Landing page match scorer

Landing Page Match Scorer

A skill for scoring Google Ads landing pages on alignment with keywords and ads, then producing specific change recommendations.

What this skill does

For each landing page connected to a Google Ads ad group, scores 7 dimensions (1–10) and produces:

1. Overall match score (weighted average)
2. Per-dimension breakdown showing weak spots
3. Specific change recommendations (exact rewrites where possible)
4. Expected QS component improvement
5. Priority ranking (Critical / High / Medium)

The seven dimensions

1. Message match - does the page headline echo the ad and keyword intent? (2× weight)
2. Keyword presence - keywords appear in H1, subheads, body? (1× weight)
3. Intent alignment - does the page satisfy what the searcher wanted? (2× weight)
4. CTA clarity - one clear next step above the fold? (1× weight)
5. Trust signals - reviews, testimonials, security badges, guarantees? (1× weight)
6. Mobile experience - load speed, layout, tap targets (1× weight)
7. Form friction - number of fields, required info, perceived effort (1× weight)

Dimensions 1 and 3 carry double weight because they're the strongest predictors of both Quality Score's Landing Page Experience component AND conversion rate.

Required inputs

1. Ad group keywords - the keywords sending traffic to this page
2. Current ad copy - headlines and descriptions
3. Landing page - URL (the skill can fetch via web_fetch) OR pasted page content (headers, body, CTA, form fields)

If only a URL is provided, use web_fetch to retrieve the page. If web_fetch is unavailable in the user's setup, ask them to paste the page text directly.

Workflow

Step 1: Gather inputs

If anything is missing, ask:

To score the landing page match, I need:

1. Ad group keywords - list 3–10 of the main ones
2. Current ad copy - paste a representative ad's headlines and descriptions
3. Landing page URL - I'll fetch the content, or you can paste the page text directly (header, body copy, CTA text, form fields)

Step 2: Fetch and parse the page

Use web_fetch on the URL. Extract:

• H1 (main headline)
• Subheadlines (H2s)
• Hero copy (first 100 words)
• Visible CTAs
• Trust signals (reviews count, ratings, customer logos, badges)
• Form fields (count and required ones)
• Below-the-fold content sections

For pasted content, parse manually from the structure provided.

Step 3: Score each dimension

Use the rubric in references/scoring-rubric.md. Quick summary per dimension:

Message match (×2): 10 = page H1 directly echoes the ad H1 / keyword intent. 5 = somewhat related. 1 = page is generic homepage.

Keyword presence (×1): 10 = primary keyword in H1, secondary in H2s and body. 5 = keyword appears in body but not headlines. 1 = keyword absent.

Intent alignment (×2): 10 = page is exactly what the searcher wanted (e.g., "pricing" keyword → /pricing page). 5 = related but not direct. 1 = wrong intent (e.g., "buy X" keyword → /about page).

CTA clarity (×1): 10 = ONE clear CTA above the fold, action-oriented text. 5 = multiple competing CTAs. 1 = no clear CTA visible.

Trust signals (×1): 10 = reviews + rating + customer logos + guarantee + security badges. 5 = some signals but not prominent. 1 = none.

Mobile experience (×1): 10 = scores 80+ on mobile PageSpeed AND visually clean on mobile. 5 = average. 1 = slow load OR broken layout.

Form friction (×1): 10 = ≤3 fields for mid-funnel ask, ≤6 for high-intent ask. 5 = 4–6 / 7–9 fields. 1 = excessive fields for the ask.

Step 4: Compute weighted overall score

overall = (msg_match × 2 + intent × 2 + kw_presence + cta + trust + mobile + form) / 9

Score interpretation:

• 8.0–10.0: page is well-aligned
• 6.0–7.9: page works but has improvement room
• 4.0–5.9: page has real problems hurting QS and conv. rate
• Below 4.0: page is the wrong destination - consider building a dedicated landing page

Step 5: Recommend specific changes

For each dimension scoring below 7, provide:

• What's wrong (specific observation)
• What to change (specific rewrite or action)
• Expected QS impact (which component improves: Expected CTR / Ad Relevance / Landing Page Experience)
• Expected conv. rate impact (rough range)
• Priority (Critical / High / Medium)

For Message Match and Intent Alignment fixes, write the actual proposed H1/H2 rewrites - don't just say "improve message match," show the new headline.

Step 6: Output

A markdown response with:

1. Headline summary: "Overall match score: X/10. Top issue: [dimension]."
2. Score table (all 7 dimensions with scores)
3. Specific changes section - per dimension scoring <7
4. Implementation order - what to fix first

If user requests a "report for my dev team" or "deliverable", produce a Word doc using the docx skill with the breakdown formatted for cross-functional sharing.

What this skill must NOT do

• Don't score without seeing the page. Refusing the score request without page content is correct behavior - guessing scores from URL alone is useless.
• Don't recommend changes that require backend infrastructure changes without flagging the cost. "Add personalized content" is a 6-month project, not a quick win.
• Don't promise specific QS point improvements. Quality Score is a black box; you can predict direction but not magnitude.
• Don't recommend cookie banners, exit-intent popups, or other patterns Google has noted as harming Landing Page Experience.
• Don't ignore mobile. Most B2C traffic is mobile; mobile failures are usually the dominant issue.
• Don't recommend A/B testing without acknowledging traffic constraints. Pages with <500 visits/month don't have stat-power for A/B.

Reference files

• references/scoring-rubric.md - full 1–10 scoring criteria for each dimension
• references/common-fixes.md - specific fixes per dimension with rewriting examples