Main area
Agency
google-ads

Quality Score — what really drives it and how to lift it

Redaktion ·

Why Quality Score matters

Two accounts, the same market, the same keywords — and yet one pays €1.80 per click while the other pays €4.40. The difference is rarely in the bid strategy. It sits inside Quality Score and the auction logic behind it. Landing on a 5 instead of an 8 means paying noticeably more for the same position — or getting buried under the next competitor even though, on paper, your max bid should win.

Quality Score isn’t a mystery, but it isn’t a single lever either. It’s a diagnostic value shown in the UI, composed of three sub-components, recalculated live in every auction, and it intervenes in three places: in Ad Rank, in CPC, and in whether your ad shows at all. This article breaks the score apart, places it inside the auction, and shows where you can realistically lift it — and where you can’t.

The mechanics: Quality Score, Ad Rank, CPC

Three terms belong together, otherwise the thread gets lost.

Quality Score is the 1–10 value you see at the keyword level. It exists per keyword, not per ad group, not per account. Three components feed in, each shown as Above / Average / Below average:

  • Expected CTR — the probability your ad gets clicked for this keyword, normalized for position and format.
  • Ad relevance — how closely the ad text matches the keyword’s intent.
  • Landing page experience — how well the destination page matches the keyword and ad in content, technical performance, and usability.

Ad Rank is what actually decides the auction — who shows where, top, bottom, or not at all. Ad Rank is a function of bid, an auction-time quality estimate, the auction context (search query, device, location, time of day, competitors), and the expected impact of ad assets and formats. The Quality Score you see in the UI is a simplification of that auction-time quality — useful as diagnosis, not identical to what actually drives the moment.

CPC — what you actually pay per click — is roughly the next competitor’s Ad Rank divided by your auction quality, plus 1 cent. Practical consequence: higher auction quality → lower CPC at the same position. The lever is non-linear: jumping from 4 to 7 is worth far more than from 7 to 8.

Expected CTR — the hardest lever

Expected CTR is normalized. Google factors out that position 1 gets more clicks than position 4 and that ads with sitelinks naturally get clicked more. What’s left is an estimate: if your ad and a competitor’s stood in the exact same position — who would get more clicks? That makes expected CTR the toughest lever: it reflects whether your ad text is competitive, not whether you happen to be well placed today.

You lift it mainly through the ad itself — headlines that pick up the keyword and state a clear benefit — and through strict match-type hygiene: a keyword that triggers on the wrong query can never produce good CTR, because the search intent is off.

Ad relevance — the easiest diagnosis

Ad relevance measures the semantic proximity between keyword and ad text. Is the keyword in the headline set? Is the topic addressed? An ad group with 80 keywords spanning five themes will almost always score “Below average” here — the text can’t fit everything.

This component is the easiest to fix because it’s structural: tighten ad group themes, cap keywords per group (Hagakure logic: one theme per group), and put your top keywords explicitly in at least one headline of an RSA.

Landing page experience — the most expensive piece

Landing page experience is the black box with the most variables: page-content match with keyword and ad, mobile usability, load time, transparency (what do you offer, who are you, privacy), navigation clarity. Google measures not just content relevance — it also measures how users behave after the click.

This component changes the slowest. A new headline takes effect in days; a new landing page in weeks. That’s exactly why many accounts permanently leave 0.5–2 points on the table here.

Expected CTR, Ad Strength, RSA — how they connect

Expected CTR (eCTR) is the predictive component of Quality Score. Ad Strength is a qualitative UI indicator (Poor / Average / Good / Excellent) Google attaches to RSAs — based on number of unique headlines and descriptions, pinning strategy, keyword presence, and message diversity. Ad Strength is not a Quality Score factor. Google has stated this repeatedly. Still, the two correlate, because a broad, well-diversified RSA gives Google more material to test in the auction — which feeds back into eCTR.

In practice: an RSA with 14 unique headlines, 4 descriptions, 0–1 pins, and at least three headline variants carrying the keyword gives the auction room to work. An RSA with three pinned headlines can never reach “Excellent” — and it lacks the material eCTR needs to develop.

Responsive Search Ads have been the only allowed format for new search ads since 2022. With them, pinning is the most common form of self-sabotage: pin two headlines to position 1 and three to position 2 and you’re letting Google test roughly 30 combinations. Without pins, it’s around 43,680. Pinning is a tool for compliance text (legal disclaimers, brand spellings) — not for CTR optimization.

Pitfalls: what really drags the score down

Three patterns show up most often in audits.

Mixed ad groups. A classic: 40 keywords covering “tax advisor Hamburg” plus “payroll Hamburg” plus “GmbH formation” in one group, paired with an RSA carrying generic copy. Ad relevance can only land “Below average” here, because the text fits everything a little and nothing properly. Fix: separate themes, one ad group per search intent, singular and plural variants of the same keyword in the same group.

Wrong match types without clean negatives. Broad match without negative-keyword hygiene pulls in queries that don’t fit the landing page. The ad gets impressions for searches it wasn’t built for — clicks don’t follow, eCTR drops, Quality Score follows. Tip: search-term reports weekly, kick irrelevant terms out as negatives.

Landing pages that don’t deliver. An ad for “outsource bookkeeping” lands on a generic “/services” page covering ten topics — specificity gets lost. Google measures (in simplified terms): does someone bounce, scroll, interact? If the answer to all is no, landing page experience drops. Fix: dedicated landing pages per theme, or at least per top keyword.

