AI ads infrastructure

AI Google Ads Quality Score Audit (in Claude)

Run a full Quality Score audit in Claude. See which keywords are dragging your account, why, and the fastest fixes for each component.

NotFair Team|

Quality Score directly affects what you pay per click and how often your ads show. Most accounts have 10–20% of spend going to keywords with QS 5 or below — Google's tax on poorly aligned ads. Claude can audit your full account in 60 seconds, surface the laggards, and explain which of the three QS components (expected CTR, ad relevance, landing page experience) is dragging each one.

The prompt

You: "Run a quality score audit. Group keywords by QS bucket (1-3, 4-6, 7-10). For everything below 7, show the spend share, the dragging component (CTR, ad relevance, or landing page), and the top 3 fix recommendations per ad group."

What Claude pulls

  • ad_group_criterion.quality_info.quality_score and the three component scores (creative_quality_score, post_click_quality_score, search_predicted_ctr).
  • Spend per keyword joined to QSso Claude can compute spend share by QS bucket.
  • The active ads in each ad group (for ad relevance diagnosis).
  • Final URLs (for landing page experience diagnosis).

Sample output

Claude: "Audit complete. Account-wide weighted QS: 6.2.

**QS 1-3 (low):** 47 keywords, $1,240 spend (8% of total). Mostly dragged by 'below average' ad relevance — your ads don't include the keyword text. Top fix: pin the keyword token into headline 1 of the RSA in 'Boarding > Overnight Stays' and 4 other ad groups.

**QS 4-6 (medium):** 312 keywords, $4,800 spend (32%). Mixed dragging — 60% landing page experience, 40% CTR. Landing page issue concentrated in your /services page; mobile load time is 4.8s.

**QS 7-10 (good):** 521 keywords, $9,100 spend (60%).

Top 3 fixes (in priority order): 1. Add keyword pinning to RSAs in 5 underperforming ad groups (would lift ~47 keywords from QS 1-3 to 4-6). 2. Fix /services mobile load time — currently 4.8s, target <2.5s. 3. Test 3 new RSA variants in 'Daycare' ad group where expected CTR is 'below average' on all keywords."

Apply the fixes through Claude

  • Update RSA assetsClaude calls updateAdAssets with new headlines that include the keyword tokens. You confirm before each ad group.
  • Pause low-QS keywords with no path to improvement"pause keywords with QS 1-2 where spend < $20 and clicks < 5 in 30 days." Some keywords are just not worth fixing.
  • Generate fresh RSA variants"write 3 new RSA variants for the Daycare ad group focused on 'cage-free' and 'overnight available' angles."

Landing page fixes happen outside Google Ads — Claude can flag the issue but the fix is in your CMS. We've found that the spend-weighted QS lift of 0.5–1.0 from a focused audit pays for itself within 2 weeks at most account sizes.

FAQ

Try MCP with Google Ads

Connect your Google Ads account to NotFair in 30 seconds and start querying campaigns from Claude.

Connect Google Ads

FAQ

Common questions about Model Context Protocol.

What is a good Quality Score?

QS 7+ is healthy. QS 8–10 is excellent and meaningfully reduces your effective CPC. QS below 5 is a flag; below 3 means you're paying a premium just to show. The metric is spend-weighted — improving QS on a $4K/month keyword matters more than improving it on a $40/month one.

How fast does Quality Score update after I make changes?

Google recalculates QS on each auction, but the displayed score in the Ads UI updates more slowly — often 24–72 hours. Don't panic if you don't see the score move the same day you ship the fix.

Can Claude fix landing page issues?

Not directly — landing pages are outside the Google Ads API. Claude can identify which pages are dragging which keywords, but the fix happens in your CMS or with your dev team. Claude can also draft a brief for the dev team if you want.

Does Quality Score matter for Performance Max?

PMax doesn't expose keyword-level QS the same way Search does. Claude can still audit asset quality and ad relevance signals on PMax, but the lever is different — it's about asset diversity and conversion signals more than the classic QS components.