I ignore most of what Google tells me to do.
Google gives every account an “optimization score.” Most of mine sit around 70%, and I leave them there on purpose. Chasing it to 100% is one of the easiest ways I know to spend a client’s money on Google’s behalf.
I’ll say it plainly, because it’s the whole reason clients pay an agency: Google is a company with a profit motive. The recommendations it surfaces, the green “apply” buttons, the optimization score, the nudges to raise your budget, are built to grow Google’s revenue per account. Not your cost per lead. I get hired to do meaningfully better than those defaults, which mostly means knowing which ones to ignore.
Let me give you a real one.
A while back, Google told me to raise the budget on a campaign from €50 a day to €84. The recommendation had a confident little number attached to it, the kind that makes raising the budget feel like free money sitting on the table. So I checked the one thing the recommendation never shows you: the campaign was already at 92.7% search impression share, and it was losing just 2.3% of its impressions to budget. That’s not headroom. That’s a ceiling. Raising the budget there would have spent more to buy almost nothing, and the optimization score would have climbed while the account quietly got worse.
That gap, between what Google’s interface suggests and what the numbers actually say, is the whole job.
So what do I act on?
Three things, by default. Blocked keywords. Disapproved ads. Anything billing or policy related. Those are real, and I clear them fast. Everything else, the broad-match expansions, the AI Max suggestions, the “add this campaign,” the auto-applied bid changes, I treat as a question, not an instruction. Sometimes the answer is yes. Often it isn’t. Either way it’s my call, made from the account’s numbers, not Google’s.
The one I’d genuinely push you to check today is Auto-Apply. If it’s on, Google can change your account without asking. I watched it flip a client’s campaign off manual bidding onto a fully automated strategy overnight, no warning, through a setting nobody remembered switching on. (Measurement-related auto-applies, like moving to data-driven attribution, I leave on. Those improve the signal without handing Google the steering wheel. It’s the bidding and budget ones that bite.)
None of this is anti-Google, and I’m not saying its recommendations are always wrong. Plenty are fine. The point is narrower: applying a recommendation is not the same thing as improving the account, and the optimization score is not a goal. It’s a measure of how closely you agree with Google.
I agree about 70% of the time. The other 30% is what you’re paying me for.