular now is over 80 clients have engaged with it. Some of our biggest advertisers are now engaged there. We mentioned, I think in the last call, we wanted to get to a $15 million ARR by the end of '27 or '26 -- and we're already at $12 million, so like this is growing well. I think it's scaling well, and again, it's definitely being driven by Meta, but it's also that social activation 90% growth number is being supported by our innovations in TikTok and YouTube as well.

所以它的擴張表現很好。尤其是 Meta，目前已有超過 80 家客戶參與。我們一些最大的廣告主現在也在那裡啟用。我們提到過，我想在上一通電話會議中，我們希望在 27 年或 26 年底達到 1,500 萬美元的 ARR——而我們現在已經到 1,200 萬美元，所以成長得很好。我認為它擴張得很好，而且再次強調，雖然明顯是由 Meta 帶動，但社群啟用（social activation）90% 的成長數字也同樣受到我們在 TikTok 與 YouTube 的創新所支撐。

Great. And then, Nicola, just kind of thinking more on the guidance side, I know over the last couple of years, you've become less exposed to kind of CPG and retail sectors, but if there's anything from kind of a macro standpoint that you would point out to us and then just any update on the advertiser who went through the agency change last quarter?

很好。接著，Nicola，從財測指引的角度再多想一下，我知道過去幾年你們對 CPG 與零售類別的曝險降低了，但從總體環境的角度，有沒有什麼你會特別提醒我們注意的？另外，關於上季經歷代理商更換的那位廣告主，有沒有任何更新？

Yeah, so I'll start by saying from a vertical perspective.

好的，我先從產業別（vertical）的角度來說。

We spoke a lot about retail and the impact it had on our business for the end of the year. That has normalized as we had expected, we already spoke about it on our last call. And overall, all of our key verticals showed growth in the first quarter of 2026. And we haven't seen material changes across the verticals. If anything, we've been able to diversify further into healthcare and technology, which allows us to not be as reliant on retail and CPG. So it feels more normalized, and it feels that we are continuing to diversify in a way that's going to help to create a more predictable business for us.

我們談了很多零售，以及它在年底對我們業務造成的影響。那已如我們預期地回歸常態，我們在上一通電話會議中也已經談過。整體而言，我們所有主要產業別在 2026 年第一季都呈現成長。而且我們沒有看到各產業別出現重大的變化。如果要說的話，我們進一步多元化到醫療保健與科技，讓我們不必那麼依賴零售與 CPG。所以感覺更正常化了，也感覺我們正持續以有助於打造更可預測業務的方式進行多元化。

In terms of macro, if you look at the verticals, I'll address the two verticals that are generally mentioned in terms of uncertainty based on what's happening in the macro for us, auto and travel are fairly small verticals, so we're not as exposed to those and just a general comment on macro. We're obviously not assuming strong tailwinds, but we're assuming an environment that's going to remain fairly stable.

就總體而言，如果看產業別，我會談兩個通常在宏觀不確定性下會被提到的產業：對我們來說，汽車與旅遊都是相對小的產業別，因此我們對它們的曝險不高；再補充一點宏觀面的總體評論。我們顯然沒有假設會有強勁的順風，但我們假設環境將維持相對穩定。

Brian Pitz, BMO Capital Markets.

Brian Pitz，BMO Capital Markets。

Mark, since I know SlopStopper is one of your favorite topics, maybe you could give us your latest expectation around penetration rate and maybe thoughts on the future opportunities around this product. And then maybe stepping back more broadly, as AI content continues to proliferate across the Internet, talk about what you're hearing from advertisers in terms of how they're adapting to this changing environment. Are they starting to get more comfortable? Maybe just a little bit more insight. Thanks.

Mark，既然我知道 SlopStopper 是你最喜歡的話題之一，也許你可以跟我們分享一下你對滲透率（penetration rate）的最新預期，以及對這個產品未來機會的一些想法。然後再更宏觀一點，隨著 AI 內容在網路上持續大量增加，談談你從廣告主那裡聽到他們如何因應這個變動中的環境。他們是否開始更自在了？也許再多給我們一點洞見。謝謝。

Thanks for the question, Brian. And yes, I love SlopStopper just because I love saying the name on these calls so much.

謝謝你的問題，Brian。是的，我喜歡 SlopStopper，純粹因為我很喜歡在這些電話會議上一直講這個名字。

As you know, we noted in the call, SlopStopper is now being applied on the measurement side to about 40% of all of our impressions. That's a pretty -- it's probably one of our fastest scaling attach rates for any different category that we've seen.

如你所知，我們在電話會議中提到，SlopStopper 目前在量測端已套用到我們約 40% 的所有曝光（impressions）。這是相當——它可能是我們看過在任何不同類別中擴張最快的附掛率（attach rate）之一。

So on the measurement side, it's really being picked up pretty quickly, and that's a great thing. And it also shows, to your second part of your question, that advertisers are still just trying to figure out how they navigate this world of AI content. And SlopStopper is built for a specific reason. It's to avoid low-quality, questionable AI content, right? Not all AI content is bad, but there's certainly stuff out there that advertisers want to avoid and that's what SlopStopper helps them do.

