Google shows a percentage in every account, the optimization score, along with a list of recommendations that raise it. Many advertisers read the number like a school grade. It is not one. And applying every recommendation just to reach 100% does not automatically make an account better.
First, so you can put this article in context: I am a Google Partner. According to the Google Partners FAQ, Partners have a minimum: the manager account registered in the program needs an optimization score of at least 70%, and Google checks the requirements daily. The same FAQ also says that Partners can apply or dismiss recommendations based on their own assessment and their clients' goals, that both raise the score, and that Google does not require 100%. That is exactly how I work: I review every recommendation, apply the useful ones, dismiss the unsuitable ones with a reason, and both count toward the score.
The facts about the score itself come from the Google Ads Help article on optimization score. Which recommendations I apply and which I do not is my practice, and I say so each time.
What optimization score measures and what it does not
According to Google, optimization score is an estimate of how well a Google Ads account is set to perform. It ranges from 0 to 100%. According to Google, 100% means the account can perform at its full potential.
It is calculated in real time, based on:
- the statistics, settings and status of your account and campaigns,
- the expected impact of the available recommendations,
- your recent recommendation history.
The score exists at campaign, account and manager account level, and according to Google only for active campaigns of certain types, including Search, Performance Max, Shopping, Display and Demand Gen. Each recommendation shows how much it would raise the score. Google gives a range from less than 0.1% up to 100%.
What matters is what the score is not. Google states explicitly that optimization score is not used by Quality Score. And it does not measure results: revenue, profit, cost per lead or lead quality do not appear in the definition. In my practice, I have seen accounts with a high score and weak numbers, and the other way round. The score shows how many open suggestions Google sees for an account and how much weight Google gives them. It does not show whether the account makes money.
Apply or dismiss: both move the score
The most important sentence in the Help article is easy to miss: according to Google, both applying and dismissing recommendations change the optimization score. An account can reach 100% by applying or dismissing all recommendations. The Partners FAQ puts it even more directly: the score goes up when you apply and when you dismiss.
That changes the logic. A high score does not mean someone implemented everything Google suggested. It can just as well mean that someone looked at every recommendation and made a decision.
According to Google Ads Help, this is how you dismiss a recommendation:
- Individually: hover over the recommendation and click the X in the top-right corner.
- All on one card: choose Dismiss all from the three-dot menu of the recommendation card.
Dismissed does not mean gone for good. Google notes that a dismissed recommendation may be shown again if it becomes relevant after some time. So the same recommendation can come back, and then you decide again.
From my practice: I keep a record of which recommendation I dismissed and why, in my own log outside the account. If it comes back, I can see immediately whether anything about the reasons has changed. That saves time and makes the decision transparent for clients.
Auto-applied recommendations: review and turn off
Besides applying recommendations by hand, there is auto-apply. According to the Google Ads Help article on auto-applied recommendations:
- It is a setting you turn on yourself. After that, Google applies the selected recommendation types regularly.
- You can turn it on or off at any time from the top bar of the Recommendations section. That is also where you choose the types.
- A queue on the Recommendations page shows which recommendations will be applied on a given day.
- What has already been applied is shown in the History tab as a summary and in the change history in detail.
- According to Google, auto-applying recommendations will not increase your budget.
According to the Help article, the types that can be auto-applied include adding broad match keywords, adding keywords, adjusting CPA and ROAS targets, and setting a target CPA or target ROAS.
These are exactly the types that change reach and targets, the levers I am least willing to hand over. In the accounts I manage, auto-apply is therefore off for these types. If you take over an account or review your own, check the History tab: it shows whether changes happened automatically in recent weeks that nobody knew about.
My review in three questions
Before I apply or dismiss a recommendation, I ask three questions. This is how I work, not a Google requirement.
- Is the data sound? According to the Google Ads Help article on recommendations, recommendations are based on performance history, settings and trends across Google. If conversion tracking counts wrong, the recommendation calculates with wrong numbers too.
- Does it fit the calculation and the goal? Google does not know your margin, your capacity or your sales process. A recommendation can be plausible for the account and still wrong for the business.
- Can I measure the effect and undo it? Adding a missing sitelink is small and measurable. A new bid strategy across several campaigns at once is not.
