The AdSense readiness audit scored 7 of 8 green against Google’s reviewer checklist. The single red item: “diverse and engaging content in a variety of formats (articles, images and videos).” Score against that criterion: zero of thirty-eight pillar pages had a diagram, chart, or image. Not a few. Zero.
Six SVG diagrams went live on 2 July. The media-diversity score moved from zero to nonzero. The first rejection landed eight days later anyway.
That was the beginning of a six-week sequence that turned an AdSense remediation effort into a full site integrity audit.
Three rejections, same verdict
AdSense returned “Low value content” on 10 July 2026. Ads.txt: Authorised. Policy issues: empty. A pure content-quality judgment with no configuration explanation on offer.
The first remediation was a literal match against Google’s inventory-value policy: the /resources/blog and /resources/guides hub pages were empty shells. Listed, navigable, nothing behind them. Real listings shipped 11 July. Soft 404s became live content.
The re-review request went in 24 July — fourteen days after the initial verdict, after a pre-submission sweep covering twenty live pages: all 200, zero broken links. Third rejection: 1 August, 16:26 BST. Dashboard-confirmed, policyIssues field empty again.
A 15-agent research review ran on 1 August. It pulled Google’s official documentation, the September 2025 Quality Rater Guidelines update, and the current 2026 spam-policy state. What came back was a clearer picture of what “Low value content” measures in practice — and of what the Q.0 integrity sweep two weeks earlier had been pointing at.
What the integrity sweep found
The Q.0 sprint emerged from a three-lens assessment on 20 July: content/SEO, design/UX, and architecture agents run in parallel against live GA/GSC telemetry. Scope was site-wide, not AdSense-targeted. The findings were uncomfortable.
Fabricated trust signals. The site was carrying authority markers without a verifiable basis. The pattern is familiar on sites that scaffold quickly and audit late: a claim that looks authoritative because it’s round, a signal that implies a credential no one has checked. Each one was stripped. The changes didn’t make the re-review submission notes. They weren’t framed as AdSense remediation because they weren’t — they were quality problems the site had independently of any AdSense process.
Stale rate articles presenting as current. A financial calculator site has specific exposure here. Rate content — mortgage rates, savings rates, inflation figures — written in one economic environment and left without clear dating carries an implicit claim: this reflects conditions now. When the rate has moved, the claim is false. The fix was honest dating, not removal. Publication dates made explicit, update dates accurate. A reader consulting a rate article from six months ago can now tell it’s six months old.
Both fixes shipped before the 24 July submission. Neither showed up as a headline remediation because neither was framed as one. They were the site stopping something it shouldn’t have been doing. That category of fix doesn’t generate a compelling changelog entry.
The diagram problem
The ArticleFigure component that shipped on 2 July added six verified SVG diagrams across the pillar estate. “Verified” meaning checked against the calculator logic each diagram illustrates — not decorative, not generic. A compound interest diagram checked against the compound interest formula. A mortgage amortisation chart that reflects the amortisation model the tool actually uses. A diagram that contradicts its calculator is a different kind of low-value content; the check wasn’t optional.
Six of thirty-eight is not a complete answer. It moved the format-diversity score from zero to nonzero. That closed the literal zero-state the pre-submission audit flagged. Thirty-two pillar pages remain text-only. That’s the backlog now — with a component and a verification pattern in place, not a question of how to approach it.
The question is sequencing. Six diagrams didn’t prevent three rejections. Format diversity is one criterion. Whether it’s the binding constraint or a contributing factor is the kind of thing the fourth review will help answer.
What the QRG update changed
The 1 August root-cause review surfaced the relevant detail from the September 2025 QRG revision: the bar isn’t “technically correct content.” It’s “content a reviewer would judge useful to a real user with that specific search intent.” That framing puts the integrity sweep in a different light.
Stale rate articles without honest dates aren’t useful to someone comparing mortgage options today. They’re misleading. Trust signals that can’t be verified don’t build trust for any reader who checks. Under the current QRG reading, these aren’t marginal quality concerns — they’re directly in scope for what a reviewer evaluates.
The AdSense reviewer didn’t need to name fabricated trust signals or undated rate content in the verdict. “Low value content” named it well enough. The three-lens assessment confirmed what the rejection was pointing at.
Before submission four
The fourth submission hasn’t gone in yet. The changes since submission one:
- Hub pages: live listings behind every navigable URL
- Fabricated trust signals: stripped
- Rate articles: publication and update dates accurate
- Format diversity: six verified SVG diagrams, ArticleFigure component in place for the remaining thirty-two
The 1 August root-cause review gives the fourth submission a more precise supporting case than any of the previous ones had — grounded in the September 2025 QRG state rather than the earlier reading the pre-submission checklist was written against.
Three rejections isn’t a good outcome. The remediation work that followed each one was legitimate, not gaming a checklist. The fabricated trust signals should have come out before the site was submitted to anyone. The stale rate articles should have been dated honestly from publication. The diagram gap was real. AdSense named all three accurately; the Q.0 sweep confirmed them independently.
The site going into review four is different in measurable ways. Whether the gap between what it is now and what Google considers sufficient has closed is Google’s assessment to make. The work done in the process was worth doing regardless of that outcome.