所以在量測端，它確實被非常快速地採用，這是件好事。而且這也呼應你問題的第二部分：廣告主仍在摸索要如何在這個 AI 內容的世界中前進。SlopStopper 的設計有其特定目的。就是避免低品質、存疑的 AI 內容，對吧？並非所有 AI 內容都不好，但外面確實有一些廣告主想要避開的內容，而 SlopStopper 就是幫助他們做到這件事。

When we think about kind of what's next for that tool, so we launched a pre-bid SlopStopper solution on YouTube. We're going to expand that to additional social platforms over the next few quarters. And it's the platforms that -- the social platforms, particularly around video, that are the most kind of challenging for advertisers to try to figure out where they should and shouldn't advertise against. So we see this as, again, another catalyst for higher attach rates for our solution and another catalyst for folks to actually lean into working with [DVD], maybe even if they're not, because it gives them another tool to use to navigate challenging content.

當我們思考這個工具接下來的方向時，我們已在 YouTube 上推出了 pre-bid 的 SlopStopper 解決方案。接下來幾季我們會把它擴展到更多社群平台。而對廣告主來說，最具挑戰、最難判斷哪些內容該投放、哪些不該投放的，正是這些社群平台——尤其是影片相關的。因此我們認為這將再次成為推升我們解決方案更高附掛率的催化劑，也會促使更多人更願意與 [DVD] 合作——甚至是原本還沒有合作的人——因為這讓他們多了一個工具來應對具挑戰性的內容。

And I think the other thing, and as I mentioned in my earlier question, there is a growing need for our solutions within the walled gardens. Not because the content there is particularly better or worse than any other platform, but the content there is becoming really overrun with a good amount of AI content, and the advertisers really are looking for tools to help navigate that. So the problems that advertisers saw in the open web are evolving to different types of challenges behind walled gardens, and the great thing is that we've got solutions to address all of them.

我認為另外一點，而且如我先前提到的，在封閉生態中對我們解決方案的需求正在增加。並不是因為那裡的內容特別比其他平台更好或更差，而是那裡的內容正被相當多的 AI 內容所淹沒，廣告主確實在尋找工具來協助他們應對。因此，廣告主在開放網路上看到的問題，正在演變成封閉生態內不同類型的挑戰；而很棒的是，我們有解決方案可以處理所有這些問題。

Maria Ripps, Canaccord.

Maria Ripps，Canaccord。

Mark, you said that you were in discussions with several LLMs regarding sort of very fine ads on their platforms. Is this the same brand safety stack that you offer today or will the agentic ad environment require sort of a fundamentally different product? And how are you thinking about sort of pricing, create sort of structuring pricing models for agentic ads?

Mark，你說你正在與幾個 LLM 討論在其平台上投放某種非常精細的廣告。這會是你們今天提供的同一套品牌安全（brand safety）技術堆疊（stack）嗎？還是說，代理式（agentic）的廣告環境會需要本質上不同的產品？另外，你們如何思考定價，也就是為代理式廣告建立／設計定價模型？

Thanks for the question, Maria, and it's a really great one.

謝謝你的問題，Maria，這個問題非常好。

So when we think about what's going on with the ad-supported LLMs, what we've seen so far, and even as recently as this week, there's been announcements that OpenAI is embracing third-party ad solutions, right, from demand solutions and creative optimization through companies like Cargo and PacVue and Smartly. The next step for them, as we've noted, is really to start looking at measurement and verification, and we're leaning into discussions with lots of different LLMs on that front.

當我們思考廣告支持型（ad-supported）的 LLM 正在發生什麼時，就我們目前所見——甚至就在本週——已經有公告指出 OpenAI 正在採用第三方廣告解決方案，對吧？從需求端解決方案到創意優化，透過像 Cargo、PacVue 和 Smartly 這些公司。如我們所提到的，他們的下一步其實就是開始關注量測與驗證（measurement and verification），而我們也正就此與許多不同的 LLM 展開討論。

And when it comes to kind of what we think about what our solution can do there, it really is to play the same fundamental role that we do in social and in streaming and in display and online video, which is acting as a trust and transparency layer. And that has to do with everything from ensuring that ads are viewable and visible by a real human to ensuring that the context where it ends up with is aligned with what that brand and who that brand is and where they want to be.

至於我們的解決方案在那裡能做什麼，我們確實是要扮演與我們在社群、串流、展示型廣告（display）與線上影片中相同的基本角色，也就是作為信任與透明度層（trust and transparency layer）。這涵蓋從確保廣告可視（viewable）且確實被真人看到，到確保廣告最終出現的內容情境（context）與品牌定位一致、符合該品牌是誰以及它希望出現在哪裡。

So from a general thesis perspective, the application of our solutions in that environment is pretty much the same. It's building trust between the buyer and the seller. When it comes to the business model around it, we've got lots of different business models when we employ with platforms, but almost all of them are advertiser-paid and they're based on volume of engagement.

