Quality Score Is Not a KPI: What to Optimize Instead
Quality Score is a diagnostic readout, not a goal. Chase conversions and the inputs behind QS instead.
The short version
- Quality Score is a lagging, rounded diagnostic, not a target; the live auction uses a separate real-time estimate you never see.
- Optimize conversions, CPA, and Search Lost IS (rank) first; QS components are levers, not goals.
- The search terms report fixes the root cause (query mismatch) that drags down both QS and conversion rate.
- Trust CPA over QS when they disagree: never pause a profitable keyword to raise a cosmetic score.
- Ignore QS under roughly 100 clicks of history, and never chase CTR at the expense of conversion rate.
Quality Score is a 1-to-10 diagnostic, not a performance target. Optimizing to raise the number is backwards: the number is a lagging summary of things you should already be fixing for conversions and CPA. Fix the three inputs behind it (expected CTR, ad relevance, landing page experience) in service of profit, and the score moves on its own. If it does not, you have learned something useful instead of gaming a vanity metric.
Why Quality Score behaves like a thermometer, not a dial
Quality Score is a rounded, keyword-level estimate Google surfaces to explain, after the fact, how it expects your keyword to perform against the query. It rolls up three components: expected clickthrough rate, ad relevance, and landing page experience. Each is reported as Below average, Average, or Above average in the keyword columns (add them via Campaigns > Keywords > Columns > Modify columns). The 1-to-10 integer is a compression of those three signals plus history.
The trap is treating a compressed, rounded, lagging summary as something to optimize directly. You cannot push QS. You can only change the underlying behavior it measures, and QS updates when Google has enough fresh impression and click data to re-estimate. That lag is why accounts that make a real improvement often see QS move days or weeks later, and why a keyword with almost no recent traffic can show a stale score that reflects conditions from a month ago.
More importantly, the live auction does not use the historical 1-to-10 QS you see in the UI. At auction time Google computes a real-time quality estimate for that specific query, device, location, and context. The reported Quality Score is a simplified diagnostic built for humans. So two things follow: raising the displayed number does not guarantee the auction treats you better, and a mediocre displayed QS on a tightly converting keyword is not a problem to solve.
Optimize these instead, in priority order
Start with the outcomes that pay you, then work backward into the inputs that QS happens to measure. The ranking below is the order I touch things in a real account.
Conversions and cost per conversion come first because they are the only metrics tied to money. Then Search Lost IS (rank), which tells you whether the auction is actually penalizing you, which is the real worry QS is a proxy for. Then the three QS components, because they are the levers, not the score.
Concretely, work the funnel in this order: match the query to the ad, match the ad to the landing page, and let CTR follow. Expected CTR is the hardest to move by hand because it is partly a function of position, which is partly a function of bid and budget. Chasing CTR directly often means writing clickbait headlines that tank conversion rate, which is the exact wrong trade.
- Conversions and CPA or ROAS: the only KPIs that map to money.
- Search Lost IS (rank) in Campaigns > Insights and reports: tells you if quality or bid is actually costing you impressions.
- Ad relevance component: fix by getting the keyword into the ad and splitting mismatched keywords into their own ad group.
- Landing page experience component: fix with page speed, mobile usability, and message match to the ad.
- Expected CTR component: move it last, and never at the cost of conversion rate.
The search terms report does more for you than Quality Score ever will
If you only have 30 minutes today, spend it in Campaigns > Insights and reports > Search terms, not staring at QS columns. Quality Score tells you a keyword is underperforming. The search terms report tells you why: it shows the actual queries your keywords matched, which is where broad and phrase match quietly spend money on irrelevant intent.
The mechanism matters here. Low ad relevance and low expected CTR are frequently caused by a single keyword matching a wide spread of queries, some of which your ad does not address. When your ad matches 'cheap running shoes' and your landing page sells premium carbon-plate racers, both ad relevance and landing page experience drop, and so does conversion rate. Pruning the irrelevant queries with negatives and tightening match types fixes the root cause. QS improves as a side effect.
