Google had record search traffic last quarter and still lost the ad race. Meta’s advertising revenue jumped 26% in the third quarter of 2026 to $63.06 billion, while Google’s Search unit grew 12% to $63.35 billion — the widest growth gap in Meta’s favour since 2021, and a result that leaves the two businesses separated by less than $300 million. The explanation is not that Google is shrinking. It is that Meta’s AI advertising tools are converting advertisers at a pace Google’s search business cannot currently match, even as its own AI answers keep users on the results page instead of sending them to ads.
The quarter that flipped the race
Start with the headline figures, because they are stark. Meta’s total revenue rose 26% to $64.60 billion in the quarter ended September 30, 2026, and advertising produced $63.06 billion of it, up 26% from $50.08 billion a year earlier. Two forces drove that growth: ad impressions across Meta’s family of apps rose 10%, and the average price per ad climbed 14%. Volume and price rising together is the healthiest form of advertising growth, and Meta managed both at once.
Google’s Search and other advertising revenue grew 12% to $63.35 billion in the same quarter, within Alphabet’s total revenue of $127.6 billion, up 25%. The comparison is not perfectly like-for-like, and honesty requires saying so: Meta’s figure covers advertising across Facebook, Instagram, WhatsApp and Messenger, while Google Search and other is one component of Alphabet’s advertising business. Including YouTube and its advertising network, Google’s total advertising revenue was $82.01 billion, still substantially above Meta’s total.
Even with that caveat, the Search comparison is the one that matters for this story. Search is Google’s core business and its historic moat, and it is the business Meta’s advertising operation has now nearly caught. The gap between $63.35 billion and $63.06 billion is about $290 million, a rounding error at this scale. A year ago, nobody would have described the two as nearly equal.
The trajectory sharpens the point. Google’s Search growth has decelerated across 2026, from 19% in the first quarter to 17% in the second and 12% in the third. Meta’s 26% is therefore not just a strong quarter; it is a strong quarter arriving while its rival’s core engine cools. Alphabet’s cloud business, up 103% to $30.7 billion, is the company’s brightest growth story, but cloud does not change the advertising picture. On the scoreboard that has defined the internet economy for twenty years, Meta just drew level.
How Meta’s AI ad machine works
Meta’s quarter was not an accident of market conditions. It was the payoff from rebuilding the machinery underneath its advertising business around AI, in three layers that now work together. The first layer is retrieval: a system called Andromeda, described by Meta in December 2024 and fully rolled out across most objectives and placements by October 2025, narrows tens of millions of candidate ads to a few thousand for each person. Crucially, Andromeda reads the ad creative itself, using computer vision and language processing to judge intent and buyer journey stage, rather than merely reading the advertiser’s targeting settings.
The second layer is automation. Advantage+, Meta’s suite of AI-powered campaign tools, is now effectively mandatory for most common performance goals, including maximising conversions, app events and link clicks. Advertisers can no longer disable it for those objectives. Interest selections have become suggestions rather than hard boundaries, and demographic inputs guide the algorithm’s starting point without restricting delivery. The advertiser supplies creative and a goal; the system decides the rest.
The third layer is generative. In July 2026 Meta described its Generative Recommender, which uses large language models to reason about what an ad contains and what a person wants, and said it had put its first generative model into the retrieval stage. Together with other model upgrades, Meta said this increased ad clicks on Facebook by 8.3% and conversions by 15.7%. Separately, more than a million advertisers used Meta’s generative AI tools to create over 15 million ads in a single month, and advertisers using AI image generation saw conversion lifts of around 7%.
Scale is what turns these layers from features into a moat. Almost 10 million advertisers used at least one of Meta’s AI tools, and industry analyses count more than 8 million using its AI creative tools specifically, up from 4 million at the end of 2024. When that many advertisers hand their decisions to the same system, the system’s recommendations start to shape the market itself. Meta’s chief financial officer, Susan Li, told investors the AI investments are already delivering substantial incremental advertising revenue through stronger engagement, better recommendations and more efficient monetisation.
The advertiser evidence: conversions up, spend up
The most telling numbers in Meta’s quarter are not the revenue figures but the behaviour behind them. Advertisers following Meta AI’s recommendations saw roughly 5% more conversions while raising their ad spend 9%. That combination is the loop every ad platform dreams of: better performance earns trust, trust unlocks bigger budgets, and bigger budgets feed the system more data to improve further. A 5% conversion lift may sound modest, but across millions of advertisers it compounds into billions of dollars.