因此，從整體論點（thesis）的角度來看，我們的解決方案在那種環境中的應用幾乎是相同的。也就是在買方與賣方之間建立信任。至於其商業模式，我們與平台合作時有許多不同的商業模式，但幾乎都是由廣告主付費，並且以互動量（volume of engagement）為計價基礎。

So as we look at the opportunity within the LLMs, our assumption is that model will extend into there as well, which will be volume-based, advertiser paid based on the scale and level of engagement of that advertiser with the impressions on those platforms.

因此，當我們看待 LLMs 內的機會時，我們的假設是該模型也會延伸到那裡；這將是以量計價、由廣告主付費，並取決於該廣告主在那些平台上與曝光所產生的規模與互動程度。

Got it. That's very helpful.

了解。這非常有幫助。

And then just following up on SlopStopper, so as we think about the product, do you view it primarily as a retention tool, sort of bundled into existing relationships or is this a product that can be priced and sold independently?

接著追問一下 SlopStopper：當我們思考這個產品時，你們主要把它視為留存工具、某種打包在既有合作關係中的功能？還是這是一個可以獨立定價並單獨銷售的產品？

And just sort of view, you just launch it on YouTube, it's coming on other platforms, but sort of what's the realistic time frame for this product to move the needle on revenue either directly or indirectly?

另外，你們目前是在 YouTube 上推出，之後也會上其他平台；但就現實而言，這個產品要在營收上（不論直接或間接）產生明顯貢獻的時間表大概是什麼？

It's another great question. So I think of SlopStopper really as two, I think, two positive impacts on our business. The first, as you noted, is a retention. It creates greater value in our engagements with our customers by giving them another category of content to have greater transparency on.

這又是一個很好的問題。我認為 SlopStopper 對我們的業務大致有兩個、我想是兩個正向影響。第一個，如你所說，是留存。它透過讓客戶對另一類內容有更高的透明度，為我們與客戶的合作互動創造更大的價值。

So for current customers who already are using our measurement, I think it gives us kind of a stickier, more ingrained relationship. That is always helpful as we go back to renegotiate deals and look at pricing increases down the road, right? So that's one.

因此，對於已經在使用我們衡量服務的現有客戶，我認為它讓我們的關係更黏著、更深入。這在我們回頭重新談判合約、以及未來評估調價時，總是很有幫助，對吧？這是其一。

The second is it helps with attach rate, right? So it helps advertisers who maybe, for example, aren't using us on pre-bid on YouTube, but now knowing that they can avoid AI slop by using our solution that drives up attach rates for our solutions there. So I think it has, again, a two-pronged impact, the way it currently is structured. A, it helps us retain and grow our relationships with current customers, but it also drives up attach rate.

第二個是它有助於提升附加率（attach rate），對吧？例如，它能幫助那些可能尚未在 YouTube 的 pre-bid 使用我們服務的廣告主；但現在知道透過我們的解決方案可以避開 AI slop，這就會提高我們在那裡的解決方案附加率。所以我認為它在目前的架構下，確實有雙重影響。A，它幫助我們留住並擴大與現有客戶的合作關係，同時也推升附加率。

If there's a third aspect as well, it is a differentiator in the market, right? It is a unique solution for DV that our competitors don't have. It allows us to win deals on a greater scale than we would if we didn't have it. So there's a direct impact to kind of the new customers that we are able to bring on as well.

如果再說第三個面向，它也是市場上的差異化因素，對吧？這是 DV 的一個獨特解決方案，是競爭對手所沒有的。它讓我們能以更大的規模贏得交易，否則如果沒有它，我們可能做不到。因此，它也會對我們能帶來的新客戶產生直接影響。

I think we're already seeing -- and so the last part of your question is like when does it create a financial impact on the business? I think we're already starting to see that a little bit. When you see kind of our measurement growing at 16% which is a great rate for us there, when you see our social activation growing which of which AI SlopStopper is part of social activation growing up 92%. Those numbers are fueled by features like this, which are unique to DV, which drive attach and create new customer engagements.

我認為我們已經開始看到——你問題的最後一部分是：它何時會對公司產生財務上的影響？我覺得我們已經開始看到一些了。當你看到我們的衡量業務成長 16%（對我們來說是很好的增速），以及你看到我們的社群啟動（social activation）成長——其中 AI SlopStopper 是社群啟動的一部分——成長到 92%。這些數字是由像這樣的功能所帶動；它們是 DV 獨有的，能推升附加並創造新的客戶合作互動。

Andrew Marok, Raymond James.

Andrew Marok，Raymond James。

Maybe just one for me on the chatbot surface again. Obviously, we've seen OpenAI kind of evolve its offering from a CPM to a CPC over time. And of course, nothing is finished yet. But from the architecture of the chatbot ad offering itself, is there anything that they could do that would be more or less advantageous for you to partner with? Are there any kind of structures that you kind of hope that they might lean toward.

我這邊再問一個關於聊天機器人版位的問題。顯然，我們看到 OpenAI 的產品隨時間從 CPM 演進到 CPC。當然，目前一切都還沒定案。但就聊天機器人廣告產品本身的架構而言，他們有沒有什麼做法會讓你們更有利或更不利於合作？有沒有任何你們希望他們傾向採用的結構？

Thanks for the question, Andrew.

