The readiness audit came back with seven green items and one red. The red one: “Diverse and engaging content in a variety of formats (articles, images and videos).” Score: 0 of 38 articles had any embedded visual content.
Seven items passing feels like near-certainty. It isn’t.
The setup
UK Calculators had crossed the metrics that should make an AdSense application straightforward. 78.6% indexation in Search Console. Eight pillar articles live in four weekdays. Thin surfaces from the initial build resolved. A newly-built adsense-telemetry skill (B-078) was reading live status from the AdSense Management API v2: approval status, no policy strikes, Auto Ads enabled.
The account-level status was READY. The site-level status was GETTING_READY: review in progress, application in the queue.
That was when I ran the detailed checklist audit.
What 7 of 8 looked like
Google publishes a reviewer checklist. Eight items. They cover the ground a human reviewer would assess: content quality, site structure, navigation, original content, author credibility signals, ad placement readiness, ad density, and format diversity.
Seven passed. Content quality cleared because the pillar articles were substantive and accurate. Navigation cleared because the site had working category structure and internal links. Terms of service cleared because the ads.txt file was correctly configured with AdSense listed as an Authorised reseller. That one was never in question.
The one failure: format diversity. The measurement: 0 of 38 articles included any embedded visual content. The site was text-only, end to end. The checklist item asks for “articles, images and videos”: a variety of formats. The site had one.
The fix
An ArticleFigure component: a semantic <figure> element with a <figcaption>, rendering captioned diagrams in article body copy. Then six SVG diagrams, each verified against the content it illustrated and distributed across the pillar articles.
The diagrams weren’t decorative. Each one mapped to a concept the article was already explaining in text: a tax calculation flow, a threshold comparison, a decision tree for an allowance check. The reasoning: if the checklist item asks for formats that help users understand content, the visual content should do actual explanatory work.
Format diversity: closed.
The site was already GETTING_READY when those six diagrams shipped. Whether the diagrams were visible to the reviewer who assessed the application is not something the API surfaces.
The verdict
10 July, 17:42 BST. Rejected. “Needs attention — Low value content.”
The ads.txt file was listed as Authorised. The rejection was a pure content-quality judgement.
I audited the site against Google’s inventory-value policy immediately. The specific language was there. The hub pages, /resources/blog and /resources/guides, were thin: article titles listed with no editorial layer, no curatorial context, nothing that gave a user a reason to be there rather than navigate directly to the pillar content. The policy language for low-value content matched those pages exactly.
The format diversity item had been closed. The hub pages had not been flagged by the checklist at all.
What the score actually said
Seven of eight is 87.5%. The temptation is to read that as “87.5% of the way there.” The checklist presents all eight items as a flat list with no weights attached. One item failed; seven passed; the ratio looks favourable.
A reviewer doesn’t use a checklist. A reviewer answers a single question: would showing an advertiser’s ad on this site embarrass that advertiser? Hub pages with no editorial content answer that badly. An index of calculator titles with no surrounding context is a low-effort surface. It doesn’t belong in premium ad inventory, and a reviewer will see that immediately, without needing a checklist item to name it.
Format diversity was the only visible red flag in the audit. It was real. It was fixed. It was not the thing doing the most work in Google’s evaluation. The hub pages were doing more, and they were invisible to the scoring tool entirely.
The 7/8 score said: seven listed criteria are above threshold. It said nothing about what a human reviewer would weigh, and nothing about the depth of pages the checklist didn’t itemise.
What the checklist doesn’t cover
Eight items. What they don’t cover: the depth of hub pages versus pillar pages. The ratio of index content to editorial content. Whether listing pages add any curatorial value on top of their links.
Trust signal integrity was missing from the list too. After the rejection I ran Phase Q, a site-wide quality sweep. It removed fabricated trust signals and reframed stale rate articles to reflect their actual status. Those were pre-existing issues. No checklist item surfaced them. A reviewer assessing the site before that sweep would have found signals that didn’t hold up.
The gap between what a checklist covers and what a reviewer assesses is where applications fail despite a passing score.
What changed
The hub pages were rebuilt: real editorial listings for /resources/blog and /resources/guides, pages that do actual curatorial work. The Q.0 sweep cleared the trust signals. The format diversity fix was already in before the rejection; the six SVGs remain.
The re-application hasn’t gone in yet. Before it does, the pre-flight audit will include a specific check of pages the checklist doesn’t itemise: hub page depth, index-to-editorial ratios, trust signal integrity. Not because Google publishes those criteria, but because a reviewer will assess them regardless of what the checklist says.
Seven of eight checklist items passing doesn’t mean 7/8 of the review is satisfied. It means seven listed items cleared a threshold. Everything the list doesn’t ask about is still visible to the reviewer.