The loop also explains the price increase. When advertisers see better returns, they bid more aggressively for the same attention, which is how average price per ad rises 14% even as impressions grow 10%. Meta is not charging more for the same product. It is selling a product that demonstrably works better, and advertisers are paying for the improvement. That is a far more durable form of pricing power than simply raising rates.
Meta’s guidance suggests the company expects the loop to continue. It forecast fourth-quarter revenue between $73 billion and $76 billion, growth of 22% to 27%. There is a blemish on the quarter worth stating plainly: a $9.89 billion charge tied to the August 2026 multi-state attorneys general agreement over teen mental health cut operating income 37% to $12.99 billion, a 20% margin. Excluding that charge, operating income would have risen 11%. The ad machine is healthy; the legal bill is separate, and large.
Google’s answer-engine dilemma
Google’s problem is not demand. Chief executive Sundar Pichai said Search queries are growing at their fastest year-over-year rate in five years, and credited AI directly. AI Mode, the company’s chatbot-style search feature, passed 1 billion monthly users, while AI Overviews, the generated summaries at the top of results, exceeded 2.5 billion monthly users by mid-2026. AI Mode queries run three times longer on average than traditional searches, and one in five AI Mode searches globally involves voice, images or documents. People are asking more, in richer ways, than ever before.
The problem is monetisation. Chief business officer Philipp Schindler told investors that the strong growth in AI Mode queries had an impact because of the lower monetisation seen in AI Mode, and that foreign-exchange movements were the single largest driver of Search deceleration relative to the second quarter. In other words, two forces slowed Search at once: currency, which Google cannot control, and a shift of queries into a format that earns less per query, which is a consequence of Google’s own product decisions.
Google’s answer is to put ads inside the answers. Schindler said the company is scaling high-quality ads within AI Overviews and AI Mode to more users and queries, including taking its monetisation footprint global for the first time, with multi-turn ads in the United States that change as users continue chatting. The company had already expanded AI Mode advertising, and the third quarter shows why the expansion is urgent: the format users love most is, for now, the format that pays least.
That urgency sits inside a much larger Alphabet story. Google Cloud more than doubled, and total revenue grew 25%, but the quarter still produced negative free cash flow of $9.1 billion after $51.4 billion of quarterly capital expenditure. Google’s full-year capital spending guide of $195 billion to $205 billion is the largest investment programme in the company’s history. The search business must fund an AI build-out of that size while its own AI products change how search earns money. That is the dilemma in one sentence.
The zero-click evidence
The claim that AI answers reduce clicks is no longer a hunch. A randomised controlled trial by researchers at the Indian School of Business and Carnegie Mellon, published through Digital Content Next in September 2026, tracked 1,065 desktop Chrome users and found that when an AI Overview appeared, outbound organic clicks fell 39.8% and zero-click searches rose 34.5%. Clicks on sponsored results, notably, stayed flat. The trial is the first causal evidence on the question, and its design — random assignment rather than before-and-after comparison — is why it carries weight.
Other measurements point the same way. Chartbeat data compiled for the Reuters Institute found Google search referrals to publishers fell 33% globally in the year to November 2025, and 38% in the United States. Similarweb clickstream data analysed by SparkToro put zero-click searches at 68% of United States Google searches in the first four months of 2026, up from about 60% in 2024. Seer Interactive found organic and paid click-through rates down 61% and 68% respectively since AI Overviews launched widely, with cited pages averaging just 0.70% organic click-through when an Overview appears.
Google disputes the gloomiest readings. Its head of Search, Liz Reid, has argued that third-party reports use flawed methodologies and isolated examples, and the company says total click volume has stayed relatively stable. It also launched an AI contribution pilot in June 2026 that pays publishers whose content grounds its AI answers, though it has not disclosed the rate or the calculation. The honest summary is that the direction is clear — answers on the page reduce outbound clicks — while the exact size of the loss is still argued over.
