Google indexed us in 3 days. Bing crawled nothing for two weeks.

The short version: We published 16 new pages on a brand-new domain. Google indexed 11 of them at a median of 3 days — several within a day. Bing crawled 0. Brave found 0 (our daily crawl data). And because ChatGPT’s search leans on Bing while Claude’s leans on Brave, a page Google had indexed within days could still be missing from the indexes that feed ChatGPT and Claude — for the entire two weeks. (This experiment measures indexing, the prerequisite for AI citation; it does not measure citations directly.) Digging in turned up two things worth more than the headline: our own crawl-tracker was silently broken, and Bing had never registered our sitemap despite it sitting in robots.txt with IndexNow pinging away. We changed one thing on 2 July and — as of 4 July — Bing is still at zero. This is a live investigation, not a solved case, and that’s on purpose.

---

What we were actually testing (and what broke)

Experiment 01 was pre-registered with two hypotheses (written before we looked at any data, so we can’t move the goalposts afterward):

• H1 (primary): pages pushed to Bing via IndexNow get indexed faster than pages left to organic crawling. Predicted: IndexNow ~48h, sitemap-only 1–3 weeks.
• H2 (secondary): the same pages enter Google and Bing within days, but Brave lags or misses them — so a page becomes citable by ChatGPT / Google AI before it’s citable by Claude.

Here’s the honest outcome, stated plainly: H1 is untestable from this run. You cannot compare “IndexNow-submitted” against “sitemap-only” when neither group got crawled by Bing at all. The independent variable never engaged. So we’re not going to dress this up as “IndexNow works” or “IndexNow doesn’t” — the experiment that was supposed to answer that question was sitting on top of a broken Bing connection, and the useful finding is the broken connection itself. H2, meanwhile, half-held in a way we didn’t predict — more on that below.

The symptom: one engine sprinted, two never showed up

Across 16 new AI-SEO glossary pages, published together on 21 June:

Engine	Indexed / found	Speed	Reads for which AI?
Google	11 / 16	median 3 days (range 0–12)	Google AI Overviews / AI Mode
Bing	0 / 16	never crawled in 14 days	ChatGPT search, Copilot
Brave	0 / 16	never found in 14 days	Claude’s web search

Two details keep the “Google sprinted” story honest. First, Google didn’t take everything — it skipped 5 of the 16 (llms-txt, oai-searchbot, claudebot, perplexitybot, schema-markup), four of which are short bot/spec-name definitions. Fast for what it wanted, indifferent to the rest. Second, one page (crawl-budget) took 12 days — the tail is long even on the engine that did well.

But the headline is the two zeros. Not “slow.” Nothing. And that matters more than it sounds, because of the next point.

”Indexed” is not one status — it’s per engine

This is the reframe worth taking away even if you remember nothing else. A page is not simply “indexed” or “not.” It is indexed by a specific engine, and different AI assistants read from different indexes:

• ChatGPT’s live search leans on Bing.
• Claude’s web search leans on Brave.
• Google’s AI Overviews / AI Mode read Google.

So on the very same day, one of our pages could be indexed by Google — the prerequisite for feeding a Google AI Overview — and completely absent from the index ChatGPT searches, because Bing had never seen it. (The engine→index mapping above is the working model from our first post, not something this experiment re-proves; what this experiment measures is which index each page actually reached.) A single “is my page indexed?” checkbox hides the thing that actually gates AI visibility. If you care about being cited by ChatGPT specifically, Google’s index telling you “you’re in” is not the answer to your question.

The twist: our own instrument was lying to us

Before blaming Bing, we checked our own tracker — and it’s a good thing we did.

Our crawl-checker called Bing’s GetUrlInfo API and counted a page as “crawled” only if the response came back with HttpStatus == 200. Reasonable-looking assumption. It was wrong.

Bing’s API returns HttpStatus: 0 even for pages it has genuinely crawled and holds in its index. We confirmed this against a control: an established page we already knew Bing had crawled came back with a real last-crawled date and a real document size — and HttpStatus: 0. Our == 200 gate could therefore never be satisfied. It was structurally incapable of ever reporting “crawled,” which means it would have kept logging “not crawled” forever — including on the day Bing finally does crawl us.

This is the entire reason this site exists: a confident metric built on a wrong assumption is worse than no metric at all. A tool that can only ever return “no” feels like data and is actually noise. We only caught it by validating the checker against a page we knew was crawled before trusting a single one of its “no” readings. The fix was small once we saw it — detect a crawl from the presence of a real last-crawled date, not from HttpStatus.

