Google Ads recommendations: apply them on your terms

Every day, Google Ads shows you recommendations and an optimization score inside your account. Some of those recommendations are valuable. Others change how your account behaves without any regard for your goals, and the impact on performance isn't always what you're told to expect. Here's how to sort the good from the costly, and why you should test before you apply.

What Google actually recommends

Google's developer documentation lists the types of recommendations: switching to target CPA, creating Performance Max campaigns, using broad match with automated bidding, fixing campaigns limited by budget. Each one shows its effect on the "optimization score," which Google describes as an estimate of how well your account is set up to perform.

On average, a low optimization score comes with weaker results

According to a study of 17,380 accounts by Optmyzr, a company that makes optimization software, accounts scoring between 90 and 100 show a ROAS 186% higher than accounts under 70, along with lower CPAs. Ginny Marvin, Ads Liaison at Google, confirms it: on average, higher scores correlate with better results. A low score is a real warning sign.

But it isn't a law, and it doesn't cause your results. In that same study, accounts under 70 had a better conversion rate, and ROAS held steady between 70 and 90.

When a recommendation ends up costing you

"Switch to target CPA." I've seen accounts lose a big share of their conversions after applying this recommendation, while a "Maximize clicks" strategy with a capped max CPC brought in more. Google admits as much in its documentation: a target set too low can make you miss clicks that would have converted, and so reduce your total conversions. It also advises evaluating performance over the last 30 days, with at least 30 conversions. Below that, the measurement isn't reliable.

"Add broad match keywords." Unless you're deliberately trying to reach terms that are only loosely related to your business, it's rarely a good idea. Your ads can show up on searches that stray far from your topic. The result: your daily budget runs out fast, Google's answer is to recommend raising it, and those clicks barely convert. Stick with exact match and phrase match. To grow your keyword list, use Keyword Planner instead.

"Create a Performance Max campaign." This single campaign serves your ads across all of Google's channels: Search, YouTube, Gmail, Discover, Maps and Display. I've seen campaigns where most of the budget went to Display, with delivery quality far below what a Search campaign gives you. On some accounts the split looks fine at launch, then a few months later as much as 99% of the budget ends up on Display. The one rule: check where your ads are actually running and act on what you find. In beta, Google is gradually rolling out settings that give you better control over the share of budget going to Display and partners.

Why being cautious is the rational choice

Google makes money when your budget gets spent, which doesn't necessarily serve your interests. Only you know your margins, the quality of your leads and the length of your sales cycle.

Consultant Georgi Zayakov, quoted by Optmyzr, says he rejects most recommendations and ignores the score, except when his Google Partner badge is at stake. Optmyzr points out that a minimum score was required for that status in 2024. Ask your agency: is the score serving your results, or its badge? Another signal: only 6% of the 500 professionals surveyed by StackAdapt, which builds a competing platform, almost always apply the platforms' AI recommendations.

How to stay in control

• Never apply "everything," and make sure auto-apply is switched off for the more sensitive types of recommendation.
• Before applying a recommendation that touches bidding, budget, structure or campaign type, run an A/B test: it's always better than applying it straight away. In Google Ads, experiments split traffic between your current setup and the version being tested. Set the duration and the success criterion before you launch.
• Give it 30 days and at least 30 conversions before you judge, and look at quality (qualified leads, sales), not just the CPA on screen. If your volume is too low to draw a conclusion, don't change the original campaign directly: test on a secondary campaign first.
• Apply the hygiene recommendations: conversion tracking, conflicting negative keywords, missing assets. There's no need to test those.
• Use the score to prioritize, not as a target: chasing 100% can push you into decisions that work against your interests.
• In Performance Max, take the time to analyze the split by channel and by placement. Google is gradually rolling out a beta percentage cap for Display and partners.

AI and automation absolutely belong in your Google Ads account, provided you stay vigilant. A/B test every change to your bidding strategy, budget or structure before rolling it out more widely, and always stay in charge of your objectives.