Thirteen of sixteen pages indexed overnight, no backlinks: the indexing bundle has a problem

A new site with sixteen pages, a sitemap, no backlinks, no audience and no history went from every page unknown to Google on its first afternoon to thirteen of sixteen indexed by the next morning, according to a log published on dev.to ^[1][2][3][4]. Nobody paid for that, and the author is explicit that nothing clever caused it ^[5], which is awkward for a product category built on submitting your pages to Google ^[6].

The log exists because Search Console answers only for today and retains nothing, so a site's day-one state is normally gone by day two ^[7]. That is the mundane discipline here: something wrote it down.

Partway through the first afternoon the author used Request Indexing on nine of the pages, then stopped when he got distracted, leaving the rest untouched ^[8]. He declines the easy conclusion in either direction. The split he reports is nine out of nine indexed among the requested pages against three out of six among the ones nobody asked about ^[9], and he lists the reasons that number cannot carry weight: the nine were not random but worked through in order of what he cared about, the groups are tiny, it is one site on one day, and every page sat in the same sitemap and the same navigation ^[10]. His stated position is that nobody running a sample this size can say whether Request Indexing helped ^[11].

The published figures also do not reconcile. Nine requested pages out of sixteen leaves seven unrequested, and thirteen indexed minus nine requested leaves four, so the reported group of six with three indexed does not match the headline counts ^[12]. Treat the group-level split as unusable and the top-line result as the finding.

The three stragglers are the useful detail. Two were tool pages edited heavily that same afternoon, and by morning Google still had not heard of them; the third was a terms page, which Google had discovered and declined to index yet ^[13]. Indexing arrives page by page, and the laggards were legal boilerplate and pages that changed after first discovery ^[14].

Now the products. Google's Indexing API documentation supports two kinds of content, job postings and live-stream events ^[15]. Send it a blog post and it returns a success response, because the response confirms receipt of the message, not action on it ^[16]. The other half of the usual bundle is IndexNow, which the author calls a genuinely good protocol; it notifies Bing, Yandex, Seznam, Naver and Yep, and does not include Google and has never claimed to ^[17]. All thirteen pages here were submitted to IndexNow ^[18], which means the only submission that definitely reached Google was the nine manual ones, and Google indexed pages from both groups anyway ^[19].

The operating advice is unglamorous: have a sitemap, link every page from somewhere a person can reach, connect Search Console on day one, and leave it alone for a week before concluding anything ^[20].

What to watch is the version of this that would settle it: new pages split randomly, half requested, published the same day with the same internal linking, repeated until the groups mean something ^[21]. The author says that test is running now and will take months, and that any article claiming an answer in a week has not got one ^[22]. Anyone wanting their own baseline can read the same per-page state free in URL Inspection, one page at a time, or pull all of it through the Search Console API ^[23].