On September 1 Google Hands Search to AI Max

On September 1 the buying stopped being work you do by hand

Google began automatically migrating legacy Search campaigns into AI Max at the start of September, rolling the change out gradually through the month. Microsoft made its own version generally available and turned it on by default for new Search campaigns, with somewhat more visible reporting. Meta moved in the same direction from a different angle, replacing manual placement exclusions with value rules that cap a reduction rather than remove a placement outright.

Three platforms, one direction. The manual controls that account managers have optimized against for fifteen years are being folded into systems that decide more and disclose less.

It is worth being precise about what changed and what did not. The auction still exists. Budgets still cap spend. Conversion tracking still decides what the system optimizes toward. What moved is the layer between intent and inventory, the part where a person used to choose which query patterns and which placements deserved money.

The controls did not disappear they moved

It is tempting to read this as loss. It is more accurate to read it as relocation. Bidding, matching and placement were never the source of advantage. They were the source of effort. A skilled buyer could extract a few points of efficiency from a well structured account, and those points were worth real money, but they were always a margin on top of something else.

The something else is the offer. What you sell, who you sell it to, what it costs, what it promises and what happens after someone says yes. No automated buying system touches any of that. AI Max can find the person. It cannot make the thing worth buying.

This publication has made the argument before in a different form. It is not an ad problem, it is an offer problem. Automation makes that argument literal. When the platform takes over the part you used to control, everything you still control becomes the whole game.

The uncomfortable part for agencies is that a lot of billable work lived in that layer. Weekly optimization, negative keyword hygiene, bid adjustments by device and hour. Some of that work was genuinely valuable and some of it was theater, and the platforms have now made the distinction public. The teams who were selling the theater will feel this quarter. The teams who were selling judgment about what to sell and how to sell it will not.

The offer is the last thing you still own

Three things sit outside the machine, and all three carry more weight now than they did in August.

The first is the offer itself. Price, packaging, guarantee, terms. An automated system optimizing toward conversions will find more conversions faster when the thing converts. Improving the offer is the highest leverage work available, and it is the work most teams have deferred because account optimization felt more tractable.

The second is the creative. Automated systems assemble and test at a scale no human can match, but they assemble from what you give them. A thin asset library caps the ceiling no matter how capable the model is. The teams who win the next two quarters are the ones feeding the system more raw material, not the ones auditing its choices.

The third is the destination. Landing experience, speed, clarity, proof. The machine delivers a visitor and then stops. Everything after the click is still yours, and it is where most of the loss happens.

There is also an opportunity here that gets missed in the anxiety. Automated systems reward businesses whose economics are clear. If your margin per customer is well understood, if your conversion event is defined correctly, if your best customers look different from your average ones and you can tell the system so, the machine will find more of them faster than any manual account ever did. Clarity is now a competitive input.

What to change this quarter

Start by auditing the offer rather than the account. Write down what you are selling, at what price, with what guarantee, against what alternative. If that page is thin, no bidding strategy will save it.

Then expand the creative supply. Assume the system will test more variations than you would have, and give it enough distinct angles to test something meaningful. Volume of genuine variation beats polish on a single execution.

Then instrument the part the platform will not report on. Reporting transparency is uneven across these systems and it is going to stay uneven. Your own measurement of what happened after the click is the only complete picture available to you, which makes first party tracking less of a compliance chore and more of a competitive requirement.

Finally, resist the instinct to fight the migration. The direction of travel is consistent across every major platform, and teams spending this quarter trying to preserve manual control are spending it on a position that will not hold. Teams spending it on the offer are building something the platforms cannot take back.

Automation is not the end of the advertising job. It is the end of one part of it, and that part was never where the money was.

← All stories