A New-Page Indexing Cohort Tracker
Group new pages by publish week and track four numbers per cohort: how many were published, how many are indexed, how many have received at least one impression, and days since publish. Reading it per cohort rather than per URL is what makes the pattern visible — one page not indexed is noise, but a cohort where three of twelve are indexed after four weeks while the previous cohort reached eleven of twelve tells you something changed between them. Usually the difference is inbound links: pages reachable only from a hub or a sitemap get indexed far less reliably than pages linked from existing indexed pages.
Indexing is not deterministic and single-URL checks invite over-reaction. A page missing after a week means nothing on its own. A cohort behaving differently from the cohort before it means something, and the difference between those two readings is entirely in how you grouped the data.
The tracker
| Column | Source | Why it is there |
|---|---|---|
| Publish week | Your own records | The cohort key |
| Pages published | Your own records | The denominator |
| Indexed | URL Inspection or the Page indexing report | The headline number |
| Received an impression | Performance report filtered to the URL, no query filter | Page totals include anonymised queries that query rows omit |
| Days since publish | Computed | Prevents comparing a 3-day cohort to a 6-week one |
| Inbound internal links | Your own site | The usual explanation for a stalled cohort |
Indexed but never seen
The fourth column matters more than it looks. A page can be indexed and receive no impressions for weeks, which is a different problem from not being indexed — it means the page exists in the index and is not competitive for anything. Conflating the two sends you to fix crawlability when the issue is that nothing on the page targets a query anyone issues. Measure it with page-level totals, not the query list: filter the Performance report to the URL with no query filter, same date range and search type. A new page receiving only rare queries can have real impressions that never appear as query rows, because Search Console withholds anonymised queries from the query breakdown — reading the query list would label a seen page as unseen and send you to the wrong fix.
What a stalled cohort usually means
In our experience the most common cause is that the cohort was published into a hub page and nowhere else. Pages reachable only from a listing page or a sitemap are discovered but deprioritised. Cohorts whose pages are also linked from existing indexed pages — in the body, not just a related list — index noticeably more reliably. Verify that on your own site before accepting it; the pattern is consistent enough for us to plan around and it is still an observation, not a law.
How long to wait
Compare cohorts at equal age rather than at a fixed calendar date. A four-week-old cohort against another four-week-old cohort is a fair comparison; this week's against last month's is not, and the days-since-publish column exists to stop exactly that mistake.
When to act
When a cohort is materially behind its predecessors at the same age. Then check the cheap explanations in order: were they linked from anywhere other than the hub, did the sitemap include them, did a template change add a noindex, did they ship with near-identical titles. Requesting indexing manually is the last resort and treats the symptom.
FAQ
How long should indexing take? — Variable, and that variability is why cohorts beat single URLs. Your own historical cohorts are the only reliable benchmark for your site.
Does requesting indexing help? — Sometimes, for individual URLs. It does not fix whatever made the cohort stall, so it is worth doing after the diagnosis rather than instead of it.
What counts as a stalled cohort? — Materially behind previous cohorts at the same age. Define 'materially' from your own spread before you need it, the same way you would for any other threshold.
Put this to work on your own website.
Fenn finds what your customers ask, drafts the articles and site fixes, and measures what ChatGPT, Claude, Gemini, Perplexity and Grok say about you — with every change waiting for your approval.