If you’re measuring AI visibility with anything — a homemade script or a paid tracker — test it against a page you know is cited or crawled before you believe its zeros.

The real cause, so far: Bing never registered our sitemap

With the tracker fixed, Bing still showed zero. So we asked what Bing actually knew about the site. In Bing Webmaster Tools, GetFeeds — the list of sitemaps Bing is tracking — came back empty. Our sitemap was declared in robots.txt, and IndexNow had been submitting URLs, but Bing’s Webmaster Tools held no registered feed and had crawled nothing.

Here’s the interesting wrinkle. Bing’s own documentation recommends referencing your sitemap in robots.txt “for automatic discovery,” and pairs sitemaps with IndexNow to give new URLs “the best chance to be discovered, crawled, and indexed efficiently.” For our new, low-authority domain, neither the auto-discovery nor the IndexNow pings produced a single crawl. Bing does add, fairly, that “no tool can guarantee when or how your content will appear” — which is exactly why we measure rather than assume. Documented behaviour and observed behaviour diverged, and that gap is what this site exists to record rather than repeat on faith.

What we changed — and what we’re still watching

On 2 July we explicitly submitted the sitemap through Bing Webmaster Tools and pushed one group of URLs through its URL Submission API.

What we do not yet know is whether that registration is what unblocks the crawl. We registered it on 2 July; as of 4 July — two days on — Bing is still at zero. That’s well within Bing’s normal window for a new site, so it proves nothing yet. This is genuinely open, and we’re going to let the daily data answer it rather than declare victory on the tidy-sounding fix.

Date	What happened
21 Jun 2026	16 glossary pages published in one deploy; sitemap in robots.txt; IndexNow submitting URLs
21 Jun – 4 Jul	Google: 11/16 indexed (~3-day median). Bing: 0 crawled. Brave: 0 found.
2 Jul 2026	Found + fixed the tracker bug; found Bing’s sitemap list empty; registered the sitemap in Webmaster Tools
4 Jul 2026	Two days post-registration — Bing still 0. Within normal range; watching.
~9–16 Jul	(pending) Did Bing start crawling within 1–2 weeks of registration?

That answer — not the fix itself — is the actual finding, and we’ll update this post with it either way.

What you can take from this, even though the story isn’t over

• Validate your instruments before you trust a number — especially a “not indexed” or “not cited” zero. Test the tool against a page you know is live in that engine first.
• “Indexed” is per-engine. Google-fast doesn’t mean Bing-fast, and Bing is the gateway to ChatGPT’s search.
• Don’t assume IndexNow + a robots.txt sitemap guarantees a crawl. Check that Bing Webmaster Tools actually lists your sitemap (the Sitemaps panel, or GetFeeds via the API). Ours wasn’t there, and nothing we’d done had put it there.
• The honest answer to “how do I get Bing to crawl a new site fast” is: we’re still testing it — and we’ll show you the data either way.

What we don’t know yet (on purpose)

We can’t cleanly separate “the registration caused the crawl” from “it would have happened with time anyway” off a single event on a single site — so we’re reporting it as an open question and letting the daily data decide, rather than dressing a guess up as a conclusion. n=1 domain, 16 pages: this is one site’s experience, not a universal law. That distinction is the entire point of the exercise.

FAQ

• Does Bing auto-discover a sitemap from robots.txt? Bing’s docs say yes — but for our new site it produced no crawling, and the sitemap wasn’t showing as registered in Webmaster Tools. We submitted it explicitly and are still confirming whether that’s the fix.
• Why does Bing matter if I already rank on Google? Because ChatGPT’s search reads from Bing. A site Bing hasn’t crawled can be invisible to ChatGPT even while Google can already cite it.
• Did IndexNow work? It submitted the URLs, but Bing crawled none of them in two weeks. For our new domain, IndexNow alone didn’t trigger a crawl — and because Bing indexed nothing at all, we can’t yet say whether IndexNow speeds Bing relative to plain sitemap crawling. That test is on hold until Bing crawls us at all.
• How do I check whether Bing has my sitemap? In Bing Webmaster Tools under Sitemaps (or GetFeeds via the API). An empty list means Bing isn’t tracking it.

Sources

• Bing on sitemaps + AI-powered search / IndexNow: Bing Webmaster Blog
• Bing Webmaster Tools for ChatGPT visibility: Stackmatix
• Our own measurement + data: HZ Experiment 01 (index-lag), tracked daily. Figures are from our site as of 4 July 2026; treat them as one site’s experience, not a universal law.