謝謝你的問題，Andrew。

You made some great points. The first is that the model is evolving, and it's evolving very quickly, right, as both advertisers experiment and as the LLMs experiment with advertising, right. So they're changing pretty quickly. The current setup of the structures actually lean very well into what we do very well, which is analyzing content and context that's text-based at scale very rapidly.

你提出了一些很好的觀點。第一是這個模式正在演進，而且演進得非常快，對吧？因為廣告主在試驗，同時 LLMs 也在試驗廣告，對吧。所以它們變化得相當快。目前的結構設定其實非常契合我們最擅長的事情，也就是以極快速度在大規模下分析以文字為主的內容與情境。

So the way that the engagements are set with consumers, the predominant nature of how consumers engage with those chatbots falls kind of very nicely into what DV has done for the last 15 years, which is analyze advertising in contextual or text environments, so that current structure actually fits well to what we do.

因此，消費者互動的設定方式，以及消費者與這些聊天機器人互動的主要型態，都很自然地落在 DV 過去 15 年一直在做的事情上：在情境式或文字環境中分析廣告；所以目前的結構其實很符合我們的能力。

Our experience working with walled gardens and getting real-time feeds of content, whether it's from folks like TikTok or others, gives us kind of a really strong legacy to build upon to be able to analyze the ads in those environments. So I guess that's a long way of saying we're pretty flexible. We've built from many different types of systems from video systems to short-form video systems to text-based open web engagements.

我們與封閉式生態（walled gardens）合作、並取得即時內容串流（real-time feeds）的經驗——不論是 TikTok 或其他平台——讓我們有很強的基礎，可以在此之上建構，進而分析那些環境中的廣告。所以，長話短說，我們相當有彈性。我們已經建置過許多不同類型的系統，從影音系統、短影音系統，到以文字為主的開放網路互動。

So the way that ChatGPT or any of the platforms are set up today are relatively easily engaged with our current system and our current classification system based on what we've done in the past. So we are ready. We've built a very challenging environments before, real-time environments, unique environments based on an individual engagement and I don't think this is going to be a significant lift for us to move beyond that.

因此，ChatGPT 或任何平台今天的架構，基於我們過去所做的事情，與我們現有系統以及現有分類系統的整合相對容易。所以我們已經準備好了。我們以前也建置過非常具挑戰性的環境——即時環境、基於個別互動的獨特環境——我不認為要跨到這一步對我們會是很大的負擔。

Matthew Condon, Citizens.

Matthew Condon，Citizens。

Hi, this is Brianna on for Matthew Condon.

嗨，我是 Brianna，代替 Matthew Condon 發問。

Just a question. You've raised the Authentic AdVantage of YouTube ACV to $10 million from $8 million last quarter, and then Meta activation, I think it was now $12 million versus $8 million shares prior. Just can you help us understand the incremental growth you're seeing within these two products? How much is it coming from new advertisers or is it more so the existing advertisers ramping?

我有個問題。你們把 YouTube 的 Authentic AdVantage ACV 從上一季的 800 萬美元上調到 1,000 萬美元；然後 Meta activation 我記得現在是 1,200 萬美元，相較之前分享的 800 萬美元。可以幫我們理解這兩個產品的增量成長來源嗎？有多少來自新廣告主？還是更多是既有廣告主在加速放量？

So it's a combination of both, so it's current DV advertisers that we've upsold to this solution launching and the previous advertisers who we brought on board actually scaling. As we've noted, we went from high [60s number of engagements to 80] plus now, so we've got new customers scaling the solution on Meta prebid. The same kind of scaling is happening on Authentic AdVantage. So it is a combination of both.

兩者都有，所以既有的 DV 廣告主，我們把這個新推出的解決方案加售（upsell）給他們；同時，我們先前導入的廣告主也確實在擴大規模。如我們提到的，我們從 60 多個合作案增加到現在 80 多個，所以我們有新客戶在 Meta pre-bid 上擴大使用這個解決方案。Authentic AdVantage 也出現同樣的擴張。所以兩者兼具。

What we love about it is once advertisers get engaged, they really do stay sticky and scale with us over time. And we've got a handful of our largest advertisers kind of growing at very solid rates across both Meta and Authentic AdVantage. I mean, when you get a 90% growth rate on something like social activation, it's going to be fueled not just by new customers, but by current customers really starting to scale their business, and that's what we're seeing here.

我們喜歡的一點是，一旦廣告主開始使用，他們確實會變得很黏著，並且隨著時間與我們一起擴大規模。而且我們有少數最大的廣告主，在 Meta 與 Authentic AdVantage 兩邊都以非常穩健的速度成長。我的意思是，當像社群啟動這樣的項目有 90% 的成長率時，推動力不會只來自新客戶，也會來自現有客戶真正開始擴大他們的投放規模，而這正是我們在這裡看到的。

Got it. That's helpful.

了解。這很有幫助。

And then just on the social side, activation grew 92%. Just on activation revenue, it slowed to 6% from the 4Q number. Just can you help impact what's going on within that line item?

