Google Ads Optimization: The Weekly Loop, and the Two Scores That Mislead

Google Ads optimization is a weekly loop over four things: what you are matching to, what you are paying, what the ad says, and where it lands. Everything else is either one of those or a report about them.

Before the loop, two numbers inside the Google Ads interface mislead more accounts than they help, and both are worth settling first because a great deal of optimization effort gets spent on them for nothing.

The two scores that are not targets

Quality Score is not an auction input. Google states it plainly: "Quality Score is not an input in the ad auction. It's a diagnostic tool to identify how ads that show for certain keywords affect the user experience." It runs 1 to 10 at keyword level and combines expected clickthrough rate, ad relevance and landing page experience.

Google goes further and says what to do with it: "Quality Score is not a key performance indicator and should not be optimized or aggregated with the rest of your data."

So the account-wide average Quality Score that appears in so many reports is a number Google specifically says not to aggregate. Use the three component columns to find keywords whose ads or landing pages genuinely mismatch the query. Do not chase the number.

Optimization score is an estimate, not a grade. Google describes it as "an estimate of how well your Google Ads account is set to perform", with scores from "0-100%". Recommendations "can help you optimize" and "may increase the overall performance", which is conditional language, not a promise. Google also confirms you can dismiss a recommendation that does not "seem relevant to your ad campaign", though "a dismissed recommendation can reappear after a certain period if it becomes relevant then."

Applying every recommendation to reach 100% is not an optimization strategy. Several recommendations routinely suggest broadening match types or raising budgets, which will reliably increase spend and may or may not increase return. Read each one against your own goal and dismiss the ones that are not it.

Sources: Google Ads Help, About Quality Score and About optimization score, read 11 October 2026.

The weekly loop

Bidding: the minimums are documented, so check them

Automated bidding is sold as hands-off and is not. It has stated data requirements, and under them it cannot work.

For Target ROAS, Google lists at least 15 conversions in the past 30 days for Search and Shopping, at least 50 in the past 35 days for Demand Gen, and at least 30 in the past 30 days for Video Action campaigns. Google defines Target ROAS itself as "the average conversion value (for example, revenue) you'd like to get for each dollar you spend on ads", with the arithmetic given as "$5 USD in sales ÷ $1 USD in ad spend x 100% = 500% target ROAS".

Two warnings from the same page are the ones accounts learn expensively. "Setting a target that's too high may limit the amount of traffic your ads may get", which is how an ambitious target quietly throttles a campaign to almost no impressions. And after a change, Google advises waiting "1-2 conversion cycles" before judging it.

That last point is the one that breaks weekly optimization habits, and it should. If your conversion cycle is three weeks, a weekly bid adjustment is not optimization, it is interference. Review weekly, change on the cycle.

Source: Google Ads Help, About target ROAS bidding, read 11 October 2026.

Attribution: check what your account is actually using

Google Ads now supports two attribution models, not the six that most guides still describe. Its documentation states: "The first click, linear, time decay, and position-based attribution models are no longer supported by Google", and conversion actions that used them "have been upgraded to use data-driven attribution."

What remains is data-driven attribution, which "distributes credit for the conversion based on your past data for this conversion action" and is "the default attribution model for most conversion actions", and last click, which "gives all credit for the conversion to the last-clicked ad and corresponding keyword."

This matters for optimization because credit assignment is now a model output rather than a rule you set. When a campaign's reported conversions shift, "the model reallocated credit" sits alongside "performance changed" as a genuine explanation, and the two call for different responses.

Source: Google Ads Help, About attribution models, read 11 October 2026.

Negative keywords: the highest-return hour in most accounts

If you do one thing from this guide, do this one.

Run the search terms report over the longest window available. Sort by cost. Read down the list and ask of each term whether a person searching it could plausibly buy from you. The terms that fail that test have been charging you, in some cases for years.

Three patterns worth looking for specifically:

• Research intent on a commercial keyword. "How to do X yourself" against a keyword for the service of doing X.
• Job seekers. Anything with "jobs", "salary", "career" or "internship" attached to your service terms.
• Competitor and brand confusion, where a similarly named product pulls your ads into an unrelated query.

Add negatives at the right level. Campaign-level for broad exclusions, ad group level for steering traffic between your own groups. Negative lists shared across campaigns are worth the setup time once you have more than a couple of campaigns.

Ad copy and landing pages

Two components of Quality Score live here, and both respond to the same discipline: match the language of the query.

For ad copy, write from the converting search terms rather than from internal vocabulary. If buyers search "emergency plumber near me" and your ad says "comprehensive plumbing solutions", the mismatch is costing relevance and clicks.

For landing pages, the test is whether someone arriving from that specific query sees their query answered without scrolling or navigating. Sending every campaign to the homepage is the most common version of failing this, and it is usually the biggest single conversion-rate lever available.

Neither of these is about hitting a Quality Score number, for the reasons at the top. They are about the click being worth what you paid for it.

What not to do

Do not optimize to Quality Score or optimization score. Google says one is not an auction input and should not be aggregated, and describes the other as an estimate.

Do not change several things at once on a low-volume account. With few conversions you cannot attribute the result to any one change, and you will learn nothing while spending to learn it.

Do not apply recommendations in bulk. Read each against your goal.

Do not judge a change before its conversion cycle has run. Google's own guidance is 1-2 cycles.

Do not trust an uplift figure that has no source. This includes figures in articles about Google Ads optimization. If a number is not traceable to your account or to a document you can open, it is decoration.

Where this fits

Most accounts that feel stuck are not short of tactics; they are spending on search terms nobody audited and judging changes faster than the data can answer. Our performance marketing work starts with the search terms report and the conversion setup, because that is usually where the money is.

Sources

All read on 11 October 2026. Google's documentation is the authority on its own products and these pages change.

• About Quality Score
• About optimization score
• About target ROAS bidding
• About attribution models