# AI Max vs a fixed-price slot: which one tells you what happened? > An AI-managed search auction reports what it chooses to; a time-boxed slot reports opens and click-throughs. The comparison, made honestly, both ways. Published 2026-09-05 · 7 min read · ai-max-vs-fixed-price, ad-transparency, ai-max Canonical: https://bidsurvivor.space/blog/ai-max-vs-fixed-price-slot --- There are two ways to buy attention this month, and they differ less in what they cost than in what they tell you afterwards. On one side is Google's AI Max, which since 1 September runs a large share of Search campaigns on the advertiser's behalf and reports what it chooses to. On the other is the fixed-price placement: a newsletter slot, a flat-fee banner, or a time-boxed auction slot where you know the price before you pay and read the result after. The AI Max vs fixed price argument is usually framed as reach against control. We think the sharper question is the one in the title: which one tells you what happened? ## First, what "fixed price" means here We are using the phrase loosely, and it is worth being exact. A fixed-price slot is any placement where the price is known before the money moves and does not change once it has. A $300 newsletter sponsorship is fixed-price. So is a $50 directory listing. So, in the sense that matters, is a slot on an open auction where you can see the standing bid and choose to beat it: the price is settled at the moment you win, and you knew it when you bid. What all of these share is that the buyer, not the platform, sets the number. AI Max, and automated bidding generally, is the opposite: you set a budget and the system decides what each click is worth. That difference in who sets the price turns out to be the same difference as who gets to see the result. ## What an AI-managed auction reports Log into a Search account this week and you get, for each campaign, impressions, clicks, cost, and conversions. Three of those four are counts. The fourth is increasingly a model: conversions are attributed across clicks by rules you did not choose, and the newest ones arrive with modelled fractions rather than whole numbers. Below that, the report thins out. The search terms report — the list of what people actually typed before your ad appeared — has omitted low-volume terms since 2020 on privacy grounds, so a small account sees a sample. Generated assets under AI Max are labelled "best", "good" or "low" rather than given click counts. Auction insights show your impression share against competitors but not what they paid. And every page ends in a recommendation, usually to raise the budget. None of this is dishonest. It is a system reporting on itself, in the units it prefers, at the resolution it prefers. We went through the structural reasons for that in [why ad platforms hide performance data](/blog/why-ad-platforms-hide-performance-data): a platform that is both the auctioneer and the only reporter has no reason to publish the numbers that would let you argue with it. The [1 September migration](/blog/google-ai-max-migration-what-you-lost) did not create that problem. It removed the settings that let you work around it. ## What a fixed-price slot reports The honest answer is: it depends on the seller, because there is no platform between you and them. A newsletter will usually tell you its subscriber count and open rate, in figures it produces itself. If you tag the link, you will see clicks in your own analytics, which is the first number in this whole comparison that belongs to you rather than to the seller. A flat-fee banner on a publisher's site tells you whatever the publisher decides to tell you, which is often nothing. A time-boxed slot on [our board](/) reports two counts per slot, and we should be clear that this is ours, so weigh it accordingly. An **open** is somebody opening your card on the board. A **click-through** is somebody leaving for your site. They are counted separately, not sampled, and both appear on the front page and on your brand's own page, alongside the price the slot cleared at. Bought or synthetic traffic is flagged at source and never counted in either. The rules are in [how BidSurvivor works](/blog/how-bidsurvivor-works). The catch is the one we keep repeating: the board is young, so those counts are small, and you can see exactly how small before spending anything. Small and readable is a different product from large and modelled. Neither is automatically better. But only one of them lets you write down, at the end of the day, a sentence of the form "this placement cost this and sent this many people". ## Running the five-question test on both In [what a transparent ad platform would look like](/blog/ad-platform-transparency) we proposed five questions for any place you buy attention: does it publish prices, does it publish results, does it publish fees, does it show you where the ad ran, and can you see everyone else's numbers. Applying them here: - **Prices.** AI Max: you learn the average CPC after. Fixed slot: you knew it before. - **Results.** AI Max: counts plus modelled conversions, sampled search terms. Fixed slot: whatever the seller counts; on our board, opens and click-throughs per slot. - **Fees.** AI Max: the platform's margin is inside the auction and not disclosed. Fixed slot: the price is the fee. - **Placements.** AI Max on Search is at least always on a results page; with partners and expansion enabled, not always the one you expected. Fixed slot: one page, which you looked at first. - **Everyone else's numbers.** AI Max: impression share only. Fixed slot: rarely; on our board, every brand's opens, click-throughs and clearing prices are public. The fixed-price slot scores higher on all five, and it would be easy to stop there and declare a winner. That would be the wrong lesson. ## Where AI Max wins, and it is not a small win Intent. A person who typed "invoice software for freelancers" into a search box is telling you something, and no fixed slot on any page can match that signal. If your product has a query, Search is where the query is, and an automated bidder with enough conversion volume will find the profitable corners of it faster than you will by hand. Scale. There are more searches in an hour than most fixed-price placements see in a year. If what you need is volume and you can afford to be wrong for a fortnight while the system learns, the automation is the tool. Fixed-price slots offer none of that. They offer a known price, a known page, and a number you can read. That is a lot for a small advertiser and very little for a large one. ## The rule we would use Buy on intent when you have a query and enough budget for the learning period. Buy fixed-price when you need to know what happened, because the budget is small enough that one wrong fortnight is the whole budget. And whichever you buy, keep one number in a spreadsheet of your own: what a visit cost, on that channel, that week. It is the only figure both kinds of channel can be made to produce, and it is the one an automated report will never volunteer. --- **Sources.** Migration and scope: [Search Engine Land](https://searchengineland.com/google-sets-ai-max-migration-timeline-for-search-campaigns-485006). Reporting, match types and asset labels: [Google Ads Help](https://support.google.com/google-ads/). Our own rules and counts: [how BidSurvivor works](/blog/how-bidsurvivor-works).