另外在社群這邊，activation 成長 92%。但就 activation 營收而言，它從第 4 季的數字放緩到 6%。可以幫我們說明一下這個科目裡發生了什麼嗎？

Yes. So activation for the quarter was up 6%, which equates the growth that we had in the fourth quarter of last year. Mark spoke of the social activation growth, which is a high percentage on a smaller base. The rest of the business, as we said in our remarks, has remained fairly steady in the first quarter and the majority of that business would be driven by mobile and online video and display business.

可以。本季 activation 成長 6%，這與我們去年第 4 季的成長相當。Mark 提到的社群啟動成長，是在較小的基期上呈現的高百分比成長。如我們在談話中所說，業務的其他部分在第一季維持相對穩定，而其中大部分將由行動端、線上影音以及展示型（display）業務所帶動。

Now, we are obviously focused on being able to verify and continuing to grow where the advertisers are spending. And so tied to that growth that you see on social activation is the social measurement growth of 23%. Those are the vectors that we're focused on so that we're able to continue to verify wherever the advertiser is spending.

現在，我們顯然專注於能夠驗證，並且在廣告主投放預算的地方持續成長。因此，與你在社群啟動（social activation）上看到的成長相呼應的是，社群衡量（social measurement）成長了 23%。這些就是我們聚焦的向量，好讓我們能在廣告主投放的任何地方持續提供驗證。

Mark Murphy, J.P. Morgan.

Mark Murphy，摩根大通（J.P. Morgan）。

Interested in behaviorally, what are you seeing out of the group of six or seven of the large retail and CPG companies that had started to drag on your growth rates? I think that was about a year and a half ago. And with the understanding, obviously, you would have lapsed that slowed down, is there any signaling from those companies relating to their own internals or how they're coping with commodity prices or consumer spending trends. I'm just wondering if you see any kind of different behavior there, and I have a quick follow-up.

從行為面來看，對於那六、七家大型零售與民生消費品（CPG）公司——先前開始拖累你們成長率的那一群——你們觀察到什麼？我想那大概是一年半前的事。在理解上，顯然你們已經度過了那段放緩期；這些公司是否有釋出任何訊號，關於它們自身內部狀況，或它們如何因應商品價格或消費者支出趨勢？我只是想知道你們是否看到那裡有任何不同的行為，另外我還有一個簡短的追問。

Mark, we're not seeing any different behavior than the overall vertical, right, for both CPG and retail. And as I said earlier retail in particular, we've sort of seen the spend patterns normalize after the end of last year, which is a positive for us, and then in general, across the verticals, basically all of our key verticals showed growth in the first quarter, so the performance we had was spread, we are diversifying away from CPG and retail because we have large clients that are scaling, especially in healthcare and technology.

Mark，我們沒有看到任何不同於整體垂直產業的行為，對 CPG 與零售都是如此。而且如我先前所說，特別是零售，我們在去年年底之後看到支出模式已經某種程度回歸常態，這對我們是正面的；此外，整體來看，在各個垂直產業中，基本上我們所有關鍵垂直在第一季都呈現成長，因此我們的表現是分散的。我們也正在降低對 CPG 與零售的依賴，因為我們有大型客戶正在擴張，尤其是在醫療保健與科技領域。

And so as much as we can diversify as a result of us signing larger brands in different verticals, that obviously helps the business. But to go back to the initial part of your question, what we saw in Q1 is more of a normalized pattern of spend for CPG and retail.

因此，隨著我們在不同垂直產業簽下更大型的品牌、能夠進一步分散，這顯然有助於業務。但回到你問題的前半段，我們在第一季看到的是 CPG 與零售更為常態化的支出模式。

Okay, that's encouraging.

好的，這令人鼓舞。

So my other question is, coming back to the prospect of working with the LLM providers, whether it's OpenAI or Perplexity or someone else, I'm curious when you think might be the earliest opportunity for some of those ads to maybe run in scale where they could conceivably be measured and verified in a way that would start to contribute noticeably?

那我另一個問題是，回到與 LLM 供應商合作的前景，不論是 OpenAI、Perplexity 或其他人，我很好奇你認為最早什麼時候，這些廣告可能有機會開始以規模化方式投放，讓它們可以被衡量與驗證，並開始帶來可觀的貢獻？

And also, is the push to do this coming more from the LLM providers themselves or do you think it's coming more from the brand advertisers?

另外，推動這件事的力量更多是來自 LLM 供應商本身，還是你認為更多是來自品牌廣告主？

Hey, Mark, it's a great question.

嗨，Mark，這是個很好的問題。

Things are moving very rapidly on the LLMs with regard to advertising. We've mentioned in the last call and this call that our advertisers are more than excited to test them out, but to scale budgets, and they've been very clear with us, and their agents been clear with us, to scale budgets, they've let the platforms know that they're going to need third-party measurement, they're going to need more transparency and more verification.

在 LLM 與廣告相關的進展上，事情推進得非常快。我們在上一通電話會議以及這次會議都提到，我們的廣告主非常期待進行測試；但要把預算擴大到規模化，他們對我們說得很清楚，他們的代理商也說得很清楚：要擴大預算，平台需要第三方衡量，需要更高的透明度與更多的驗證。

So the drive to kind of integrate into the platforms is really coming from our partners and the brands who want the same level of transparency, the same level of currency type measurement on the LLMs that they get everywhere else. I mean, on Meta, on YouTube, on TikTok, on the open web, on streaming, they get that kind of agnostic verification, and they're demanding it on the LLMs.

