Process

From decaying pages to a measured refresh

Built for agencies that already know which pages are slipping — and don't have writer time for the long tail.

Intake

Send URLs or a Search Console CSV (page, clicks, impressions, position over ≥56 days). We never need password access; a read-only export is ideal.

Decay queue

We score pages by estimated recoverable clicks (volume × decline × refreshability). You see the ranked list before we write anything.

Treat / control split

Matched pages are held as untouched controls. Without that, “it went up” is just a story.

Brief → draft → check

An AI drafting agent writes the refresh. A separate checker agent runs check_claims.py: unsourced numbers, outdated product names, guarantee language, broken links. FAIL → fix or don't deliver.

Your editor

You review the draft and checklist, edit as needed, publish under your brand. No person at DecayQueue reviews drafts — your editor is the quality gate for publication.

Measurement

At day 28 and 56 we compare treated vs control click change from the GSC data you share. We report the number either way.

What the checker catches

Unsourced numbers

New stats need a working source link in the same paragraph, or they're flagged.

Outdated terms

e.g. AdWords → Google Ads, Google+ shutdown, Mobile-Friendly Test retired — cited from Google.

Guarantee language

“Guaranteed #1 rankings” and similar claims are blocked before delivery.

Broken links

Outbound URLs that return 404/410 get flagged.

The checker verifies provenance, not absolute truth. Items it can't see still need your editor's judgment. See a real (internal) checklist on the sample report page.

Diagrams

Decay queue Pages ranked by estimated click recovery, some marked treat and some control. Sample decay queue · ranked by estimated click recovery /guides/local-seo · −42% clicks · TREAT /blog/gmb-checklist · −31% · TREAT /blog/schema-101 · −28% · CONTROL /resources/adwords · −19% · TREAT Illustrative data only — not a client result.
Control-group measurement Two groups of pages: refreshed and left untouched, compared at day 28 and 56. Measurement: treated vs untouched controls Treated (refreshed) Day 0 → 28 → 56 clicks Controls (untouched) Same period, no refresh We report the difference. We don't promise rankings or traffic.

Start the free pilot