Flight log / Transmission 056

Google Ads optimization score: which recommendations to accept

Optimization score is the one grade you can raise without doing your homework. Google's own help page says you reach 100% by applying or dismissing every recommendation, and a Partners-registered agency needs at least 70% to earn Partner status.

Optimization score is the percentage at the top of your Recommendations page, and it is very good at making a healthy account feel unfinished. A 64% looks like a failing grade. It isn't one. It's Google's estimate of how much of its own advice you've taken, weighted by how much Google thinks each piece would help.

This guide covers what the score measures, how the uplift math works, who the number really matters to, which recommendation types a B2B lead gen account should usually accept or reject, how auto-apply works, and a monthly routine that keeps the page useful without letting it drive.

What is optimization score in Google Ads?

Google Ads Help defines optimization score as an estimate of how well your Google Ads account is set to perform. It runs from 0% to 100%, and 100% means, in Google's words, that the account can perform at its full potential. It shows at the campaign, account and manager account levels, and only for active Search, Display, Video Action, App, Performance Max, Demand Gen and Shopping campaigns.

Two things it is not. It is not Quality Score, and Google states plainly that optimization score isn't used in calculating Quality Score. And Google's help pages don't say it feeds Ad Rank or the auction. It's a to-do list with a percentage attached, not a ranking factor.

The score also has a focus. Google infers your business objective from your bid strategy. On a fully automated Smart Bidding strategy the goal is explicit; otherwise Google derives it from the Smart Bidding strategy it would recommend for you. A red minus icon next to the score means it has defaulted to a conversion focus. You can override the inferred objective by choosing max conversions, max conversion value, target CPA, target ROAS or target impression share, which matters if Google has guessed wrong about what you're optimizing for.

How is optimization score calculated?

Google doesn't publish a formula or weights. What its help page does list is the inputs. The score is calculated in real time from:

• account and campaign statistics, settings and status,
• the relevant impact of available recommendations,
• recent recommendation history,
• and broader trends in the ads ecosystem, which is why the score and the list can change when you haven't touched anything.

Each recommendation carries a score uplift between under 0.1% and 100%, Google's estimate of its impact. Apply a recommendation worth 5% and the score rises 5%. The uplifts can add up to more than 100, because applying some recommendations invalidates others. Then comes the line that explains most of the confusion: Google says you can reach 100% by applying or dismissing all recommendations.

Read that again. Dismissing counts. Here's what that looks like in a hypothetical B2B account where nothing about the campaigns changes for 10 weeks except a cleanup of the Recommendations page.

In this illustrative account the score jumps from 66% to 88% in one week because someone dismissed 14 recommendations that didn't fit the strategy. The share of leads that sales accepts holds near 29% the whole time, and cost per lead doesn't move either. Same account, same results, a 22-point better grade. That's the score working as designed: it measures agreement with the recommendation list, not performance.

Who does a high optimization score actually help?

Agencies, mostly. Google's Partners program requires a Partners-registered manager account to hold a minimum optimization score of 70%, alongside a 90-day ad spend of $10,000 across managed accounts and Google Ads certifications for at least 50% of account strategists. That's a fair bar for a badge, and it also means some agencies have a business reason to keep the number high that has nothing to do with your cost per lead.

That doesn't make the score useless. It's a decent inbox. Some of what lands there is genuinely good: a conflicting negative keyword blocking your best term, a conversion tracking upgrade you've been putting off, a bid strategy target that's drifted far from actuals. The skill is sorting, and the score's ranking by uplift won't do the sorting for you.

Here's why. In a hypothetical account, the biggest uplifts tend to sit on the recommendations that expand reach and spend, while the repairs that protect lead quality are worth fractions of a point.

In this illustrative snapshot, adding broad match keywords is worth 9.8 points and removing conflicting negatives is worth 0.4. For a B2B account where one wrong-intent keyword can fill the CRM with students and job seekers, the value runs close to the opposite direction. Treat uplift as Google's opinion of the size of the change, not your expected return.