因此，推動我們去整合進平台的動力，主要是來自我們的合作夥伴與品牌端；他們希望在 LLM 上也能獲得與其他地方相同水準的透明度、同等的「通用貨幣」式衡量。我的意思是，在 Meta、YouTube、TikTok、開放網路、串流媒體上，他們都能取得那種中立（agnostic）的驗證，而他們也正在要求 LLM 必須提供同樣的能力。

So things are evolving rapidly there. We saw this week that ChatGPT and OpenAI opened their ad platform to numerous third parties to buy -- to allow for buying and creative optimization on those platforms. So it's clear that they're definitely embracing the marketplace, they're embracing third parties to come in and help build that business. And when you throw a number out there like $100 billion by 2030 in ad revenue, they're going to need partners to do that. So we're very positive and bullish on the opportunity. It certainly hasn't materialized yet, and we've been very clear on that. But we do believe that if we can use history as a guide with what's happened with social with us and streaming and mobile, et cetera, that we think this will be a great opportunity for DV down the road.

所以那邊的發展非常迅速。我們本週看到 ChatGPT 與 OpenAI 將其廣告平台開放給多家第三方，讓他們能在這些平台上進行採買——並進行購買與創意優化。因此很明顯，他們確實正在擁抱市場，也在擁抱第三方進來協助建立這門生意。而當你拋出像「到 2030 年廣告營收達 1,000 億美元」這樣的數字時，他們將需要合作夥伴來達成。所以我們對這個機會非常正面、也相當看多。它當然還沒有真正實現，我們也一直很清楚地說明這點。但我們確實相信，如果可以用歷史作為指引——回顧社群、串流、行動等領域在我們身上發生的事——我們認為這將會是 DV 未來很好的機會。

Tim Nollen, SSR.

Tim Nollen，SSR。

Could I switch topics to CTV? Actually, you've had an announcement or two during the quarter. And I'm just curious, what are you bringing to TV measurement, to CTV that is new and different versus what has existed thus far?

我可以把話題轉到 CTV 嗎？其實你們在本季有一兩項公告。我很好奇，你們在電視衡量、在 CTV 上帶來了哪些相較於目前既有做法更新、不同的東西？

And I'm using the term measurement loosely, there's a lot of new ways to measure TV. I'm just curious what is the opportunity for DV in a much more complicated TV market these days than it used to be.

我這裡對「衡量」這個詞用得比較寬鬆，因為現在有很多新的方式可以衡量電視。我只是想了解，在如今比過去更複雜的電視市場中，DV 的機會是什麼。

And just relatedly, you mentioned MTMs for CTV were up 28%, I think, in the quarter. Could you just add a little bit of context for us? I assume that's accelerating on the new products and kind of where is that trending for the rest of the year.

另外相關地，你提到本季 CTV 的 MTMs 成長了 28%，我想是這樣。你能否再補充一些背景脈絡？我猜這是新產品帶動的加速；那麼今年剩餘時間的趨勢大概會如何？

Yeah, Tim. Thanks for the question. We've said that we've got three pillars we're focused on for growth, social, AI, which we spend a lot of time on, but streaming is incredibly important to us as well. Impression growth last quarter was up 28%, and it was driven by higher attach rates for some of our new solutions.

好的，Tim。謝謝你的問題。我們說過，我們聚焦的成長有三大支柱：社群、AI——我們在這上面花了很多時間——但串流對我們也同樣非常重要。上季的曝光（impression）成長為 28%，主要由我們部分新解決方案更高的附掛率（attach rates）所帶動。

So the launch of verified streaming TV, which gives advertisers the ability to actually measure and ensure that their ads are being delivered on a high-quality, full-episode player, not on an outstream or an embedded video someplace, that it is a truly streaming TV environment that is gaining traction and driving attach rates up on the post-bid measurement part of our business.

例如「已驗證的串流電視」（verified streaming TV）的推出，讓廣告主能夠實際衡量並確保他們的廣告是投放在高品質、完整劇集（full-episode）的播放器上，而不是投放在外嵌（outstream）或某處嵌入式影片上；也就是說，這是一個真正的串流電視環境。這項方案正在獲得採用，並推升我們業務中事後競價（post-bid）衡量部分的附掛率。

We've also seen with our tools like the automated Do Not Air lists, which allow advertisers to create dynamic exclusion lists of programming that they don't want to be around. Our attach rate has almost tripled for that across our pre-bid solutions on specific DSPs. So we've seen attach rates grow, which drives pre-bid. We've seen -- and when pre-bid attach grows, post-bid attach measurement grows as well.

