What Moves a Stuck Quality Score (and When to Stop Trying)

The 1–10 number in your keyword table is a reading, not a control. This piece covers what it measures, what changes it, what doesn't, and how to decide what to do with keywords that won't budge. Platform details reflect Google Ads Help as of October 2026.

A keyword column full of 3s and 4s invites a particular question. The ads look reasonable and the landing pages exist, so is there some other lever? A bid, a setting or a few more months of account history that would nudge those 4s up to a 5?

Not a hidden one. The only levers are the things the score measures: how well the keyword, the ad and the landing page fit the search, judged against whoever else shows for it. That can sound like advice you've already heard. But the reason behind it changes what you should do with the keywords that stay low, and what you tell the people who ask about them.

What the number is

Google calculates Quality Score for each keyword from three components:

• Expected clickthrough rate: how likely your ad is to be clicked when it shows.
• Ad relevance: how closely the ad matches the intent behind the search.
• Landing page experience: how relevant and useful the page is to people who click.

Each component gets a grade of Above average, Average or Below average. The grade comes from comparing you with other advertisers whose ads showed for the exact same search over the last 90 days. The 1–10 score summarizes those three grades.

Two things follow from that definition. First, the score is relative. A 4 doesn't mean you missed a fixed bar. It means your ad and page compare poorly with whoever else showed for that search, and stronger competitors can move your grade without you touching anything. Second, the score is built only from searches that exactly match the keyword. Searches your keyword picks up through broader matching don't enter it.

The score you see is not what the auction uses

Google's About Quality Score page is blunt. The score "is not an input in the ad auction," and it "should not be optimized or aggregated with the rest of your data." What the auction uses instead, according to Google's guide to using Quality Score, is real-time evaluations of the same three components, made for each search.

If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.

Google's own guidance compares the score to a warning light on a car's engine. That's the right picture. You can't make the engine run better by working on the light. And because the auction weighs the context of each search, which the reported score may not capture, a keyword can do better or worse in practice than its single number suggests.

One keyword status seems to contradict all this. "Rarely shown due to low Quality Score," which Google glosses as "Keyword isn't showing ads often due to a low Quality Score." Google's documentation supplies the pieces to square it:

• ads must meet a certain level of quality to show at all;
• lower-quality ads face higher Ad Rank thresholds;
• Google says its assessment of your ad quality "is summarized in" your Quality Score.

The most consistent reading is an inference, not something Google states outright. The status names the summary, but what keeps the ad off the page is the underlying quality failing to clear the auction's bar. Either way, the remedy runs through the three components, not through the number.

What you see in the account, and what it actually means

A costly move around Quality Score can start with something real in the account that gets read the wrong way. Six observations matter most, because each tends to end in a bid increase, a rebuild or a pause. Landing pages and keyword statuses come up later, where they change a decision.

A high first-page bid estimate sits next to a low score. It looks as if expensive keywords get punished. The cause runs the other way. Google says the first page bid estimate is based on the keyword's Quality Score and the competition, and a very high estimate "may mean that your keyword's Quality Score is poor." Bidding up to meet it doesn't touch the score either. Your bid "may affect your Ad Rank, but it doesn't impact the assessment of your ad quality" (About ad quality). Read a high estimate as a symptom, and check the components before you raise anything.

Ads near the top get more clicks. That tempts two moves: bidding to the top to build CTR history, or bidding up because a low position seems to be dragging CTR, and so the score, down. Neither works as intended. Google says expected CTR "is normalized for your actual position on the page" (What matters for ads quality). A top ad is expected to get more clicks than one lower down, and it is graded against that expectation. In Google's words, "You don't need to bid for higher positions to increase Quality Score." Buying the top spot may get you data sooner, but it doesn't buy a better grade. Google adds that you're therefore "free to bid to performance: the clicks, conversions and costs that work best for your business."

A new account shows low scores, or dashes. It's tempting to blame the account's age. Age isn't a documented input, but history is. For newly launched keywords, Google says it doesn't start from scratch. It starts from your related keywords, ads and landing pages, and "having a great account will influence our initial expectations of performance" until there's enough of the keyword's own data to replace them. A brand-new account has no such history to draw on, so its early grades rest on little evidence. A "—" means there aren't yet enough exactly matching searches to score the keyword at all. For a dash, Google suggests finding higher-volume keywords in Keyword Planner, or broadening match types. That doesn't contradict the exact-search rule. In Google's words, match types "don't change how your Quality Score is calculated," but broadening them "helps your keyword participate in more auctions, increasing the chances of gathering enough exact-match search data to generate a score." The broader-match searches themselves still don't count. Don't treat early scores as a verdict. In an established account, a new keyword whose related keywords, ads and landing pages are in good shape will probably start from their standing.