Which Google Ads recommendations should you accept?

Sort by category, not by uplift. Google's Recommendations page groups them into Bidding & Budgets, Ads & assets, Keywords & Targeting and Repairs. Here's how they tend to land for a B2B lead gen account spending $5K to $100K a month with offline sales stages behind the form.

A useful rule: accept recommendations that fix something broken, review the ones that change how you bid, and test (never bulk-apply) the ones that widen who sees your ads. If a bidding recommendation looks reasonable, our comparison of Target CPA and Maximize Conversions covers when each one fits. If you decline broad match, the work moves to your negatives, so keep your negative keyword lists current.

Dismissing is a valid answer. A dismissed recommendation can come back later if it becomes relevant again, and the optimization score page shows how much of your score came from dismissals. That number is honest bookkeeping: it shows where you disagreed with Google, which is fine.

Should you turn on auto-apply recommendations?

Auto-apply lets Google apply chosen recommendation types to your account on a regular basis. Per Google Ads Help, eligibility is checked daily, the day's queue is visible on the Recommendations page where you can dismiss items, and you pick types either through the "Maintain your ads" and "Grow your business" bundles or one by one. Auto-apply works at the account level only, and raising budgets isn't included. The History tab summarizes what was applied, and Change history lets you inspect or update individual changes.

The auto-applicable list spans all four areas: ads and assets (such as "Improve your Responsive Search Ads" and "Use optimized ad rotation"), bidding (target adjustments and strategy switches), keywords and targeting ("Add broad match keywords," "Add keywords," "Expand your reach with Google search partners," "Use Display expansion," plus the removal recommendations), and measurement ("Upgrade your conversion tracking"). Google also notes that the "Add responsive search ads" recommendation stopped suggesting or applying new ads from January 26, 2026.

Our take: turning on the "Grow your business" bundle in a B2B account is handing the steering wheel to a system that grades success by conversions it can see. Here's how that can play out in a hypothetical account where auto-applied broad match keywords went live in month 4.

In this illustrative example, form leads climb about 50% after the switch, and the optimization score looks great. Sales-qualified leads stay at roughly 24 a month. If the account bids on form fills, Smart Bidding now has more "wins" to chase, so the broad queries get more budget, not less. The fix isn't to ban automation; it's to feed bidding the conversions that matter, which is what enhanced conversions for leads and offline imports are for.

If you want auto-apply at all, keep it to the low-risk repair types: removing redundant, non-serving and conflicting keywords. Review the History tab weekly for the first month.

A monthly routine for the Recommendations page

1. Check the score's focus first. If the icon shows a guessed or default objective, set the objective yourself so recommendations aim at the right goal.
2. Filter to Repairs. Fix anything broken there, then look for conflicting negative keyword and tracking recommendations in the other categories.
3. Open Bidding & Budgets. Compare each target change against what your CRM says the last 60 to 90 days of leads became.
4. Open Keywords & Targeting. Decline network and targeting expansions by default. Note any keyword suggestion that matches a real search term you've seen convert, and add it manually.
5. Open Ads & assets. Use the diagnosis, write your own copy.
6. Dismiss the rest so next month's list starts clean, and check the auto-apply History tab for surprises.

Twenty minutes, once a month. It pairs well with the network and placement checks in our invalid clicks guide, since search partners and Display expansion are where both topics meet.

Where optimization score advice breaks down

• Small accounts get thin forecasts. Recommendations rely on your data and Google's forecast models. In a low-volume account, a forecasted lift is a guess with decimals.
• Lead quality is invisible to it. Unless you import what happened after the form, Google only knows a conversion fired. Every recommendation is aimed at that.
• Some accept calls are genuinely good. Blanket rejection is as lazy as blanket acceptance. Broad match with Smart Bidding and strong negatives does work in some accounts; test it in a contained campaign.
• The score moves on its own. Google says ecosystem trends can change it. A drop with no account changes is not an alarm.

FAQ