How recommendations and auto-apply work
The recommendations tab is Google’s list of changes it thinks would improve the account: add these keywords, switch this campaign to a different bid strategy, raise this budget, add these assets, remove keywords it considers redundant. Each item carries an estimated effect and contributes to Optimization Score, a figure Google calculates from how many of its own suggestions you have accepted or dismissed.
Auto-apply is a separate set of switches. Turn one on and Google stops suggesting that type of change and starts making it, without asking, whenever its models decide the moment is right. Some auto-apply settings are cosmetic, such as adding automated assets. Others are structural — bid strategy changes, keyword additions, match type expansions — and those alter where the money goes.
Why they matter
The list is genuinely useful as a to-do list you did not have to write. It catches disapproved ads, missing conversion tracking, campaigns limited by budget, and assets you never filled in. Reading it weekly is a good habit even if you accept almost nothing.
What it is not is an objective ranking of what your account needs. Optimization Score measures agreement with Google’s defaults, not profitability, and dismissing a recommendation with a reason is a legitimate answer that lowers the score. A carefully built account with tight match types and a deliberate structure will always sit below a loosely built one that accepted everything. Judge the account on cost per acquisition, not on that figure — the same discipline covered in a proper Google Ads audit.
Where auto-apply goes wrong
The damage is rarely dramatic; it is drift. Keyword recommendations add terms that were deliberately excluded, match type changes widen targeting that was narrow on purpose, and a bid strategy swap resets learning at a moment nobody chose. Because the change was not made by a person, nobody remembers it happening, and the account slowly becomes something other than what was designed.
Two things make this harder to catch. Auto-applied changes appear in change history under Google’s own name rather than a colleague’s, so they are easy to skim past. And the effect shows up as a gradual shift in traffic quality rather than a broken campaign, which means the first sign is usually a rise in cost per lead that gets blamed on competition instead.
What to do about it
Open the auto-apply settings and look at what is already enabled — accounts often arrive with switches turned on by someone earlier, or by a setup wizard. Leave on the ones that cannot change targeting or spend allocation, such as fixing a broken final URL. Switch off anything that adds keywords, changes match types, changes bid strategy or raises budgets, and make those decisions yourself.
Then treat the recommendations tab as a queue, not a score. Work through it once a week, apply what genuinely matches the plan, and dismiss the rest with the reason attached so Google stops repeating it. Check change history at the same time and filter for automated changes, so nothing enters the account without someone seeing it.