There is also a counterweight that matters for advertisers. Agency data from Brainlabs found that while organic sessions fell 10.5% after AI Overviews became common, referral traffic from AI platforms rose 163%, and visitors from ChatGPT, Copilot, Gemini and Perplexity completed key actions at about 1.5 times the organic-search rate. And our own earlier coverage found that AI Overviews and AI Mode click-through patterns are more nuanced than the headlines suggest. The traffic is not vanishing; it is moving, and the winners will be those who follow where it goes.
The chatbot layer: your new first page
The deepest shift in this story is not Meta versus Google at all. It is the arrival of a third layer between brands and buyers: the chatbot answer. A May 2026 survey of 4,000 shoppers across nine European countries found AI search — ChatGPT, Gemini and Claude — named as a journey starting point by 19% to 27% of respondents, ahead of Facebook, TikTok or Instagram individually. Adobe found 47% of Gen Z consumers discovered a new product or brand through ChatGPT. These are self-reported snapshots, not revenue measurements, but they describe a real change in where consideration begins.
The referral data backs the direction. eMarketer’s industry figures show United States ecommerce visits from ChatGPT grew 210.5% year over year in the first quarter of 2026 to 20.5 million visits, with ChatGPT holding 92.2% of AI referral share, and ChatGPT traffic converting 31% higher than non-branded organic traffic. BCG’s analysis of leading European brands found referrals from external agents under 1% of website visits in late 2025, but estimated that with hidden influence included — shoppers who choose inside an assistant and then visit directly — agents’ effect could reach four times the referral figure, and could overtake organic search traffic in the United States by as early as 2028.
The counterweight deserves equal space. Comscore data analysed by eMarketer puts AI engines at just 3.3% of United States online discovery time, and Chartbeat puts AI chatbot referrals at about 0.01% of publisher traffic, a share that was slipping during 2026. The chatbot layer is small today and possibly overhyped in the short term. But its growth rate, its conversion quality, and its position at the start of the journey are why builders ignore it at their peril. Small channels that convert well and grow this fast have a habit of becoming large ones.
This is also where Google’s and Meta’s stories collide with a third force. ChatGPT itself is becoming an advertising surface: OpenAI has widened ChatGPT advertising with multi-image formats, and since March 2026 Shopify merchant catalogues have fed directly into ChatGPT’s shopping results. The ad race of the next five years may not be Meta versus Google at all, but Meta and Google versus the assistants that sit between them and the customer. For now, though, the money still flows through the two incumbents, and Meta is winning the current round.
The spending war behind the ad war
Neither company is standing still on investment. Google spent $51.4 billion on capital expenditure in the third quarter alone and guides $195 billion to $205 billion for 2026. Meta plans $130 billion to $145 billion of capital expenditure in 2026, the top of that range matching the figure in the original brief for this article. Both companies are pouring unprecedented sums into AI infrastructure, data centres and chips, and both expect advertising to pay for it.
The risk is asymmetric, and it cuts against Meta. Advertising still makes up about 98% of Meta’s sales — precisely 97.6% in the third quarter, $63.06 billion of $64.60 billion — so if ad growth ever stalls, Meta feels it first and hardest. Alphabet, by contrast, can fund its AI build from cloud, which more than doubled, and YouTube as well as search. Meta’s concentration is the price of its focus: the same single engine that produces 26% growth also carries nearly all of the company’s risk.
That asymmetry is also Meta’s argument for urgency. Its AI ad tools are not a side project; they are the defence of the only business that matters to the company. Every point of conversion lift, every advertiser converted to Advantage+, every creative generated by AI is simultaneously a revenue gain and a moat repair. Google can afford a slow quarter in Search. Meta cannot afford a slow quarter in ads, and this quarter it did not have one.
Our read: the moat and the drawbridge
Search was Google’s moat for two decades. Chatbots just found the drawbridge. For twenty years, owning the starting point of commercial intent — the search box where people typed what they wanted — was the most defensible position in advertising. Chatbots now intercept a growing share of those questions before Google ever sees them, and advertisers follow attention wherever it goes. Meta’s quarter shows what happens when a company built for feeds and recommendations meets a world where AI does the targeting: it thrives.
Three things could reverse this story, and each is worth watching. First, Google could succeed in monetising AI Mode and AI Overviews globally at rates comparable to traditional search, closing the monetisation gap Schindler acknowledged. Second, Meta’s growth could hit the limits of its own concentration, whether through regulation, advertiser fatigue with automation, or a market downturn that hits its single revenue engine. Third, the chatbot layer could mature into a genuine third ad platform, with ChatGPT’s shopping surfaces taking budgets from both incumbents rather than merely referring traffic to them.