A score moves while nothing in the account changed. Remember what the grade compares: you against whoever else showed for that exact search over the last 90 days. Competitors changing their ads can shift it, and so can that window moving forward. For new keywords, the borrowed starting estimate also gives way to their own data. Before reacting, add the "(hist.)" versions of the Quality Score columns and segment by day to see which component moved. Rebuilding a keyword because its number twitched is work aimed at the light, not the engine.

A keyword moved into a new ad group comes back with a different score. That makes restructuring look like a lever. Google says structure itself doesn't matter. Moving keywords into new ad groups or campaigns without changing the ad text or landing page "has no effect on their quality," and there's no such thing as an ad group-level, campaign-level or account-level Quality Score. What can change quality is the ad the searcher sees: moving a keyword under new ad text can change it. Restructure only when it lets you write an ad that fits the search better.

Negatives lift the account's CTR, but keyword scores don't budge. Both halves are true. Negatives stop your ads from showing on loosely related searches, which raises blended CTR and saves money. But the score is built only from searches that exactly match the keyword, and Google says changing match types "will not impact Quality Score." By the same logic (an inference, not a Google statement), negatives that block non-exact searches leave the score alone. Add them for the money. That's reason enough.

Diagnose a stuck keyword: read the parts, not the total

In the keyword table, add the Exp. CTR, Ad Relevance and Landing Page Exp. columns. Google treats both "Below average" and "Average" as signs there may be room to improve. So work on any component that isn't Above average, starting with the weakest:

• Ad relevance. The ad doesn't speak closely enough to the search. Write the ad in the searcher's language. Split ad groups whose keywords can't all be served by one ad. Google's example: engagement rings and wedding bands belong in separate groups.

• Expected CTR. The ad is on topic but not chosen often enough. Ad writers have long argued over whether to put the searched words in the ad or to write something more appealing. It isn't either/or. The ad first has to show that it fits the search, then give a reason to pick you: a specific offer, a concrete benefit, and a call to action that matches what the page does. Google also warns that a more specific ad can lower CTR while raising conversion rate, so aim for the balance that meets your goal, not for the highest CTR.

• Landing page experience. The click lands somewhere that doesn't deliver enough. Google's checklist:

  • feature what was searched for (someone who clicked an ad for flannel shirts should land among flannel shirts);
  • follow through on the ad's offer;
  • work well on a phone, since Google says the user's device is taken into account;
  • load quickly.

  Conversion rate is a useful proxy for your own tests, though Google notes it doesn't change the landing page status.

• All three Above average. The diagnostic has nothing left to flag. Google doesn't document whether that fixes the score at its maximum, and further quality gains can still show up in what you pay per click and where your ad appears, as well as in your own CTR and conversion numbers.

• The score shows "—". There isn't enough exact-search data yet. See the new-account point above.

Pass/fail belongs to policy, not to the grade

Landing page quality is often described as an on/off switch. There are really two separate checks, and only one of them works that way.

Policy: can the ad run at all? Google's destination requirements lead to disapproval when a destination doesn't work, is unnecessarily difficult to navigate, or can't be crawled by Google Ads. Crawling is where quiet mistakes happen. Google's ads crawlers, AdsBot-Google and AdsBot-Google-Mobile, ignore the global robots.txt user agent (*), so a blanket disallow doesn't stop them. A robots.txt group that names them does, and that limits Google Ads' ability to check the page. Google also says information from any of its crawlers may be used to assess ad quality. Keep the AdsBot tokens out of any group that disallows your landing pages.

Experience: how well does the page serve the click? This is the graded component above, on the same three-level scale. It doesn't require repeating the search word for word: Google says that phrase "doesn't need to be on your landing page." It asks whether the page delivers what the search was looking for.

Delete, pause, or bid up?

The score can't settle this, because it leaves out the one thing that decides it: what a click is worth to you. Start from cost per conversion instead.

Here's a constructed example. A keyword converts 3% of its clicks, and the account's target is $40 per conversion. The most this keyword can pay per click and still hit target is $40 × 3% = $1.20. Now suppose its first-page bid estimate is $2.50.

If columns extend beyond the page, scroll sideways. With a keyboard, focus the table and use the arrow keys.

Bidding $2.50 doesn't mean paying $2.50. Actual CPC is only what's needed to clear the Ad Rank thresholds and beat the ad below you. But two points from Google's documentation push the other way:

• Lower-quality ads face higher thresholds.
• With low-quality ads, your actual CPC can sit close to your maximum "even when there is low competition."

