Google Ads Optimization Score: Apply or Ignore? (2026)

Google reports that advertisers who raise their Optimization Score by 10 points see, on average, more conversions — and in 2026 that single statistic is the reason millions of accounts chase a number that has little to do with profit. The Optimization Score is a percentage Google assigns to how closely you follow its own suggestions, not a measure of whether your campaigns make money. Understanding that gap is the difference between using Recommendations as a sharp audit tool and being quietly nudged into spending more.

This guide breaks down what the score actually measures, which Recommendations protect performance, which ones only grow spend, and why auto-apply deserves suspicion. To see which suggestions matter for your account against real performance data, run our free 5-axis Google Ads audit.

Updated 2026-05-15 with current Recommendations categories, auto-apply behavior, and Optimization Score weighting observed across US, UK and European accounts.

What does the Optimization Score actually measure?

The Optimization Score is a number from 0% to 100% that Google calculates for each campaign and rolls up to the account level. It estimates how closely your setup follows the Recommendations Google would make — and that is all it estimates. It is not a measure of conversions, revenue, or efficiency; it is a measure of agreement with Google's playbook.

The scoring mechanism — Each pending Recommendation carries a weight, shown as a percentage uplift. Apply it and your score rises by that weight; dismiss it and the weight disappears from the calculation, so your score rises too. This means you can reach a high score either by applying suggestions or by dismissing the ones you reject — a detail most advertisers never notice.

What it does not measure — The score knows nothing about your margin, your break-even, or whether a suggested budget increase would be profitable. It cannot see your CRM, your true cost of goods, or your offline conversions. A campaign losing money can sit at a high score, and a tightly profitable one can sit lower simply because the owner declined volume-chasing nudges.

Why it exists — Google built the score to encourage adoption of features that, on average and across all advertisers, increase activity on the platform. That averaging is the catch: what lifts results across millions of accounts is not necessarily what lifts yours. For the metric that does drive the auction, see our Quality Score guide, which is a different system entirely.

Why is chasing 100% the wrong goal?

Treating 100% as the target inverts the relationship between the tool and the goal. The score is a proxy Google designed; your goal is profit. When you optimize the proxy directly, you apply suggestions because they raise the number, not because they raise your return — and the two diverge fastest exactly on the Recommendations that grow spend.

The 100% trap — To reach a perfect score you must apply, or deliberately dismiss, every pending Recommendation. The fastest path is usually to accept the big-weight ones, which are often budget increases, broad match, and asset automation. An advertiser who clicks toward 100% frequently ends up with more spend, looser targeting, and a score that looks great in a report while cost per conversion drifts up.

Average uplift, not your uplift — The conversion-lift figures Google attaches to Recommendations are modeled averages across many accounts. Your account has its own margins, seasonality, and competitive set. A suggestion that adds conversions on average can still add unprofitable ones for you, especially when it widens reach into lower-intent inventory.

What to optimize instead — Anchor every decision to cost per conversion and return on ad spend, the same way you would judge any spend. A score of 78% with a healthy ROAS beats a score of 100% bought with inflated budgets. To set the targets that should drive these calls, see our guide to lowering your CPA.

Which Recommendations are usually worth applying?

Not every Recommendation serves Google more than you. A specific set fixes real problems, costs nothing, and is reversible — these you should clear quickly because leaving them in place actively harms performance.

Conversion tracking fixes — When Google flags broken, missing, or duplicated conversion tracking, treat it as the highest priority. Smart Bidding is only as good as the signal you feed it, and a broken tag starves the algorithm and hides real sales. Our Google Ads audit checklist covers the full measurement review these flags point to.

Conflicting negative keywords — A Recommendation that surfaces a negative keyword blocking your own active keywords is pure value. These conflicts silently suppress eligible traffic, and resolving them restores impressions you already paid to compete for. Apply them after a quick check that the unblocked term is genuinely one you want.

Disapprovals and policy fixes — Recommendations that flag disapproved ads, missing assets, or policy issues protect serving. A disapproved ad earns nothing, so clearing these restores eligible volume with no downside. The same goes for clearly relevant negative keywords Google surfaces from your search terms — for the structural version of that work, see our account-level negatives guide.

Which Recommendations should you routinely dismiss?

The other side of the list is where the commercial nudge lives. These Recommendations are not scams — they help some accounts — but they grow spend faster than profit by default, so they deserve a deliberate test, never a reflex apply.