In practice a single review pass of an active account routinely surfaces dozens of negative-worthy terms from a few hundred reviewed, because broad match casts wide and most queries it catches are not your buyer. You do not need to find all of them. You need to catch the spenders. Tools like QueryCut automate the term-by-term triage, but you can do the same manually: sort by cost, read top to bottom, and add negatives for anything that does not match intent.
Decision rules you can apply this afternoon
Rules beat vibes. Here is the heuristic set I use, with the tradeoff behind each number stated plainly so you can adjust it to your own volume.
These are practitioner thresholds, not statistics. The logic is always the same: wait for enough data that a single random event will not flip your decision, but not so long that you keep bleeding money on an obvious loser.
The judgment call Google hides from you: you cannot see the real-time auction quality estimate, only the rounded historical score. So when a keyword has a low displayed QS but strong CPA, trust the CPA. The money is real; the score is a lossy summary.
- Ignore QS below roughly 100 clicks of keyword history. Below that, one missed or extra conversion swings both CPA and the score, so any read is noise.
- If a search term has spent 3x your target CPA with zero conversions and its last click was more than 30 days ago, add it as a negative. Under that spend, or with recent clicks, wait.
- If a keyword converts at or below target CPA but shows QS of 3 or 4, leave it alone. Do not rewrite a profitable ad to chase a diagnostic.
- If QS dropped on a keyword that used to score 7+, check the search terms report first: a drop usually means match type widened or query mix shifted, not that your ad got worse.
- If ad relevance is Below average, the fastest fix is splitting that keyword into its own ad group with an ad that uses the keyword in the headline.
The failure mode: optimizing the score and losing the account
The classic mistake is building the account around Quality Score as a scorecard, then making changes that lift the number while hurting the business. Pausing low-QS keywords that happen to convert is the most expensive version: you remove profit to clean up a cosmetic metric. A low-QS keyword with good CPA is telling you the diagnostic and the money disagree, and the money wins.
The second failure mode is writing to CTR. You can lift expected CTR with aggressive, curiosity-gap headlines that pull clicks from unqualified searchers. CTR goes up, QS may tick up, conversion rate falls, and CPA rises. You optimized the proxy and broke the target.
Be honest about the limits. You cannot know the exact weighting Google uses to turn three components into one integer, you cannot see the live auction estimate, and you cannot force an update on a keyword that is not getting impressions. What you can control is intent match: the right query reaching the right ad reaching the right page. Do that for CPA, and Quality Score takes care of itself.
Questions people also ask
Does a higher Quality Score lower my cost per click?
Indirectly and not reliably. Ad Rank factors in a quality estimate, so better quality can reduce what you pay for a given position. But the figure the auction uses is a real-time estimate for each specific query, not the 1-to-10 number in your UI. Optimize for conversions and the quality inputs follow; do not expect the displayed score to predict your CPC precisely.
Should I pause keywords with a Quality Score of 1 or 2?
Only if they also fail on money. Check conversions and CPA first. A keyword scoring 2 that converts at your target is doing its job and should stay. Pause when the term has spent well past your target CPA with no conversions, not because the score is low.
How long after I make changes does Quality Score update?
It updates as Google accumulates fresh impression and click data for that keyword, which can take days to weeks depending on traffic. Low-volume keywords can show stale scores for a long time because there is not enough new data to re-estimate. This lag is why QS is a poor real-time KPI.
If not Quality Score, what single report should I check first?
The search terms report, under Campaigns > Insights and reports > Search terms. Sort by cost and read down. It shows the actual queries your keywords matched, which is where wasted spend and low relevance originate. Fixing query mismatch there improves conversions and the QS components at the same time.
Is expected CTR worth optimizing directly?
Last, and carefully. Expected CTR is partly a function of ad position, which depends on bid and budget, so it is hard to move by hand. Writing for clicks alone risks pulling in unqualified traffic that lowers conversion rate. Improve it through better keyword-to-ad relevance, not clickbait.