For a low-quality keyword, then, the bottom row is the realistic risk. Bidding to the estimate pays off only if, at the price you actually end up paying, the conversion rate reaches what the table demands: about 6.25% at $2.50. Plan on the conversion rate you've measured, not one you hope a new position will bring. Unless you also change something that lifts conversions, such as the ad's promise or the page, bidding up buys position rather than fixing anything. It won't lift the score either, because expected CTR is graded for the position you're in.

With that arithmetic in hand:

• The keyword pays at its current price, despite a low score. Keep it, and work its weakest component. Google says higher-quality ads typically cost less per click, so the same position should get cheaper.
• It doesn't pay, and a component is Average or Below average. Quality and price aren't separate problems here, because better quality can lower what each click costs. Use the arithmetic to size the gap: how far would the price have to fall, or the conversion rate rise, to reach target?
  • If the gap looks closable, work in order:
    1. Give the keyword one real repair: its own ad group, ads written for exactly that search, and a page that delivers it.
    2. Review it once it has had enough clicks that a conversion or two either way wouldn't change the verdict.
    3. If it still doesn't pay, cap its Manual CPC bid at break-even ($1.20 in the example). While the measured conversion rate holds, that caps cost per conversion at target, unless bid adjustments or "Manual bidding for Search Partners" raise the cost above the cap.
    4. Pause it only if even that bid can't serve or deliver useful volume.
  • If no plausible repair could close the gap (if, say, the price would have to halve; the figure is an illustration, not a threshold), pausing now is a reasonable judgment about economics. It doesn't require believing the quality can't improve.
• It doesn't pay, and every component is already Above average. The score has nothing more to point at. Unless something else about the offer can change, cap its Manual CPC bid at break-even, with no bid adjustments or "Manual bidding for Search Partners" raising it. Pause or remove it if even that bid can't serve or deliver useful volume.
• Status: "Eligible (limited): Below first page bid estimate." Google says the keyword can show but "isn't currently showing ads on the first page of Google search results, likely due to Ad Rank." An ad-group default bid set below first-page estimates can put keywords here. It doesn't lower their scores. Whether to raise the bid is the arithmetic above.
• Status: "Rarely shown due to low Quality Score." On the reading inferred earlier, the underlying ad quality isn't clearing the bar. Google suggests improving the three components or replacing the keyword with a more relevant one.
• Status: "Low search volume." This is not a quality problem. Google says it's "unrelated to quality score, bid, or creatives." A narrower keyword can fit intent better, but go too narrow and this is where it ends up. Google's suggested fixes are a less specific keyword, a broader match type, or new ideas from Keyword Planner.

If a bid strategy sets your bids

Everything above assumes you set the bids. A bid strategy changes who does the arithmetic, not the quality underneath it. Google's current guidance:

• Conversion-focused Smart Bidding sets bids in each auction toward conversions or conversion value. The strategies Google lists include Target CPA, Target ROAS, Maximize Conversions and Maximize Conversion Value. With Target CPA, you name the cost per conversion and the strategy pursues it, so the bid-up question becomes the strategy's job. These strategies need conversion tracking set up correctly. Google's guidance also implies they need enough conversion data, because it suggests Maximize Clicks for new campaigns that lack sufficient conversion data for Target CPA or Target ROAS.
• Maximize Clicks gets as many clicks as the budget allows. Google says it can suit new campaigns short on conversion data, or traffic goals, but that it isn't a good choice if your goals include a cost per conversion.
• Manual CPC lets you set bids keyword by keyword, which is the control the arithmetic above needs.

A new account with a cost-per-conversion goal and little conversion data therefore faces a real choice. Manual CPC, with the arithmetic above, keeps the goal in view while conversions accumulate. Maximize Clicks gathers traffic and data, but it bids for clicks, not for your goal. The trap is running a clicks strategy against a conversion goal, then reading the resulting bill as a Quality Score problem.

What to say when someone asks for "the Quality Score"

Clients and managers often want one number. Google's guidance is not to aggregate this one. A more useful report has two parts:

• how much spend sits on keywords with a component marked Below average or Average, and which component it is;
• the cost per conversion of those keywords, next to everything else.

That tells the reader where the fixable problem is and what it's costing. If someone points to "Rarely shown due to low Quality Score" as proof that the number itself matters, the answer is the reading above, which is an inference rather than a Google statement: the status is a symptom of the underlying quality, and the remedy runs through the components. The conversation stays on that quality and what it costs, rather than on a 1–10 number that Google says isn't an auction input.

The number tells you where to look. The three components tell you what to fix. Cost per conversion tells you whether the fix was worth making.