Google Ads keeps changing settings you already fixed, and the auto-apply recommendations feature is usually why.
What Google Ads auto-apply recommendations actually do
Auto-apply recommendations is a Google Ads setting that lets the platform implement its own suggestions without asking you first.
Turn it on and Google can flip ad rotation back to "optimize," widen language or location targeting, add keywords, or change bid strategies on its own schedule.
Google documents the feature here, and by default several of these categories are already switched on for most accounts. Most account managers never went looking for that toggle, which means most accounts are running on autopilot they never chose.
Why it matters for B2B accounts: every one of those changes is optimized for Google's definition of performance, which is usually clicks and conversions in the aggregate. It is not optimized for your definition, which is qualified pipeline from the right buyer.
Ready for paid ads that pay off?
Book your free auditThe pattern we keep seeing
We manage enough Google Ads accounts to see the same failure mode repeat across completely unrelated businesses.
A team locks ad rotation to even, so every headline gets a fair test. Three weeks later it is back on optimize. Nobody touched it. Auto-apply did.
A team narrows location targeting to exclude a country that never converts. A month later spend is flowing back into it, because a recommendation quietly reset "presence" targeting toward "presence or interest."
A team sets a tight keyword list on purpose, because broader match was pulling in irrelevant traffic. Auto-apply adds keywords back based on search-term matches Google thinks are relevant.
None of this shows up as an alert. It shows up three or four weeks later as a lead-quality problem nobody can explain, because the account "hasn't changed."
The fix is not just turning it off
Turning off auto-apply recommendations stops new changes, but it does not undo the ones already sitting in your account. Both steps matter.
1. Go to Recommendations in the account, open the settings gear, and review every auto-apply category individually rather than toggling one master switch.
2. Turn off auto-apply for anything that touches targeting, bid strategy, ad rotation, and keywords. Reporting and housekeeping recommendations are lower risk.
3. Check the change history for the last 90 days and look specifically for changes you did not make. That is your list of settings to manually revert.
4. Re-check targeting, rotation, and match type settings against what you actually intended, not what the account currently shows.
5. Put a recurring 15-minute audit on the calendar. Weekly for active accounts, every two weeks at minimum. Settings drift back on their own timeline, not yours.
A useful companion check while you are in there: flag any campaign with zero impressions and zero spend over the trailing two weeks. Zombie campaigns like this clutter reporting and hide the real performance signal underneath them, and they tend to accumulate in accounts that have gone through several strategy changes without cleanup.
Why this is worse for B2B than for ecommerce
An ecommerce account with a broad catalog and thin margins can sometimes absorb a recommendation-driven expansion. More traffic, even slightly off-target, still converts some of the time.
A B2B account selling into a specific buying committee cannot. If a recommendation widens your location targeting from "presence" to "presence or interest," you start paying for clicks from people who are merely curious about your target market, not operating inside it. If it resets ad rotation, your carefully tested headline gets diluted by weaker variants Google wants a turn at proving out.
The tighter and more deliberate your targeting has to be, the more expensive an unsupervised recommendation gets. That is the reality for most mid-market and enterprise B2B advertisers, and it is exactly the profile of account where auto-apply causes the most damage per dollar.
How to build the audit into a real process, not a one-time cleanup
A single review fixes the account for a few weeks. A recurring process keeps it fixed.
→ Assign one owner for auto-apply and recommendation settings, not a rotating cast of whoever logs in that week.
→ Log every manual override in a shared doc or the account notes, so the next person understands why a setting is locked the way it is.
→ Pair the recommendation audit with your existing negative-keyword or search-term review, since both catch the same underlying problem: spend drifting away from the intent you actually paid to target.
→ Treat any Google-suggested quick win the same way you would treat a stranger's advice about your business. Sometimes it is right. Read it before you take it.
Google's recommendations are not malicious. They are built to move an account's Optimization Score, which correlates loosely with account activity and Google's own preferred settings, not directly with your pipeline. Score and revenue are not the same thing, and an account can hit a high score while quietly bleeding budget on unqualified traffic.
The accounts that hold their lead quality steady over time are rarely the ones running the newest tactic. They are the ones where someone actually checks, on a schedule, that the settings still say what they meant to say.
If you want a second set of eyes on what your account has drifted into, our paid media audit walks through exactly this kind of settings and change-history review alongside targeting, tracking, and landing page fit. You can also see how we structure engagements on our pricing page.
