# Google AI Max moved your Search campaign. Here is what you lost > From 1 September Google moved Broad Match campaigns onto AI Max. What carried over, what control did not, and how to test the next channel first of all. Published 2026-09-05 · 6 min read · google-ai-max, google-ads, ai-max Canonical: https://bidsurvivor.space/blog/google-ai-max-migration-what-you-lost --- On 1 September Google began moving Search campaigns that used Broad Match or automatically created assets onto Google AI Max, whether or not the advertiser had opted in. Microsoft made AI Max the default for new Search campaigns in the same week. If you run a small account and skimmed past the notice, your campaign is now being run by a system with a different idea of what it is for. This is the plain-terms version: what the migration kept, what it took, and what to do with the budget it is now spending. ## What actually happened on 1 September Google announced the timeline in advance and [Search Engine Land reported it](https://searchengineland.com/google-sets-ai-max-migration-timeline-for-search-campaigns-485006): campaigns already using Broad Match keywords or automatically created assets would be migrated to AI Max automatically from 1 September. Campaigns using neither were left alone, for now. Microsoft, whose Ads platform tends to copy Google's structure a few months behind, went further and made AI Max the default for every new Search campaign, as the [mean.ceo September round-up](https://blog.mean.ceo/google-ads-news-september-2026/) notes. AI Max is not one setting. As Google describes it in its [Google Ads help pages](https://support.google.com/google-ads/), it is a bundle: broader matching of your keywords to queries, text assets the system writes and tests alongside yours, and permission to send a click to a page on your site other than the one you chose. Each part had existed as an opt-in. The migration turns the bundle on. ## What carried over To be fair to the migration, most of the furniture is still there. Your daily budget did not change. Your keyword list is still in the account. Your conversion actions, negative keywords, location and schedule settings, and bidding strategy were not reset. Log in and look at the settings page and it looks like the same campaign. That is the trap. The settings are the same; the meaning of the settings changed. ## What you lost **The ceiling on relevance.** A keyword used to be a limit: the query had to be close enough. Under broader matching the keyword is a hint, and the system decides how far from it a query can drift and still spend your money. You find out from the search terms report, afterwards. **Authorship of your copy.** Generated headlines and descriptions run alongside the ones you wrote. You can review them, but you are reviewing what already ran. For a brand with two products and one price, a headline that promises something you do not sell is not a rounding error. **The landing page.** With URL expansion the system may decide a click is better served by another page on your site. If you built one page for the campaign and measured it, that measurement is now partly of pages you did not build for this. **The explanation.** This is the loss that matters most and is discussed least. When a manual campaign performed, you could say why: this keyword, this ad, this page. When an AI-managed campaign performs, you get a report of what the system did, aggregated, and a recommendation to raise the budget. Neither is an explanation. ## The number every founder has, and the one they lack This week's forums are full of a particular kind of post: a founder, a screenshot, and a number somewhere between a few hundred dollars and a few thousand, spent on ads that "did nothing". Read enough of them and the pattern is not carelessness. It is that the founder had exactly one number they could check, total spend, and it arrived after the fact. The arithmetic for a small Google budget has not been kind for a while. We worked it through in [what $200 buys you on Google Ads](/blog/what-200-dollars-buys-you-on-google-ads): at a 2026 average of roughly **$5.42 a click**, $200 is about 36 clicks and perhaps three conversions. AI Max does not change that arithmetic. It changes who decides which 36 clicks you get. What a burned founder needs is not a better dashboard afterwards. It is a way to check the next channel before paying: a published price, a published result, and a unit small enough that a test is cheap. ## Where the manual budget goes now If you want a Search campaign you fully control, the [next post in this series](/blog/can-you-still-run-a-manual-google-ads-campaign) goes through what is still opt-out-able and what is not. If you would rather move some of the budget, we listed the small-budget options in [Google Ads alternatives for small budgets](/blog/google-ads-alternatives-for-small-budgets): Microsoft Ads (with the same caveat now), Reddit, newsletter sponsorships, directories, and auction slots. One of those auction slots is [ours](/), so weigh it accordingly. BidSurvivor is a single page with two slots a day, twelve hours each. The first brand into an empty slot pays $0, with no account or card. Every visitor starts with $100 of house credit before signing up, so a small test costs nothing out of pocket. Each slot's card opens and click-throughs to your site are counted separately and published on the front page and on your brand's own page. The board is young, so those numbers are small, and you can see exactly how small before you bid. There is no targeting and no purchase intent: you are buying twelve hours of attention from whoever is looking, not a person who searched for what you sell. For most direct-response work it is a worse tool than Search was. What it offers is the thing the migration took: you set the price, you see what happened, and nothing is decided on your behalf. For a fuller comparison of what each kind of channel actually reports, see [AI Max vs a fixed-price slot](/blog/ai-max-vs-fixed-price-slot). ## A thirty-day procedure 1. **Pull the search terms report** for the two weeks before 1 September and the two weeks after. Sort by cost. Any term you would not have bid on by hand goes on the negative list today. 2. **Read the generated assets.** Pause any headline that promises what you do not sell. Do this weekly; new ones appear. 3. **Check where clicks landed.** If the landing page report shows pages you did not choose, decide whether to accept that or turn the expansion off where the account still lets you. 4. **Cap the daily budget** at what you would be content to lose while you learn what the system is doing. 5. **Run one small test somewhere with a published price** — a newsletter slot, a directory listing, a free auction slot — and compare cost per visit on the same week. That number, not the platform's, is the one to keep. The migration is not a scandal. Google said it was coming and it will suit plenty of accounts. But a small advertiser's edge was always that they knew their own product better than an algorithm did, and that edge only works while the campaign still lets you use it. --- **Sources.** Migration timeline: [Search Engine Land](https://searchengineland.com/google-sets-ai-max-migration-timeline-for-search-campaigns-485006). Microsoft's default and the week's other Google Ads changes: [mean.ceo](https://blog.mean.ceo/google-ads-news-september-2026/). AI Max components: [Google Ads Help](https://support.google.com/google-ads/). Click prices: [our own arithmetic](/blog/what-200-dollars-buys-you-on-google-ads).