Six Google clicks in 8 landed days against a bar of 25 in 28: 5 on one post from August 11, 1 on a July post, and 0 impressions on the 21 posts published since August 26

The bar, written down before the window opened

On 2026-09-23 this company reads a four-leg checkpoint on the window 2026-08-26 to 2026-09-22. Leg C is the organic-search leg: at least 25 Google web-search clicks on the property inside the window, from an in-window Search Console pull, with no basis change after 2026-09-16. The other legs are a settled payment, real external checks started, and AI-engine referrals; they are reported elsewhere. This post is the leg C reading at day 8, published because the pace is a fact worth having on the record before the read, not after it.

6 clicks in 8 landed days

Search Console's last two days are marked fresh by Google and can still move, so the pace is computed on the 8 days through 2026-09-02 and the fresh days are shown but not counted. The prior 28 days (2026-07-29 to 2026-08-25) carried 13 clicks on 2800 impressions, 0.464 a day.

Which pages earned them

Every click sits on a /news post, and every clicked post predates the window. 5 of the 6 are on one post about reading Search Console's generative-AI data through the API, published 2026-08-11, at an average position of 5.9 on 69 impressions in the window. The other click is on a post from 2026-07-28 at position 42.6. The 904 impressions on the site's comparison, guide, data and vertical pages produced 0 clicks.

Cross-check: our own analytics records fresh landings on /news whose referrer is Google web search, on a viewer-filtered basis, bucketed in Pacific time to line up with Search Console's days. On the 5 clicked (page, day) cells, 3 agree, 1 shows a Search Console click with no landing on our side (an analytics blocker is the ordinary cause), and 1 shows more landings than clicks (2 landings against 1 click on 2026-08-28). Across the window our analytics counts 6 Google landings on /news against Search Console's 6 clicks. The two channels do not disagree about the size of leg C.

The 21 posts published inside the window: 0 impressions

We published 21 posts between 2026-08-26 and 2026-09-04. None of them has a Search Console row in the window: 0 impressions, 0 clicks. The registry's 127 posts by publish cohort, against their window impressions:

The pattern is by age, not by count. Posts from 2026-08-01 to 08-13 carry 5 of the 6 clicks on 184 impressions; posts from 2026-08-14 to 08-25 carry 245 impressions and 0 clicks; posts from inside the window carry nothing. Indexation is the reason we can name. In the daily read of the 14 posts our rebuilt /news index links, day 8 reads 1 of 14 indexed (10 "Discovered", 3 "unknown to Google", 2 Googlebot rows on the 14 in 7 days), the reading published yesterday. The 8 posts published 2026-09-01 to 09-04 were inspected in the same run today:

Page one is not the constraint on the posts that are indexed

35 /news posts average a page-one position in the window. Together they hold 330 impressions and 5 clicks, and 5 of the 5 are the one post above. The next-largest page-one post, on the 2026-08-13 generative-AI impressions logging bug, sits at position 7.7 on 99 impressions with 0 clicks. This is the reading our 2026-08-22 post and the 2026-09-01 series made on the wider set: page-one impressions on this site convert to clicks on one URL and almost nowhere else.

What this means if you care about AI visibility

• A publishing cadence and a click count are different clocks. 21 posts in 10 days added 0 impressions inside the same window; the 6 clicks came from posts 29 and 15 days old at the window's open. If a client's organic bar is measured on a 28-day window, the pages that can move it are the ones indexed before the window, not the ones shipped during it.
• Read the pace on landed days only. Search Console marks its last two days fresh; 2 fresh days here read 0 and 0 clicks and are excluded from the 0.75 a day. A pace read that includes them moves twice before it settles.
• Google clicks are one channel. The same site was cited by ChatGPT, Gemini and Claude enough to send 7 net referrals in 8 days in the 2026-09-02 read, on pages Google has not indexed. Whether an engine names your business when a buyer asks is what our free 60-second check reads per engine.

Limits

• One property, 8 landed days, 6 clicks. Single clicks move the per-day figure by 0.125; the reading is a pace, not a rate.
• Search Console's fresh data can be revised for about two days; the 2026-09-03 and 09-04 rows will be re-read on 09-23 and may differ from the table here.
• Impressions are per query and per position, so a page with 0 impressions has not appeared in any result Google served, indexed or not. The registry's publish date is a proxy for first indexation; Google's discovery date is not visible to us.
• The PostHog cross-check is viewer-filtered and misses readers behind analytics blockers; it is a lower bound on landings, which is why clicks above landings is the ordinary direction.
• Search Console's day boundary is Pacific time; the ledger and PostHog cuts are bucketed to match, but a click near midnight can sit on either side.

The next readings, pre-committed

2026-09-11: the day-15 read of the 14-URL /news index denominator, bar 7 of 14 indexed, published either way. 2026-09-16: the checkpoint basis freezes; no change to what counts as a click after that date. 2026-09-23: the in-window Search Console pull for 2026-08-26 to 2026-09-22. If it reads 25 or more, leg C is met and this post's pace figure was wrong in the useful direction; if it reads under 25, the leg is not met and the sentence "the clicks came from posts indexed before the window" is restated with the final per-page table. Two lines are falsifiable before then: if any of the 21 in-window posts earns a Search Console impression by 09-23, it is named with its first-impression date; if none does, the sentence "21 posts, 0 impressions" is restated at 28 days.