---
name: improve-landing-page-experience
description: >-
  A one-time repair job for the landing-page component of Quality Score: find the keywords Google has
  rated Below Average on landing-page experience, diagnose which baseline requirement the page fails
  (message match, clarity, speed, mobile, focus, trust), fix the foundations, and confirm the rating
  recovers over 14-30 days. VigilDog reads the per-keyword rating and spend via gads_run_gaql_query and
  gads_get_keyword_performance, pulls the serving ad copy with gads_get_ad for message-match
  comparison, and tracks behavioral proxies through ga4_run_report; speed tests and page edits are
  human work. Use it when the Below Average label, 70%+ bounce, or a collapsed mobile CVR shows up. Not
  for ad-to-keyword relevance (improve-ad-relevance), weak ad creative (improve-expected-ctr), a full
  conversion-focused audit (audit-and-optimize-an-existing-landing-page), or ongoing testing
  (run-a-landing-page-a-b-test).
---
# Repair the Landing-Page Component of Quality Score

## Purpose

Google scores each keyword's landing-page experience by asking whether the page delivers what the ad
promised, quickly, usably, and credibly - and a Below Average rating drags Ad Rank and inflates CPCs.
This skill locates the rated-down keywords, identifies which foundational requirement the page fails,
repairs it, and verifies the rating climbs back to Average or better. It is a repair, not a program:
ongoing optimization belongs to testing skills.

## When to run

- Any spend-carrying keyword shows landing-page experience = Below Average.
- Bounce rate on a paid landing page runs at 70% or higher.
- Mobile CVR is under half of desktop CVR for the same page.
- Conversion rate sits far below vertical norms and the click-side metrics look fine.
- Runs fine in parallel with improve-ad-relevance or improve-expected-ctr - unlike those two, this
  component depends on the page, not the ad.

## When NOT to run

- The weak Quality Score component is ad relevance → improve-ad-relevance.
- The weak component is expected CTR → improve-expected-ctr.
- You want a full conversion audit with prioritized fixes beyond the QS baseline →
  audit-and-optimize-an-existing-landing-page.
- The page already rates Average+ and you want incremental lift → run-a-landing-page-a-b-test.
- For Display/Video traffic: the QS framing barely applies - judge pages there on conversion rate.

## Prerequisites

- Search campaigns with enough impression volume for Google to rate the keywords.
- Ability to edit the landing pages (or a channel to whoever can).
- GA4 collecting on the affected pages.
- Access to an external page-speed tester (PageSpeed Insights / Lighthouse) - outside VigilDog.

