AI Search Ads in 2026: ChatGPT Ads, Google AI Max, and the New Paid Surfaces Marketers Cannot Ignore

Two Shifts at Once, Pulling in Different Directions

Paid search used to mean one thing: you bid on a keyword, Google showed your ad above the blue links, and a user clicked. That model is still alive, but in 2026 it is no longer the whole story. Two separate shifts arrived within months of each other, and they pull marketers in different directions. The first is paid placement inside AI answers, which OpenAI started with ChatGPT. The second is paid automation inside the existing Google Ads system, which Google delivered with AI Max for Search. One adds a brand new surface. The other rewires a surface you already use.

These are easy to confuse because they both carry the words AI and ads, but they solve different problems. ChatGPT advertising is about being seen at all inside a conversational answer that did not previously contain commercial units. Google AI Max is about whether you let an algorithm decide which queries you match, what your headlines say, and which landing page receives the click. The first is a visibility question. The second is a control question. Understanding which one you are dealing with is the difference between a smart test and a wasted quarter.

There is a thread that ties both together, and it is the point this entire piece keeps returning to. Neither surface lets you buy your way out of weak content. ChatGPT keeps ads separate from the answer it generates, and AI Max routes its paid clicks to whichever of your pages Google judges most relevant. In both cases the model is reading your site, your entities, and your structured data to make decisions. If those signals are thin, paid spend amplifies a weak foundation. If they are strong, paid spend compounds an advantage you already earned. Our complete guide to AI search optimization covers that foundation in depth.

Neither surface lets you buy your way out of weak content.

How ChatGPT Ads Actually Work

OpenAI launched advertising in ChatGPT on February 9, 2026, for US users on the Free and ChatGPT Go tiers. This was the first time a major conversational assistant carried commercial units at scale, and the rollout was deliberately conservative. Ads do not appear for every prompt, they are not woven into the model's reasoning, and they are restricted to the lower-cost tiers rather than the paid Plus and enterprise plans. The signal from that design is clear: OpenAI wanted to add a revenue line without giving users a reason to distrust the answers.

The placement reflects that caution. Ads appear in clearly labeled, subtly tinted boxes at the bottom of responses and do not influence the actual answer. The model produces its response first. The ad sits below it as a distinct unit, visually separated and marked as advertising. This matters for how you should think about the surface. You are not buying your way into the answer. You are buying a slot beneath an answer that was generated independently. If the answer above your ad recommends a competitor by name, your ad is sitting underneath that recommendation. That tension is exactly why organic AI visibility and paid placement are not substitutes.

The pilot did not stay in the US for long. In May 2026, the ChatGPT ads pilot expanded to the United Kingdom, Mexico, Brazil, Japan, and South Korea. That international expansion tells you OpenAI is satisfied enough with early results to widen the experiment across very different ad markets and languages. For marketers in those regions, this is the moment to start tracking how often your category triggers an ad slot and what the answer above it says. The early data on which advertisers get shown is still thin, but the structural lesson is already settled: the answer drives the user, and the ad is a supporting actor.

There is a deeper strategic read here too. By keeping ads at the bottom and separate, OpenAI preserved the one thing that makes ChatGPT useful, which is trust in the answer. That decision protects the organic citation as the most valuable real estate on the page. If you want to be the brand the model names inside its response, you optimize for AI citations the same way you would for any other AI surface, and our AI citation optimization guide walks through the mechanics for ChatGPT, Perplexity, and Google.

ChatGPT Shopping and the New Referral Stream

Alongside ads, ChatGPT rolled out shopping updates that change how commerce shows up inside the assistant. For clear buying-intent queries, ChatGPT now surfaces product modules with improved product results, citations, visual product details, pricing, reviews, and direct links to buy. When a user asks for the best running shoe under a certain price or a comparison between two laptops, they are no longer reading a wall of prose. They are seeing a structured product unit with images, prices, and a path to purchase. This is a meaningful shift in how product discovery happens for the people who have moved their research into ChatGPT.

The part that matters most for marketers is the referral behavior. ChatGPT now links to brand websites more frequently, creating new referral traffic and AI visibility opportunities. For a long time the worry about conversational AI was that it would absorb every click and leave brands invisible. The shopping updates push in the opposite direction by sending qualified, high-intent users to brand sites with a buying motion already in progress. That referral stream is new inventory you did not have a year ago, and it is earned, not bought. It flows to the brands the model decides to cite and link.

So how does ChatGPT decide which brands to surface? Which sources show up is affected by established domains, clear entity definitions, and prompt specificity. Established domains carry trust the model has learned to rely on. Clear entity definitions mean the model can confidently identify what your brand is, what it sells, and how it differs from alternatives. Prompt specificity is the user's contribution: a vague query produces a general answer, while a specific one pulls in the brands that best match the detail. You cannot control the prompt, but you can control how clearly your brand is defined as an entity, which is the work most teams skip. Our guide to getting featured in AI search results goes deeper on this, and it applies directly to ChatGPT shopping. For commerce teams specifically, our AI ecommerce SEO guide maps these signals to product pages.

