By December 31, 2026, most commercial queries that trigger a Google AI Overview in the US will show a paid unit inside or directly beneath it. Check 50 such queries at the end of the year. If fewer than half show that placement, I was wrong.

I am making the call because I watched paid placements absorb the old results page one layout change at a time. On February 19, 2016, Google removed right-rail text ads on desktop and moved to up to four ads above organic results plus three below, cutting total slots from 11 to 7. The stated reason was layout consistency across devices. The effect was more paid space where people looked first and less organic space they could see without scrolling. Shopping and Local pack ads followed the same pattern: put a paid placement near concentrated intent, then let advertisers compete for it.

Attention now sits inside the AI answer. At Google Marketing Live in May 2025, Google expanded Search and Shopping ads inside AI Overviews to US desktop, with global English expansion planned, and began testing ads in AI Mode. Across 3,119 queries, Seer Interactive found organic CTR on queries with AI Overviews fell from 1.76% to 0.61% between June 2024 and September 2025. That does not tell us exactly what any one citation is worth. It does tell us the old supply of organic clicks is under pressure.

My bet is that ads will follow attention into the answer, while the remaining organic citations become harder to win and more valuable to hold. Here are five dated predictions, including what would prove each one wrong.

Prediction 1: Paid units reach most commercial AI Overviews by the end of 2026

The mechanism is inventory, not conspiracy. Google already has advertisers and campaigns. Ads in AI Mode can appear below or within responses where relevant, and eligible Performance Max, Shopping, and Search broad-match campaigns do not need a separate AI Mode placement purchase. I used to tell clients a new placement meant a new thing to manage. I was wrong about that assumption: automatic eligibility makes expansion much easier than asking every advertiser to build something from scratch.

The evidence today is a rollout, not saturation. The May 2025 AI Overview expansion reached US desktop first, global English expansion was planned, and AI Mode ads were still tests. But the route is familiar. Start with commercial queries, see whether people tolerate the placement, then broaden it. Advertisers do not need to love the format; they need to see enough useful demand to keep bidding.

My call: by December 31, 2026, paid units appear inside or directly beneath most US AI Overviews on commercial queries. To check it, choose 50 commercial queries that trigger an Overview and run them in an incognito US desktop session on December 15, 2026. If fewer than 25 show the paid unit, this prediction fails. If you run eligible campaigns now, check where your ads may appear rather than assuming you opted out of an answer placement.

Prediction 2: Commercial Overviews show fewer distinct organic sources

The mechanism is screen space and reading patience. An Overview with unpaid links, a carousel, and a paid unit becomes a long mobile scroll. Something has to give. My prediction is that Google will protect the legibility of the answer and the paid placement while showing fewer distinct organic sources on commercial queries. I watched a version of that squeeze on the old SERP: moving paid units into the most visible space did not eliminate organic listings, but it changed which ones anyone noticed.

There is already a reason to care about holding a citation. In Seer’s query set, cited brands kept 35% more organic clicks and 91% more paid clicks than non-cited brands, even as overall organic CTR on Overview queries fell. That is not proof Google has begun trimming citations. It is evidence that being named in the answer matters while the wider pool of organic clicks contracts.

The next step is a prediction, not an observation: the count of distinct cited domains on commercial Overviews falls by Q3 2027. Sample 30 comparable commercial Overviews in Q3 2025 and again in Q3 2027. If the median count of distinct organic cited domains is flat or higher, I missed. For now, stop celebrating a growing total citation count if you are absent from the few queries that pay the bills. Track those queries individually.

Prediction 3: ChatGPT adds ads beside answers while Perplexity keeps citations unpaid

The mechanisms differ, which is why this one is worth watching. ChatGPT began an ad pilot on February 9, 2026, with sponsored units below answers. It then launched a self-serve Ads Manager on May 5, 2026, and removed its six-figure minimum. Reported US desktop ad penetration rose from 14% in May to 26% in June. Early pricing sat around $60 CPM with a $200,000 minimum; a later minimum was $50,000 with CPC bidding. By June, reported unit CTR was 0.50%, and 66.3% of ads appeared after the second prompt. That last detail matters: the ad can appear after the user has supplied more context than a short search query ever could.

Perplexity took the opposite route. It tested sponsored follow-up questions beginning in November 2024, then phased them out and confirmed in February 2026 that it was abandoning advertising in favor of subscriptions over trust concerns. In that model, there is no native Perplexity ad platform to buy your way into an answer. You need to earn the citation.