Three levers that actually work

Account structure: one theme per ad group

By far the biggest lever on ad relevance, and indirectly on all three components. Cut ad groups tight enough that one RSA can really fit every keyword in the group. Hagakure (one ad group per theme, paired with broad match types and Smart Bidding) is the modern standard — SKAG (one keyword per ad group) is over-tuned and starves Smart Bidding of data.

In practice that means 5–20 keywords per ad group, all sharing the same intent. If you’re tempted to drop in a new keyword that’s slightly off-theme — open a new ad group instead.

RSA hygiene: feed the algorithm, minimize pins

A productive RSA looks like this:

  • 12–15 headlines (the maximum), genuinely diversified: three carrying the keyword, three with USP, three with CTA, the rest with trust signals / social proof.
  • 4 descriptions, each actually using 80–90 characters.
  • 0–1 pinned headlines, only when compliance demands it.
  • At least four asset types active: sitelinks (with descriptions!), callouts, structured snippets, plus lead forms or image assets where they fit.

Ad Strength “Good” or above is the minimum target. “Excellent” isn’t the end state — well-performing RSAs can sit at “Good” because the set is intentionally focused. But “Average” and “Poor” are leaking measurable performance.

Landing page: match search intent, not brand layout

The landing page has to deliver in the first 1–2 seconds after the click: H1 contains the keyword (or a very close variant), the sub-headline answers the typical question behind the search, a clear CTA sits above the fold. Mobile LCP under 3 seconds, no layout shift. Trust visible: address, imprint, reviews — whatever the market expects.

A pragmatic rule: lay the ad text and the landing page header side by side. If they don’t carry the same message, you have a gap. That gap costs Quality Score — and conversion alongside it.

Diagnosis: what the UI tells you — and what it doesn’t

In the account you see Quality Score at the keyword level with three status columns: Expected CTR, Ad Relevance, Landing Page Experience — each as Above / Average / Below average. You have to enable those columns explicitly (Columns → Quality Score). A Quality Score of 7 with “Average” across all three is a completely different diagnosis than a Quality Score of 7 with “Above average” eCTR and “Below average” landing page — same score, very different levers.

| Diagnosis pattern | Likely cause | First lever | |---|---|---| | Ad relevance “Below” | Ad group too broad, keyword not in headline | Split the group, keyword into 2–3 headlines | | eCTR “Below” | Generic ad, weak RSA diversity, wrong match types | Rework RSAs, add negatives | | Landing page “Below” | Generic LP, slow, no intent match | Dedicated LP, LCP < 3 s, H1 = keyword match | | Score swings wildly | Low search volume, statistical noise | Look at ad-group trends, not daily numbers |

Important: for low-volume keywords Google often shows no score at all (—). That’s not a bug — it just means the data base isn’t enough. In new accounts that can stay this way for weeks.

Two practical examples

Account A — B2B SaaS, mature account, average Quality Score 5. Diagnosis: 60 % of ad groups had > 30 keywords, RSAs ran 2–3 pinned headlines per position, one generic landing page covered three product lines. Actions: ad groups cut to one product theme each (from 18 to 47 groups), pins removed except for one compliance headline, three dedicated landing pages built with product-specific H1. After eight weeks: average Quality Score 7.1, average CPC −22 %, conversion rate +14 %.

Account B — local service business, new account, first four weeks. Diagnosis: score — on most keywords (insufficient data), eCTR “Below average” on the few measurable ones. Actions: RSAs filled out to 14 headlines, search-term report surfaced 60+ irrelevant queries (job searches, free-this-or-that searches) → added as negatives. After four more weeks: eCTR moved to “Above average” on 70 % of keywords, CPC down 35 % — the first optimization was structural, not creative.

FAQ

Does the bid strategy affect Quality Score?
No. Smart Bidding uses its own auction-time signals; the displayed Quality Score isn't directly changed by it. Indirectly, yes: Smart Bidding can lift CTR and conversion rate at the auction level, which over time pulls the displayed score along.
Should I pause keywords sitting at Quality Score 1–3?
Often yes, but not reflexively. Check first: is there a better ad group for this keyword? Does the landing page match? If yes → move it. If no and conversion rate is also weak → pause. Low Quality Scores in your account don't drag the average down in some mystical sense — they're a symptom you should treat.
Does serving a "personalized" landing page via URL parameters help?
Not really. The bot evaluating the landing page sees the default. Server-side personalization only helps if it improves the actual content match — dynamic H1, dynamic subline. Plain URL tracking parameters change nothing about landing page experience.
Quality Score 10 is the goal, right?
No. Quality Score 10 is a vanity metric, rarely profitable. A 7–8 with strong conversion rate beats a 10 with weak conversion rate every day. Optimize for profit (POAS, ROAS), not for the score.

Conclusion

Quality Score is a diagnosis, not a switch. It emerges from three components — expected CTR, ad relevance, landing page experience — that respond at different speeds. Ad relevance moves in days through structural fixes (split ad groups). eCTR needs ad-text work and clean match-type hygiene and shifts in 1–4 weeks. Landing page experience is the slowest dial — and often the one with the largest lever.

Ad Strength isn’t Quality Score, but it’s a useful heuristic for whether your RSA gives the auction enough material. “Good” is usually enough, “Excellent” is nice, “Average” or “Poor” leaves performance on the table.

If you take one thing from this article: cut your ad groups tighter. That’s the single structural change that lifts all three Quality Score components at once — and gives Smart Bidding cleaner signals at the same time. Everything else — headline tests, landing page refresh, Ad Strength tuning — only works properly once the structure is right.

Themenuebersicht

google-ads