Inflated budget increases — A budget Recommendation assumes more volume is always good. Below your break-even, it simply buys more unprofitable clicks. Apply one only when your cost per conversion has headroom against margin; otherwise dismiss it without worry, knowing dismissal never penalizes the account.

Aggressive broad match — Switching keywords to broad match, often paired with Smart Bidding, can find new converting queries — but without strong negatives and trustworthy conversion signal, it pours budget into low-intent traffic. Treat broad match as a controlled experiment in one campaign, not an account-wide flip.

Automation and expansion — Automatically-created assets, final-URL expansion, Display expansion, and Search Partner reach all hand control to the algorithm and widen inventory into lower-quality placements. Each can work, but each blurs your control and attribution. Dismiss them unless you have a specific, measured test in mind, and revisit only with a clear hypothesis.

Why is auto-apply risky, and how do you turn it off?

Auto-apply is the feature that turns the Optimization Score from a suggestion engine into an autopilot you did not ask for. It lets Google implement Recommendations automatically, without your review — and that is exactly where the risk concentrates.

What you lose — With auto-apply on, Google can raise budgets, change bidding targets, switch match types, and add assets while you are not looking. Structural edits ship without sign-off, you lose the audit trail of who changed what, and a bidding change can reset Smart Bidding learning at the worst possible time, costing days of stable performance.

The governance problem — For agencies and teams, auto-apply breaks accountability. When performance shifts and nobody on the team made the change, diagnosis becomes guesswork. Every meaningful edit should be attributable to a person and a reason, judged against margin before it ships, not discovered after the fact in the change history.

How to turn it off — Open the Recommendations page, go to the auto-apply settings, and uncheck the categories — especially budgets, bids, and match types. If you keep any enabled, limit it to genuinely safe hygiene like removing redundant keywords, and never budget or bidding automation. For a structured way to review what changed, lean on our audit checklist on a fixed cadence.

How do you use Recommendations as an audit signal instead?

The most valuable way to use the Recommendations page is to flip its purpose: stop treating it as a to-do list and start reading it as a free, always-on audit. Google is surfacing things it noticed about your account, and even the self-serving cards often point at a real underlying signal worth investigating.

Read every card for the signal underneath — A budget Recommendation tells you a campaign is limited by budget — useful to know even if you choose not to raise it. A broad-match suggestion hints Google sees unserved query volume. A tracking flag is a genuine measurement alert. The card is the messenger; your job is to read the message, not obey the instruction.

Run a monthly pass — Schedule a recurring review of the list to catch new tracking breaks, fresh keyword conflicts, and disapprovals early, before they cost a week of wasted spend. This converts a sales surface into a diagnostic one, and it costs nothing beyond fifteen minutes of attention.

Decide on merit, then move on — For each card, apply, dismiss, or note-and-test, judging against your own cost per conversion and return on ad spend. Dismissing clears noise so the genuinely useful flags stay visible. To pair this read with the structural checks that catch what Google won't surface, work through our audit checklist alongside it.

The apply-or-dismiss decision table

Work this table top to bottom — it is ordered from the Recommendations that almost always help to the ones that almost always need scrutiny, so the safe applies sit at the top and the commercial nudges sit at the bottom.

Putting it together: score as a tool, not a target

The Optimization Score is most useful when you demote it from a goal to a signal. Read every Recommendation, apply the handful that fix measurement and hygiene, scrutinize anything that touches budgets or bids, and dismiss the spend-growth nudges that do not fit your margin. None of those dismissals hurt your account; they only lower a cosmetic percentage.

Keep control — Turn off auto-apply so no structural edit ships without your sign-off, and review the list yourself on a monthly cadence. A 78% score with a profitable ROAS is a far better outcome than a 100% score bought with inflated budgets and broad match you never wanted.

Anchor to profit — Every apply-or-dismiss call should answer one question: does this lower my cost per conversion or raise my return on ad spend? If a Recommendation cannot, dismiss it and move on. To find the wasted spend these nudges often hide, run the SteerAds free 5-axis audit, and quantify the leak with our wasted ad spend calculator.

Sources

Official sources consulted for this guide:

• support.google.com — about Optimization Score

• support.google.com — about Recommendations

• blog.google — ads and commerce updates

• ads.google.com — Google Ads

FAQ