我們也看到，像是自動化的「Do Not Air」清單等工具，讓廣告主可以建立動態的排除清單，把他們不希望廣告出現於其周邊的節目內容排除掉。在特定 DSP 上，我們這項工具在事前競價（pre-bid）解決方案的附掛率幾乎成長了三倍。因此我們看到附掛率提升，帶動 pre-bid。而當 pre-bid 的附掛率提升時，post-bid 的附掛衡量也會隨之成長。

And I think it's because that advertisers have been demanding more transparency on CTV, right? Believe it or not, they probably get less granular verification data on CTV than they get on social or even on like short-form video and YouTube. So I think our solutions are starting to touch a nerve with advertisers. It's driving up attach rates. And it's increasing, obviously, the number of impressions that we're measuring across streaming TV as well. This is all good for us. Attach rates mean more money. More people are using our solutions more.

我認為這是因為廣告主一直在要求 CTV 有更高的透明度，對吧？信不信由你，他們在 CTV 上拿到的驗證資料細緻度，可能比在社群、甚至像短影音與 YouTube 上還要少。所以我認為我們的解決方案開始切中廣告主的痛點。這正在推升附掛率。也顯然增加了我們在串流電視上所衡量的曝光數量。這些對我們都是好事。附掛率代表更多收入。更多人更頻繁地使用我們的解決方案。

And, Tim, on your question for the volume of impressions, so, yes, it grew 28% in the first quarter, and we do expect it to continue to outpace the overall revenue growth of the company just because it is the area that is growing, so we do expect that to continue.

Tim，關於你問到的曝光量：是的，第一季成長了 28%，而我們確實預期它會持續跑贏公司整體營收成長，因為這是正在成長的領域，所以我們預期這個趨勢會延續。

Okay, and just a quick follow-up. When you're talking about, I use the term CTV again kind of loosely, but you mentioned streaming, Mark. Are you specifying this from video that you've been measuring in other areas like in social?

好的，另外再快速追問一下。當你在談論時，我再次比較寬鬆地使用 CTV 這個詞，但你提到了串流，Mark。你指的是你們在其他領域（例如社群）也一直在衡量的那種影片嗎？

Yeah, CTV we designate as something that actually ends up on a large player in a living room. Streaming TV includes CTV, but it includes high-quality branded entertainment that may end up, for example, on a mobile device or a tablet, but it's Hulu. It's not a TikTok video. It's Paramount. It's not a reel. That's the difference. So streaming TV, think of it includes all high-quality TV. CTV includes stuff that ends up in your living room on a big screen.

是的，我們把 CTV 定義為實際會出現在客廳大型播放裝置上的內容。串流電視包含 CTV，但也包含高品質的品牌娛樂內容，例如可能會出現在手機或平板上，但它是 Hulu。它不是 TikTok 影片。它是 Paramount。它不是短片（reel）。差別就在這裡。所以談到串流電視，你可以把它理解為涵蓋所有高品質電視內容。而 CTV 則是指最後會在你客廳的大螢幕上播放的內容。

On the screen on the big screen. Okay, great. Thanks for the clarification.

在螢幕上，在大螢幕上。好的，很棒。謝謝你的釐清。

Youssef Squali, Truist.

Youssef Squali，Truist。

Hi, this is Rob on for Youssef. Thanks for taking the questions.

嗨，我是代替 Youssef 的 Rob。謝謝讓我們提問。

On the gross margin expansion due to AI, I'm just curious if we could unpack that. And then is this the new norm for 2026 or are you still targeting the 80%? And then I'm just curious on the drivers behind the sequential trend in large advertising customers and ARPU for that as well. Thanks.

關於 AI 帶動的毛利率擴張，我想了解能否再拆解說明一下。另外，這會是 2026 年的新常態嗎？還是你們仍以 80% 為目標？我也想請教大型廣告客戶數與其 ARPU 的連續性趨勢背後的驅動因素。謝謝。

Yeah, so on the gross margin, we achieved 82% in the quarter and the way we're able to do that is that we are using AI tools to essentially allow us to verify and classify content a lot more efficiently. And our expectation is that even though the volume of impressions that we measured will continue to grow, we'll be able to maintain a healthy gross margin, whether it's 80% or 82%, it's going to depend a little bit on the volume that comes from the new batch of verification that we'll have to do, including on the AI platforms. So it's hard to say, but one thing is certain is that we will be able to continue to remain efficient as the volumes grow and so 80% is a safe benchmark, and it's a very healthy benchmark.

好的，關於毛利率，我們本季達到 82%。之所以能做到，是因為我們使用 AI 工具，讓我們能更有效率地驗證與分類內容。我們的預期是，即使我們衡量的曝光量（impressions）將持續成長，我們仍能維持健康的毛利率；至於是 80% 還是 82%，會在某種程度上取決於我們需要進行的新一批驗證所帶來的量，包括在 AI 平台上的驗證。因此很難給出定論，但有一點很確定：隨著量體成長，我們仍能持續保持效率，所以 80% 是一個安全的基準，而且也是非常健康的基準。

And then on the other question, which was trending on large advertisers, what we're seeing, and we call this out in the last call, which is the average dollars per client for the top 100 is continuing to grow year over year. We look at that on an annual basis, but it is certain that our large advertisers are being upsold to the new solutions and are part of the growth rates that we mentioned on social activation. So the ARPU is growing as we're able to offer them new products, especially on social and soon on CTV.

