The biggest AI answer engine is still a search engine.
Search ads now run alongside AI answers. More research happens before the click than it used to, and more queries end without one. The moment someone is ready to act still runs through Google, so that’s where the budget has to be most precise.
Inside the account, most advertisers miss in one of two directions.
Too much manual control
Burning hours on levers the algorithm already owns: match types, manual bids, device and daypart adjustments.
Too little direction
Handing the wheel to AI with no thought about where it’s driving.
We set the target.
The objective that matches the real business goal, with conversion values that know a $50,000 enterprise lead from a free trial.
We feed the model.
Clean conversions, CRM outcomes, and the assets and feeds AI Max and PMax build from.
We draw the boundaries.
SKAGs are dead; structure isn’t. Brand, non-brand, new and returning split out, so budget goes where we decide it should.
We test deliberately.
Holdouts across Search, PMax and Demand Gen, and new betas tested one variable at a time.
Everyone gets the same Smart Bidding. What you feed it is the part a competitor can’t buy.
Tell a PMax campaign to maximize leads, and it will: as many as possible, with no regard for cost per lead or whether any of them are any good.
Google decides the bid. We decide what it’s bidding for.
Where the algorithm stops and we start
We size search on incrementality.
Brand search that would have converted anyway gets funded accordingly. Non-brand earns its share of budget against social and YouTube using the same measuring stick.