What Google AI Max Actually Does

Google AI Max for Search is a different animal entirely. It is not a new surface and it is not a new campaign type. Google AI Max for Search is an AI layer over standard Search campaigns. You switch it on inside campaigns you already run, and it changes how those campaigns operate under the hood. Specifically, it adds Search Themes, URL Expansion, and AI-generated ad copy, and it extends Smart Bidding. Each of those features takes a decision that used to be yours and hands part of it to Google's models.

Walk through what each one does. Search Themes let you describe what you want to match in plain language rather than building exhaustive keyword lists, and Google expands from there. URL Expansion lets Google send a query to a more relevant page on your site than the final URL you set, which means the algorithm picks the landing page. AI-generated ad copy means Google writes headlines and descriptions on top of the assets you provide. Extended Smart Bidding pushes the bidding automation further into broad and unmatched query territory. Taken together, AI Max trades keyword-level control for reach and automation. That trade is good for some accounts and bad for others, which is the whole debate.

The product matured fast. AI Max moved out of beta in April 2026, and in May 2026 Google added AI Brief and text disclaimers. Moving out of beta means Google is confident enough to put AI Max in front of every advertiser, not just testers. AI Brief and the new disclaimers are a transparency response, giving advertisers more visibility into what the system generated and making it clearer where AI wrote the copy. That sequence, ship the automation then add the transparency, matters because it tells you the disclosures came after pushback, not before. If you run paid search at any scale, our guide to managing Google Ads with Claude Code shows how we audit these settings programmatically rather than clicking through every campaign.

The Honest Data Behind AI Max

Here is where you need to read the numbers carefully, because the headline figure and the field reality do not agree. Google claims advertisers activating AI Max typically see about 14% more conversions or conversion value at a similar CPA or ROAS, and up to 27% lift for exact-match-heavy campaigns. Those are real, specific claims, and they come from the company that built the product. The 27% figure for exact-match-heavy campaigns is the more interesting one, because it suggests AI Max helps most where advertisers were previously leaving reach on the table by matching too narrowly.

Now the counterweight. Independent testing found that 84% of advertisers report neutral or negative results, so results vary widely. Read those two facts side by side and the picture sharpens. A vendor-reported average lift of 14% can coexist with a majority of advertisers seeing neutral or negative outcomes, because averages are pulled upward by a smaller group of large winners. In plain terms, AI Max can produce a meaningful gain for some accounts and quietly erode performance for many others. The variance is the story, not the average. Anyone who tells you AI Max is a guaranteed 14% improvement is quoting half the evidence.

The practical takeaway is not to avoid AI Max. It is to treat it as a controlled experiment with a clear kill switch. Because URL Expansion and Search Themes loosen your match control, the accounts most at risk are the ones with broad inventory, ambiguous landing pages, or weak conversion tracking. The accounts most likely to win are those with clean conversion signals, tight landing page relevance, and a narrow product focus that the automation cannot misroute. If you do not know which group you are in, that uncertainty is itself the answer: test small, measure hard, and be ready to turn it off. A proper audit of your site and tracking tells you which side of that 84% you are likely to land on before you spend.

Why Earning the Organic Citation Still Decides Both

This is the point most coverage of AI ads misses. Paid surfaces in 2026 are not isolated from your organic presence. They depend on it. Consider ChatGPT first. The model generates an answer, names the brands it trusts, and then shows an ad below. If your brand is the one the model cites inside the answer, you get the credibility of the recommendation and the referral link, for free. The ad slot beneath is a secondary play. The most valuable position is inside the answer, and you earn that with content and entity clarity, not with budget.

Now consider AI Max. URL Expansion means Google sends a paid click to whatever page on your site it judges most relevant to the query. Think about what that requires. Google has to understand your pages well enough to choose the right one. If your site has clear, well-structured, topically distinct pages, the automation routes clicks intelligently and your paid performance improves. If your pages are thin, overlapping, or ambiguous, the automation misroutes clicks and your money funds bad landing experiences. The quality of your organic content directly determines how well your paid automation performs. The two are not separate budgets. They are the same asset viewed twice.

This is why we keep telling clients that the order of operations matters. You do not fix paid by spending more on paid. You fix paid by making your organic foundation legible to the machines that now mediate both channels. The same content depth and entity clarity that wins a ChatGPT citation also gives AI Max a better page to route to. The same structured data that helps Google understand your products also helps ChatGPT surface them in shopping modules. One investment, multiple payoffs. Our AIO optimization service is built around exactly this premise, because the foundation is what makes everything downstream work.

The two are not separate budgets. They are the same asset viewed twice.

The Entity and Structured Data Work That Underpins It

If the foundation decides both paid surfaces, the natural question is what that foundation actually consists of. Two things sit at the center: entity clarity and structured data. Entity clarity means a machine can read your site and answer, without ambiguity, what your brand is, what it sells, who it serves, and how it differs from alternatives. ChatGPT's source selection is affected by clear entity definitions, and Google's routing depends on understanding your pages. Both are entity problems before they are anything else. A brand that is defined consistently across its site, its schema, and the wider web is one the models can confidently surface.