至於另一個問題，也就是大型廣告主的趨勢，我們看到的情況是——我們在上一通電話會議也提到——前 100 大客戶的每客戶平均金額仍在年對年成長。我們以年度為基礎來看這個指標，但可以確定的是，我們正在把大型廣告主向上銷售到新的解決方案，他們也是我們在社群啟用（social activation）所提到成長率的一部分。因此，隨著我們能提供他們新產品，特別是在社群上、以及很快在 CTV 上，ARPU 也在成長。

Justin Patterson, KeyBanc.

Justin Patterson，KeyBanc。

Thanks for taking my question. This is Jacob on for Justin.

謝謝讓我提問。我是代替 Justin 的 Jacob。

I guess kind of hitting on that last point about how you're using AI internally on classification, can you talk about maybe some of the areas that DoubleVerify is using AI internally in terms of ramping engineer productivity, maybe how you're keeping that -- keeping scaling token costs in mind while kind of ramping adoption internally these tools?

我想延續你剛才提到的最後一點，也就是你們如何在內部用 AI 做分類。你能否談談 DoubleVerify 在內部使用 AI 的一些領域，例如提升工程師生產力？以及在內部推動這些工具採用的同時，你們如何把 token 成本擴張的問題納入考量？

Yeah, thanks for the question, Jacob.

好的，謝謝你的問題，Jacob。

We look at, obviously, two large buckets of how we are leveraging AI. First is in kind of internal execution, where we focus on efficiency and effectiveness and better client engagement. And the second major bucket is kind of AI product development, which we've talked about a lot on this call.

我們看待 AI 的運用，顯然可分為兩大類。第一類是內部執行，重點在效率與效能，以及更好的客戶互動。第二大類是 AI 產品開發，我們在這通電話會議中已談了很多。

When it comes to kind of internal AI execution, we are focused really on agentic development and using agents to create code. So far, we've seen 40% faster software development. We're triaging IT tickets at rates we've never seen before. And it's allowed us to, as we've said before, maintain a headcount that will continue to show efficiencies over the coming year.

在內部 AI 執行方面，我們非常聚焦於代理式（agentic）開發，並使用代理來產生程式碼。到目前為止，我們看到軟體開發速度提升了 40%。我們處理 IT 工單的分流速度也達到前所未見的水準。這也讓我們如先前所說，能維持人力規模，並在未來一年持續展現效率提升。

So A, just in general engineering operations, we've been able to balance the cost of tokens with kind of the impact on it. And we look at everything from an ROI perspective when we lean in on using AI.

因此，A，就一般工程營運而言，我們已能在 token 成本與其帶來的影響之間取得平衡。而且當我們加大 AI 使用時，我們會從投資報酬率（ROI）的角度來檢視所有事情。

You also mentioned core classification, and we've been using that to help our core systems kind of do what they do much faster. It's increased our productivity by 4x in classification. We're driving labeling, what we do when we label content by about 2000 times faster. And we've been very clear that we've got a decent number of contractors that have been helping us with the labeling and feeding our models, that number of contractors will be reduced by over 100 by the end of the year.

你也提到了核心分類（core classification），我們一直用它來協助核心系統更快地完成其工作。它讓我們在分類上的生產力提升了 4 倍。在標註方面，也就是我們標註內容所做的工作，速度提升約 2000 倍。我們也很清楚地說過，我們有相當數量的承包商協助標註並餵資料給模型；到今年年底，這些承包商的人數將減少超過 100 人。

So we're seeing efficiencies by using AI on an engineering team across core classification. And eventually, we're going to see this with client interaction as well as we start moving towards more natural language interfaces that enable better client interactions with less client service engagement and overhead. So it's a big impact on our operations. That's why we're seeing not only increasing margins but faster go-to-market with product and more efficient client engagements.

因此，我們在工程團隊、跨核心分類的工作上，透過 AI 看到了效率提升。而且最終，當我們開始朝向更自然語言的介面邁進時，也會在客戶互動上看到同樣的效果；這能以更少的客戶服務投入與管理成本，帶來更好的客戶互動。所以這對我們的營運有很大的影響。這也是為什麼我們不僅看到毛利率提升，也看到產品更快上市（go-to-market）以及更有效率的客戶互動。

I will now turn the call back over to management for closing remarks.

我現在把電話會議交回管理層作結語。

Thank you all for joining us this evening. As we look ahead, we remain confident in the performance of our business, and our priorities are clear: deepen adoption of the core products with core customers, accelerate the growth of our solutions for social, streaming TV and AI, and drive industry-leading margins by leveraging the power of AI. We appreciate your continued support and look forward to connecting with many of you at upcoming conferences.

感謝各位今晚加入我們。展望未來，我們對業務表現仍充滿信心，我們的優先事項也很明確：深化核心客戶對核心產品的採用、加速社群、串流電視與 AI 解決方案的成長，並透過運用 AI 的力量推動業界領先的利潤率。我們感謝各位持續的支持，也期待在即將到來的各場會議上與許多各位交流。

Ladies and gentlemen, that concludes today's conference call. Thank you all for joining. You may now disconnect.

各位女士先生，今天的電話會議到此結束。感謝各位參與。您現在可以掛線。