After 11 Days Frozen, Google Finally Re-read Our Sitemap — and the Indexed Count Went Down

Our indexing coverage number had said the same thing for eleven straight days: 208 pages indexed, 2,270 not, 2,478 known. We kept pushing "request indexing" at every session we ran — no movement. This morning the number finally moved — in the wrong direction.

164 indexed. 3,594 known. The known pool grew by 1,116 pages overnight and the indexed count dropped by 44.

What actually happened. Our sitemap was regenerated on Oct 16 (3,162 URLs). Google, which hadn't re-read the map since early September, finally picked it up — an inspection panel confirmed a fresh Googlebot crawl dated Oct 16 on one of our app pages. A re-read after weeks means re-evaluating everything the map contains, and during re-evaluation the buckets reshuffle: "discovered, not indexed" jumped to 1,891 pages and "crawled, not indexed" to 1,020.

Why the indexed count falling isn't the disaster it looks like. Two things can be true at once: Google now knows about 1,100+ pages it had never seen, and it temporarily moved 44 pages out of the index while it re-decides what they are. The first is progress we waited a month for. The second is the cost of the first.

What we're watching now.

1. Whether the indexed count climbs back over the next one to two weeks as the re-evaluation settles — the 29 pages we've pushed through "request indexing" sit in the priority queue inside exactly this window.
2. The two big buckets (1,891 discovered / 1,020 crawled) are where the next gains come from if they convert; the 506 old 404s we've reported before still sit in their own exclusion bucket, untouched by the reshuffle.
3. Whether the daily quota still matters. It does — 10 more pages requested today — but the re-read did more for discovery in one morning than all twenty-nine pages of our "request indexing" queue combined.

The lesson we're keeping: when a coverage number freezes for days, the interesting change isn't in the number — it's in whether Google has re-read your sitemap at all. Check the map's "last updated" date against the crawl dates in URL inspection before assuming the requests are being ignored. We assumed; the map was just waiting to be picked up.