What the deck calls omnichannel AI visibility, I call two different ways to end up caring about the same line of text: the cited source. By Q4 2027, I expect commercial ChatGPT answers to routinely pair a sponsored block with two to four cited sources, while brands test citation-targeted pages against ad creative for the same prompts. This prediction fails if ads still appear on fewer than 15% of sampled ChatGPT buyer prompts in December 2027, or if Perplexity relaunches a scaled native ad platform. Do not copy one platform’s paid plan onto another and call that a strategy.

The mechanism is fewer choices plus context. A classic results page offered a row of links and left the reader to compare them. An AI answer narrows the visible sources and puts each citation beside a claim the reader may want to verify. That makes a cited link a different kind of invitation to click: less browsing, more checking.

Seer’s numbers give me a reason to test this, not a result I can claim in advance. Cited brands held onto more clicks than non-cited brands in its Overview query set while overall organic CTR fell. That comparison does not establish that a citation gets a higher CTR per impression than a classic blue link. It leaves that question open. I am betting the reduced set of visible choices will push the citation ahead on comparable commercial queries.

Say you spend $20,000 a month in a category where you rank fifth organically. That fifth-place listing is already easy to overlook. If the same query becomes an answer with paid units and three cited sources, being fifth in the old list offers little comfort. You need to know whether you are one of the sources a buyer can actually see.

My call: by June 2027, advertisers able to measure both will see higher clicks per citation impression than per classic organic impression on the same commercial query set. A clean like-for-like comparison showing parity or worse for citations in the first half of 2027 proves me wrong. Do not label ordinary Search Console impressions as citation impressions to manufacture the test. If you cannot distinguish the two, measure what you can and leave the prediction unscored.

Prediction 5: Buying back lost answer visibility gets more expensive

The mechanism is auction pressure. ChatGPT’s early ad program shows how quickly access can change: a roughly $200,000 minimum fell to $50,000, CPC bidding arrived, and reported US desktop ad penetration rose from 14% to 26% in a month. A lower entry requirement lets more advertisers compete. It does not, by itself, prove prices will rise. My bet is that competition for the commercial answers that convert will push the effective price of that visibility above comparable classic Search placements.

Google’s route could make that pressure sharper. When eligible campaigns can enter an answer placement without a separate build, the pool of potential bidders grows with little work on the advertiser’s side. Organic citations matter in that environment because they offer a way to remain visible without winning every paid auction. I have seen the practical version of that tradeoff in home services: an account with a strong local listing and review profile had another route to the buyer, while an account with no organic presence had to lean harder on paid placement. I would rather have both routes than explain a rising acquisition cost with a prettier dashboard.

By December 2027, I expect the effective CPC or CPM for in-answer paid units on a tracked set of commercial queries to exceed the price of classic Search placements serving comparable intent. I expect the gap to be widest where organic citations are fewest. Like-for-like auction data showing in-answer placements clearing at or below classic Search prices through 2027 would prove me wrong. Until then, treating organic answer visibility as optional is a bet that you can always afford the auction later.

What to do now, before the auction sets the price

If I am right about even three predictions, the move is not to panic about ads. It is to work on the pages that could earn a citation while there is still a route to visibility that does not require another bid. Start where money changes hands. A vague commercial page is unlikely to help a model answer a specific buyer question, no matter how much traffic the page once collected.

Miniature auction inside a glowing AI answer box, with bidders displacing citation cards

For most businesses, I would start with five to ten URLs: the core service or category page, two or three comparison pages, a pricing or cost page, and a best-or-top-picks page if that fits what you sell. Give each page a direct answer near the top, then put the supporting detail below it. Make the claim clear enough to quote in a sentence. Keep the HTML fetchable and the important answer readable without running your JavaScript. That is the practical structure behind how to get cited in Google AI Overviews. It is also a sensible way to prepare pages for assistants that need to extract and cite a source. Make the answer easy to find before you worry about making the page clever.

Who can safely ignore this

Not everyone needs to move this up the list. If your buyers still call from a yard sign or a referral before they search, or your core terms show no Overview and no assistant traffic after you check them for a month, put this work at the bottom. I mean that. Chasing citations on queries with no answer box is work with no return.

For everyone else, do the small version in the next 90 days. Pick ten money pages. Add a quotable answer block to each, with a number where you have a useful one. Check monthly whether your business is cited on the queries that pay you, and keep that sheet beside your Search term report. If three of these predictions land, those pages give you a route to visibility your competitors cannot buy away just by raising a bid. If I am wrong, you still have clearer commercial pages.

I keep paid and citations in one system at groas for that reason. The auction can reprice on every search while a team checks its account once a week. Earn the citation now, or plan to compete for the space beside it later.