Forecasters are already pricing in a changing order. eMarketer projected Meta would reach 26.8% of global digital ad spend in 2026, overtaking Google for the first time. That is a forecast, not a result, and it measures the whole digital ad market rather than the Search comparison at the heart of this article. But it captures the direction: the company that best turns AI into advertiser results is gaining share, and right now that company is Meta.
What to do about it
For advertisers, the practical takeaway is blunt. Meta’s AI recommendations are no longer an optional experiment. With nearly 10 million advertisers using the tools and followers of the recommendations seeing higher conversions alongside higher spend, ignoring the AI layer means competing against advertisers who are not ignoring it. The 5% conversion edge is the price of standing still, and it compounds every quarter it is left unclaimed.
For builders and brands, the advice is sharper still. Stop optimising only for Google rankings and test how ChatGPT and Claude describe your product, because that answer is becoming your new first page. Open a fresh chat, ask the questions your buyers actually ask, and note which brands appear and what the answer says about yours. Feed the assistants clean product data, clear descriptions and real prices, and track whether you are named for your most important buyer questions. A business that has never checked its chatbot description is flying blind in the channel where its customers increasingly start.
For publishers, the lesson is diversification. Search referrals are a shrinking foundation, not a dependable one, and the randomised trial evidence says AI answers will keep taking a share of outbound clicks. The counterweight is that AI-referred visitors convert at higher rates, so the goal shifts from maximising raw search sessions to winning high-intent visits wherever they now originate — including from inside AI answers themselves. And for anyone tempted to hide from AI answers entirely, note that users can already bypass AI Overviews when they want to; the audience for direct answers is not going away, but neither is the audience that prefers them.
Background
Meta and Google have traded advertising leadership narratives for years, but the growth gap has rarely tilted this far toward Meta since 2021. Meta’s AI advertising tools rolled out across its apps over recent years, from Andromeda’s retrieval engine in December 2024 to its full deployment in October 2025 and the Generative Recommender in July 2026, while Google’s AI Overviews expanded across Search and AI Mode passed a billion monthly users. The third quarter of 2026 is the first in which the two strategies can be cleanly compared on the same scoreboard, and the scoreboard says Meta’s approach is currently converting better.
The quarter’s results also land as both companies pour unprecedented capital into AI infrastructure. The ad race now funds the AI race, and the AI race in turn decides who wins the ad race, a feedback loop that raises the stakes of every quarter. Alphabet’s cloud growth and Meta’s advertising growth are both, in the end, bets that AI will pay for itself. This quarter, Meta’s bet paid faster. For the full Alphabet picture, including the cloud surge and the cash burn behind it, see our earlier Alphabet Q3 coverage.
Frequently asked questions
Did Meta actually overtake Google in ads? Not in total advertising. Meta’s $63.06 billion of ad revenue nearly matched Google Search and other advertising at $63.35 billion, but including YouTube and its network, Google’s total advertising revenue was $82.01 billion. Meta closed the gap on Google’s core search business, not on Alphabet’s entire advertising operation.
Why is Google’s search growth slowing if queries are at a record? Because queries are shifting into AI Mode and AI Overviews, which monetise at lower rates than traditional search, and because foreign-exchange movements dragged growth relative to the second quarter. More questions do not automatically mean more ad revenue when the answers keep users on the results page.
Are AI Overviews really cutting publisher traffic? The best available evidence says yes. A randomised trial of 1,065 users found 39.8% fewer outbound organic clicks when an Overview appeared, and Chartbeat data showed search referrals down by about a third year over year. Google disputes the scale of the loss and points to stable total click volume, but the direction is consistent across methods.
Should a small brand care about chatbot answers yet? Yes, as a growing channel rather than a replacement. AI referrals are still a small share of traffic, but they convert at higher rates and are growing fast. Checking how ChatGPT and Claude describe your product costs nothing, and the description they give is increasingly the first impression a buyer gets.
Sources: Adweek on Meta Q3, Medianama on Alphabet Q3, eMarketer on AI referrals, BCG on AI shopping agents, DCN trial coverage.
0 Comments