## Procedure

1. **Map the damage.** Use `gads_run_gaql_query` on `keyword_view` selecting the keyword text, the
   post-click quality rating, cost, and clicks, filtered to Below Average, sorted by cost. This is
   the priority list - fix the pages behind the most expensive rated-down keywords first.
2. **Group keywords by final URL.** Use `gads_get_keyword_performance` (with final URLs) so the work
   is organized per page, not per keyword; one page fix usually clears many keywords.
3. **Pull the promise.** For the top URLs, fetch the serving ads with `gads_get_ad` (or a GAQL pull of
   ad headlines per ad group). Record what offer, wording, and action each ad promises.
4. **Message-match audit.** Compare each top keyword and its ad headlines against the live page
   headline: does the page immediately confirm the visitor is in the right place, is the advertised
   offer visible, does the keyword's theme appear in the headline? The comparison can be done by the
   model from fetched page content; record match / partial / miss per keyword.
5. **Clarity audit.** Look at the page cold for a few seconds and answer: what is offered, who is it
   for, what should I do next, why believe it? Two or more unanswerable questions marks the hero
   section as a repair target.
6. **Behavioral audit.** Use `ga4_run_report` on the affected page paths: bounce rate, engagement
   rate, average session duration, and CVR split by device. These tell you whether the failure is
   at-arrival (bounce) or during-consideration (engaged but not converting).
7. **HUMAN STEP (outside VigilDog):** technical audit with PageSpeed Insights on each affected URL,
   mobile and desktop. Record the scores and the three field metrics: largest-paint time, input
   responsiveness, and layout shift. Also test readability and tap-target usability on a real phone.
8. **Focus and trust audit.** Count every clickable element versus links pointing at the conversion
   goal; check for full site navigation, footer link farms, and social icons. Verify company name,
   contact details, privacy policy, and some credibility proof are present - Google reads
   transparency signals.
9. **HUMAN STEP (outside VigilDog):** repair, in order of measured damage: (a) align the page headline
   and offer to the top keywords' ads - or add dynamic keyword-driven headline insertion via URL
   parameter if many keyword variants share the page; (b) rewrite the hero for instant clarity;
   (c) strip distracting links toward a one-goal page; (d) add missing contact/trust elements;
   (e) compress images, defer scripts, and cache to pass the speed floor; (f) fix mobile font sizes,
   tap targets, and form usability.
10. **Escalate if the offer is the problem.** If no compelling value statement, proof, or
    risk-reversal exists to put on the page, pause: the page cannot outrun a weak offer. Route to
    create-an-irresistible-offer.
11. **Validate over 14-30 days.** Weekly, re-run the step 1 query and watch labels move off Below
    Average; track CVR and bounce trends via `ga4_run_report`. Conversion rate improving is the
    leading indicator - Google treats it as a proxy for a satisfying page.
12. **Route the exit.** Labels recovered but other QS components still weak → improve-ad-relevance or
    improve-expected-ctr. All components healthy → ongoing CRO via run-a-landing-page-a-b-test. No
    movement after 30 days → re-audit; the likely misses are message match or insufficient traffic
    for a re-rating.

## Decision rules

- **Priority:** fix pages by rated-down spend, not by page count - one URL behind 60% of Below
  Average cost is the whole job.
- **Clickable-to-goal ratio:** aim near 1:1 on a dedicated page; above roughly 10 clickable elements
  per conversion action, restructure rather than trim.
- **Speed floor:** mobile PageSpeed score of at least 50 (70+ preferred); largest-paint under 2.5s,
  input response under 200ms, layout shift under 0.1. Do not chase 100 by deleting conversion
  elements - a slightly slower page that converts beats a fast one that doesn't.
- **Mobile floor:** 16px minimum body text, 44px minimum tap targets, no horizontal scrolling, forms
  completable with a thumb.
- **Patience window:** labels typically refresh 14-30 days after fixes, faster on high-traffic
  keywords; do not re-churn the page inside that window.
- **Homepage rule:** homepages rarely rate well as paid destinations (too many goals, too many
  exits); brand campaigns are the main exception.
- **Split-page trigger:** when keywords on one URL carry fundamentally different intent (buy vs
  compare), give each intent its own page instead of one compromise page.

## Common failure modes

- **Rewriting the page but not the match.** Teams polish design while the headline still ignores the
  keyword; the rating stays down. Message match is the highest-frequency root cause - check it first.
- **Fixing the ad instead of the page.** This component is about the post-click experience; ad edits
  belong to the other two QS skills.
- **Speed-tunnel-vision.** Passing the speed floor with a page that answers nothing still rates
  poorly; clarity and relevance carry more weight than the last 20 speed points.
- **Judging in week one.** The label lags fixes by weeks; premature verdicts trigger useless
  re-churn. Watch CVR as the early signal instead.
- **Leaving trust gaps on lead-gen pages.** Asking for contact details while showing no address,
  phone, or policy links suppresses both conversions and the transparency assessment.

## Related skills

- **Parallel:** improve-ad-relevance, improve-expected-ctr (the other two QS components; this one can
  run alongside them), improve-quality-score (the umbrella triage).
- **After:** run-a-landing-page-a-b-test (ongoing optimization once baseline passes).
- **Instead, when broader:** audit-and-optimize-an-existing-landing-page (full conversion audit).
- **Upstream blocker:** create-an-irresistible-offer (when the offer, not the page, is weak).