Structured data is how you make that clarity machine-readable. Product schema with accurate pricing, availability, and review markup feeds directly into the kind of product modules ChatGPT shopping now displays. Organization and Article schema help establish your domain as an established, identifiable entity rather than an anonymous page. The reason this work pays off across both ChatGPT and Google is that both are reading the same signals. You are not building two foundations. You are building one and letting two platforms benefit. Start with our schema markup generator to get the technical implementation right, then verify entity coverage across the site.

There is also a competitive dimension that is easy to overlook. If a rival brand is cited in ChatGPT for your core category and you are not, that is not bad luck. It is a signal that their entity is clearer or their domain more established for that topic. The fix is to understand what the model is rewarding in their content and close the gap deliberately. This is where competitor intelligence stops being a vanity exercise and becomes operational. Our competitor intel service maps which brands AI surfaces in your category and why, so you are closing real gaps rather than guessing.

How to Test AI Max Without Burning Budget

Given that 84% of advertisers report neutral or negative results, the worst possible approach is to flip AI Max on across your whole account and hope for the 14% average. Instead, isolate the test. Pick one or two campaigns where you have clean conversion tracking and a clear sense of current performance, and enable AI Max on those alone. Leave the rest of the account untouched as a control. This gives you a real comparison rather than a blended number that hides whether AI Max helped or hurt.

Watch the right metrics, not the vanity ones. AI Max will often increase impressions and clicks because it loosens matching through Search Themes and URL Expansion. More volume is not the same as more value. Hold the test to conversions or conversion value at your target CPA or ROAS, which is the exact frame Google uses in its own claim. If conversions rise at a stable CPA, AI Max is earning its place. If you see more clicks but flat or declining conversions, the automation is buying you traffic you do not want, and the disclaimers Google added in May 2026 will not change that math.

Pay special attention to URL Expansion. This is the feature most likely to quietly degrade performance, because it sends clicks to pages you did not choose. Audit the search terms report and the landing pages AI Max selected. If it is routing high-intent queries to weak or off-topic pages, either improve those pages or restrict the expansion. The point is that AI Max is not a fire-and-forget setting. It is a powerful automation that rewards supervision and punishes neglect. For teams that want to monitor this at scale, our technical SEO automation guide shows how we script these audits so nothing slips through.

A Playbook for Showing Up in ChatGPT

Since ChatGPT ads are still a narrow pilot and the answer drives the user, your near-term priority should be organic visibility inside the assistant rather than the ad slot beneath it. The work breaks into three parts. First, become an established entity for your category. That means consistent naming, a clear definition of what you do across your site and schema, and enough authoritative content that the model has learned to associate your domain with the topic. Established domains are explicitly part of what affects which sources ChatGPT shows.

Second, build content that answers the specific questions buyers actually ask, because prompt specificity shapes which brands get surfaced. A user who asks a precise question gets a precise answer, and precise answers pull in the brands whose content matches that precision. If your content only covers your category at a generic level, you will be surfaced for generic prompts and skipped for the specific, high-intent ones where purchases happen. Depth and specificity are not just SEO virtues anymore, they are the qualifications for being named in a buying-intent answer. Our content strategy approach with Claude Code is built to produce that depth at scale.

Third, make your products legible to the shopping modules. If ChatGPT is now showing visual product details, pricing, reviews, and direct buy links for buying-intent queries, the brands with clean product schema, accurate pricing data, and genuine reviews are the ones eligible to appear. This is the same structured data work that helps Google, which is the recurring theme of this piece. Once the organic foundation is solid, the paid ad slot becomes a tactical add-on for the markets where it exists, rather than a substitute for visibility you never earned. Measuring whether any of this drives revenue is its own discipline, covered in our AI revenue attribution guide.

What Marketers Should Do Now

Pull the threads together and the action plan is concrete. Start by auditing whether the machines can read your brand. Run your site through an AI readiness check, confirm your entity is clearly defined, and fix the structured data gaps that stop ChatGPT and Google from understanding your products. This is the foundation that makes every paid surface perform better, and it is the step most teams skip in their hurry to spend. Without it, you are pouring budget into systems that cannot route it well.

Next, treat the two paid surfaces according to their maturity. ChatGPT ads are an early pilot, live in the US since February 2026 and now in five more countries since May. Your job there is mostly to earn the organic citation and the shopping module, and to monitor whether ad slots appear in your category. Google AI Max is generally available and demands an actual decision. Run a contained test, hold it to conversions rather than clicks, and watch URL Expansion closely. Given that 84% of advertisers see neutral or negative results, your default posture should be skeptical until your own data proves otherwise.

Finally, stop thinking of organic and paid as separate budgets fighting for the same money. In 2026 they are two views of one asset. The content that wins a ChatGPT citation gives AI Max a better page to route to. The schema that powers a shopping module also helps Google understand your products. The entity clarity that gets you named in an answer also makes your automated bidding smarter. Invest in the foundation once, and both the new AI ad surfaces and the old ones reward you for it. If you want a partner to build that foundation and run the tests properly, that is exactly what we do. Start with a free optimization consultation and we will map the gaps before you spend a dollar on either surface.

Frequently Asked Questions