Indexation report

https://coinliqui.com · started 2026-09-28 07:02 UTC

A. Coverage

What exists, and whether a crawler can still fetch it. No credentials — this section always runs.

What the covered set did since the last reading

Against 2026-W39 — 143 URLs then, 143 now.

Joined (6): /funding/bch, /funding/kbonk, /funding/pengu, /liquidations/bch, /liquidations/kbonk, /liquidations/pengu

Left (6): /funding/fil, /funding/morpho, /funding/strk, /liquidations/fil, /liquidations/morpho, /liquidations/strk

A URL that joined since the last reading has not had time to be indexed, and will appear below as Discovered — currently not indexed for reasons that are not about the page.

B. Search Console

Indexed share, per template

The 6 URLs Google has not indexed, with its own reason for each. Read the reason before acting: Discovered — currently not indexed is a queue, and the answer is usually to wait and watch the indexed-share series below this table; Crawled — currently not indexed is a judgement about the page, and the answer is to change the page.

Indexed share, reading by reading

One row per completed inspection, newest first: 8 readings. The store keeps the last 26.

Search performance, 2026-09-19 to 2026-09-26

These averages are not comparable to the previous reading. 6 URLs joined the covered set since it, taking the total from 143 to 143. A template that starts appearing for more queries appears for the deepest ones first, so an impression-weighted average falls even when no existing query lost a place. The band table below counts queries rather than weighting them, and is the half of this section that survives a change in the covered set.

Queries by impressions

15 of 26 queries Search Console named from 2026-09-19 to 2026-09-26, most impressions first. Equal impressions go to more clicks, then the better average position, then alphabetical order. The cut falls inside a tie: 15 queries had 1 impression, and 4 of them are listed. The band table below counts every one of them.

Where the queries sit

The 26 queries Search Console named from 2026-09-19 to 2026-09-26, by the position each averaged: 61 impressions, 0 clicks. That is not every search. The page total above is 300 impressions and 1 click: Search Console leaves out queries too rare to name without identifying who searched, and the page table counts a search once for each page it showed. Impressions say how often Google showed the page; clicks say how often that mattered.

One push away

No query-page pair Search Console named from 2026-09-19 to 2026-09-26 averaged a position between 11 and 25. Nothing here is close enough that writing more of the same page would move it — see the band table above for where the named queries actually are.

C. Crawler fetches

Not collected here: CF_ANALYTICS_TOKEN is not set.

By decision, not omission. A long-lived analytics credential inside a worker that runs unattended every five minutes is a key carried for no good reason, so the zone analytics are read from the operator's machine instead — npm run crawlers, using the wrangler OAuth token already there. Nothing was handed to this worker.

What that costs: this section would have kept a weekly series, and the local reader samples a single day — the free plan refuses a wider range. The first sample, taken 2026-08-31, is why it matters at all: ~819 VERIFIED AI-side fetches a day against Googlebot's 19, and 704 requests wearing a crawler's name that Cloudflare could not verify. None of it reaches Search Console, GA4 or the pageview counter, because a crawler runs no JavaScript.

D. Legal reading age

All 3 documents were last read on the same day, 40 days before this report was generated. 1 document is overdue.

Open each in a browser and update src/data/terms-baseline.json. A plain fetch is not a reading: both documents that matter here served a shell or a 403 to one, and an automated watcher would have reported no change indefinitely.

What to read first

· Section A must be all green. A URL a crawler cannot fetch is not an indexing problem.

· Indexed share by template, not by page. One template stuck in Discovered — currently not indexed past week 6 is a thin-template problem; scattered pages are just latency.

· Average position is only comparable when the covered set is. The guidance here used to be "position before impressions — it moves earlier and more honestly", and the week of 31 August 2026 refuted it: 54 URLs joined, every template's average collapsed — /liquidations from 40.2 to 263.5 — and not one existing query had lost a place. A template that starts appearing for more queries appears for the deepest ones first. Read the BAND table instead when URLs joined: it counts queries rather than weighting them by impressions, so a query moving from 51+ into 11-25 is a real move either way.

· Crawler fetches are the leading indicator. If they are zero, nothing downstream can move, and the cause is access rather than quality.

· "One push away" is the only section that suggests an action. Everything else here measures what was served or received. A query at position 11–25 is a page that is already relevant and losing to something beatable; for a domain with no external links that band is the only one on-page work can move. An empty section means the honest answer this week is to keep writing and wait — read the band table above it for why.

· Section D is the one nothing else can catch. Every other failure on this site has a technical symptom. A change to the terms this site depends on has none — the pages keep rendering perfectly — so the only detector is somebody re-reading the document.