What Optimization Score measures
Google looks at your account, compares it with the settings and features it believes you should be using, and expresses the difference as a score on the Recommendations page. Each open recommendation carries an estimated uplift, and applying or dismissing it moves the score. Dismissing counts the same as applying, which is the first clue about what the number really is.
The recommendations come from account settings, bidding, keywords, ads, budgets and measurement. Some are genuinely useful: fixing a broken conversion action, adding assets to an ad, or removing conflicting negatives. Others are growth suggestions dressed as fixes, such as raising budgets, widening match types, or switching a campaign to a different type. The score does not distinguish between the two, because it measures adoption rather than outcome.
It is worth being clear on what it is not. It does not look at your profit, your margins, whether the leads you receive are any good, or whether the conversions being counted are real. It is a checklist score computed from your account alone.
Why Optimization Score matters
As a prompt, it earns its place. Recommendations surface things that are easy to miss in a busy account — a disapproved ad, a conversion action that stopped recording, an ad group with no active keywords, an asset missing from a campaign. Working through the list is a fast way to find real faults.
It matters for a second reason that has nothing to do with performance: agencies and Google representatives are often measured on it, so it appears in conversations and reports as though it were a performance metric. Knowing what it actually measures lets you push back on advice that exists to move the score rather than the business.
Where Optimization Score goes wrong
The damaging pattern is auto-apply. Left switched on, recommendations change match types, budgets and campaign settings without a person deciding whether that suits the business. An account that was deliberately tight in order to control cost per lead can be widened overnight, and the score improves while results decline.
The second mistake is treating a high score as proof of good management. An account can score well while advertising the wrong products, sending traffic to a weak landing page, or counting page views as conversions. The third is dismissing everything for a quick score, which loses the genuinely useful prompts along with the rest.
How to act on it
Read every recommendation, apply the ones that fix something broken, and dismiss the ones that do not fit with a note of why. Turn auto-apply off unless you have chosen a specific recommendation type you are happy to be applied without review. Treat any suggestion to raise budgets or loosen matching as a proposal to be tested, not an instruction.
Judge the account on cost per lead, lead quality and revenue instead, and use impression share to see whether growth headroom actually exists before you buy more of it. If the score is being quoted at you as evidence, a proper Google Ads audit answers the question the score cannot: whether the money is producing customers.
