Google Ads Recommendations: Don't Accept Them

Short answer before you read further: don’t accept the suggestions in the Recommendations tab. Not the harmless-looking ones, not the ones marked urgent, and definitely not all of them at once. Every single recommendation pursues the same goal, more money flowing through your account, and most of them take away your control over where it goes while they’re at it.

1. Accepting Google Ads recommendations means giving up control

The Recommendations tab looks like a maintenance area. Green checkmarks, a percentage in the top right, a button on every card. But it isn’t a maintenance area, it’s the platform’s sales floor, and that percentage is the sales pitch.

The optimization score doesn’t measure how profitably your account runs. It measures how many of Google’s suggestions you’ve implemented. I’ve seen accounts at 65% making excellent money, and accounts at 100% burning cash. That number belongs to Google, not to you, and it sits right where you see it on every login. Why that’s not a one-off but a pattern is covered in detail in the 5 lies Google Ads tells you.

The real problem isn’t any single recommendation, it’s what accepting them does to your account structurally:

• You lose traceability. An accepted recommendation changes keywords, bids or settings without you having read a preview of the change. Two weeks later you no longer know which of five simultaneous changes moved the numbers.
• You never change one thing at a time. That’s the ground rule account work stands or falls on. The Recommendations tab is built to make you break it.
• You dilute your structure. Recommendations don’t know your campaign logic. They push keywords into ad groups you deliberately kept them out of.
• You interrupt the learning phase. Every change to bids and targets sets automated bid strategies back. Accept recommendations weekly, and your account stays permanently unsettled.

One detail most people don’t know: you can dismiss a recommendation instead of accepting it. For the optimization score, dismissing counts the same way, the number goes up either way. If the percentage bothers you, that’s the calm route. Ignoring it entirely is the even calmer one.

2. The recommendations one by one, and what they do to your account

The cards rotate, the pattern doesn’t. Here are the types that show up in almost every account, and what they actually trigger.

Increase budget. The classic, often paired with an estimated number of extra conversions. That estimate assumes additional clicks convert as well as your existing ones. They almost never do, because you skim the best demand first and then bid into weaker auctions after. Whether your account can handle more budget is decided by your cost per lead against your margin, not a card in the account.

Add keywords or switch to broad match. The most expensive suggestion on the list. It widens the range of search queries you pay for, into territory you never chose. On accounts without high conversion volume, that’s a direct route to clicks on “training”, “salary” and “diy”.

Remove negative keywords. Rarer, but particularly unpleasant, because it undoes your own work. Your negative list is the result of real observations in the search terms report. Google only sees lost reach there.

Switch bid strategy or adjust targets. Usually “maximize conversions” instead of manual bidding, or a higher target CPA. Both can be right on a clean account and a disaster on one with half-finished tracking, because the automation then optimizes toward the wrong signal. Check first whether what you’re measuring is even accurate: the conversion tracking health check walks you through that in fifteen minutes.

Add Performance Max or AI Max. Sounds like a new campaign, but mostly means losing control. With AI Max you no longer decide on search queries, with Performance Max you no longer decide on placement. When PMax pays off anyway, I’ve written up separately at Performance Max: when the campaign is worth it. Straight out of the Recommendations tab, it’s never worth it, because nobody there has looked at your feed or your tracking.

Auto-generate ad assets. Google then writes its own headlines and descriptions from your website. For accounts with a carefully maintained ad structure, that’s a step backward, and legally shaky phrasing tends to originate exactly here. See how an ad actually looks beforehand with the ad preview tool (in German).

Apply recommendations automatically. The one point I don’t debate. Auto-apply lets the platform add keywords, change bids and adjust settings without asking you. Check your account for whether this is switched on, and turn it off. A third-party system with a commission incentive has no write access to your budget.

Which of these automations are currently active in your account and working against you takes 2 minutes to find out: the Google Ads settings check (in German) asks 11 yes/no questions and delivers a fix list, no account access needed.

Going through the list, you’ll notice the same thing I did: not a single suggestion type lowers your costs. None suggests excluding a keyword, pausing a campaign or cutting the budget, even though that’s the right move in every second account. A list that only points in one direction isn’t optimization. It’s a catalog.

3. What to do instead

Ignoring recommendations doesn’t mean never changing anything. It means the change is triggered by your own observation, not by a card. The process I run in every account, unchanged:

1. Start with the search terms report. It shows what you actually paid for. It’s the only list in the account that comes from your own data, not from Google’s target.
2. Then the number that matters. Cost per lead against your margin, not click price and not optimization score. If you don’t have that number in your head, you can’t judge any recommendation, however it’s worded.
3. One change, then quiet. Let it run two to four weeks before judging it. On small budgets lean toward four, because three extra conversions in week two isn’t a trend, it’s noise.
4. Bulk changes with a preview. If you genuinely need to exclude fifty keywords, do it in the Google Ads Editor and read the change preview line by line before posting. That’s exactly the control auto-apply takes away from you.

Sometimes a recommendation is genuinely correct, that happens. Implement it yourself anyway, at the right point and at a time you can measure it. The difference isn’t stubbornness, it’s accountability. A change you made deliberately can be evaluated and reversed. An accepted recommendation is, four weeks later, just a kink in a chart nobody can explain.

The same goes for the friendly phone call reading you the same list out loud. Who’s calling and why the script always lands on the same four points is covered in the call from a Google Ads rep.

The bottom line

• Accepting recommendations costs you control, traceability and usually more budget. The Recommendations tab is a sales floor, not a maintenance area.
• The optimization score rates your compliance, not your profit. Dismissing a recommendation raises the number just as much as accepting it.
• Auto-apply must be switched off in every account. It’s write access to your budget for a system that earns from your spending.
• Make changes based on the search terms report and your margin, one at a time, with a preview, and with enough time to see the effect.

Not sure which automation is currently making decisions in your account? I’ll look at the account, the measurement and the economics independently and tell you what can stay and what should go. Book a free intro call (in German).