These recommendations are often worth it
In the Help article on recommendation types, Google lists dozens of variants. In my practice, I usually apply the following after reviewing them, because they fix measurement or serving:
- Fix tracking: I always take recommendations to set up or fix conversion tracking seriously. Without clean measurement, every automated bid strategy works with the wrong signals. I do the implementation myself and not with one click, because I want to know exactly what is being counted.
- Enhanced conversions: The recommendation to set up enhanced conversions is often useful because it can improve measurement. The prerequisite is that consent and data protection are properly sorted out.
- Missing assets: If Google points out missing sitelinks, callouts or images, I add them, but with my own copy. In my experience, complete assets are almost always a gain, but I still do not leave the wording to automation.
- Disapproved ads and assets: Warnings about disapproved ads, disapproved assets or broken landing pages are not an optimisation idea, they are an error. I fix those right away.
These I often dismiss, and why
I am more critical with recommendations that change reach, budget or targets. Dismissing does not mean the recommendation is wrong in principle, but that it does not fit this account right now.
- Broad match without clean data: I dismiss the recommendation to add broad match keywords or turn on AI Max as long as conversion data is wrong or too thin. Broad match keywords need good signals, otherwise, in my experience, they mostly bring wasted spend. What AI Max actually changes and how Google upgraded campaigns in September is covered in my post on the AI Max upgrade.
- Loosening target CPA or target ROAS: Google often recommends raising the target CPA or lowering the target ROAS to get more volume. If the target is derived from the calculation, I usually dismiss it. More conversions at a price that does not pay off are not a gain.
- Automatically created text: I dismiss recommendations to turn on text customization, meaning automatically generated ad text, wherever claims must be exactly right: prices, deadlines, qualifications or legally sensitive wording.
- Adding keywords in bulk: I review keyword suggestions one by one. Many fit, some target a service the business does not even offer. I do not apply them as a package.
- Raising the budget: I check the recommendation to raise the budget against margin and capacity. If the campaign is profitable and the business can handle more enquiries, it is often right. If not, I dismiss it. It is neither good nor bad in principle.
Overview: my tendency by recommendation type
In summary, this is my tendency. It comes from my practice and is not a Google rule. In individual cases, I decide based on the three questions above.
| Recommendation | My tendency | Reason |
|---|---|---|
| Fix tracking | apply | without clean measurement, every optimisation is guesswork |
| Enhanced conversions | apply if consent is sorted | better measurement |
| Missing assets | apply | with my own copy |
| Disapproved ads | fix | an error, not an option |
| Raise budget | review | only if margin and capacity support it |
| Add keywords | review | one by one, not as a package |
| Loosen target CPA | usually dismiss | the target comes from the calculation |
| Automatically created text | usually dismiss | where claims must be exactly right |
| Broad match without data | dismiss | tracking and data first |
Table scrolls sideways
What this means for you as an account owner
If an agency or a freelancer manages your account, how they handle recommendations is a good test question. Do not just ask about the optimization score, ask how it comes about: which recommendations were applied, which were dismissed, and why? Is auto-apply on, and if so, for which types? More questions I would ask before signing a contract are in the 10 questions to ask before hiring a Google Ads agency. What makes management worth the money is summarised in my post on what makes a good Google Ads agency.
If you manage your account yourself, this is how I would do it: go through the recommendations once a week, decide on each one, and turn off auto-apply for types that change reach, budget or targets. A high score is then a by-product of clean work, not a goal in itself. Whether the account is actually set up well is better answered by your own checklist than by the score, for example my 100-point Google Ads audit checklist.
If you want a second pair of eyes on your account, there is the Google Ads audit (from €500). In ongoing Google Ads management (from €1,000 per month), I handle recommendations as described here.
Sources
- Google Ads Help, optimization score: support.google.com/google-ads/answer/9061546
- Google Ads Help, Google Partners program FAQ: support.google.com/google-ads/answer/9762278
- Google Ads Help, how to become a Google Partner: support.google.com/google-ads/answer/9702452
- Google Ads Help, auto-applied recommendations: support.google.com/google-ads/answer/10279006
- Google Ads Help, recommendations: support.google.com/google-ads/answer/3448398
- Google Ads Help, recommendation types: support.google.com/google-ads/answer/3416396
As of: 29 September 2026
