Experiments · 01
Index lag: how fast a new page reaches Google, Bing & Brave Running · day 11 of 28
Every AI citation starts with being indexed. So before testing citations, we published 16 brand-new pages and tracked, day by day, when each search index picked them up — and whether an IndexNow ping gets pages into Bing faster than the sitemap alone. This is an interim readout; the experiment runs to day 28.
A new site can be citable by Google's AI in days — and invisible to ChatGPT for weeks.
Google indexed 10 of 16 new pages in a median of 3 days. Bing crawled zero — not even the homepage — despite verification and an IndexNow ping. Because ChatGPT reads from Bing, the site simply isn't there for ChatGPT yet.
Read the full write-up: the debug, the tracker bug, and the open question →
Cite this dataset
HZ Signal (Haobo Zhang), Index lag: how fast a new site reaches Google, Bing & Brave — Experiment 01, hzsignal.com/experiments/index-lag/, 2026. Interim readout (day 11): Google indexed 10/16 new pages, median 3 days; Bing crawled 0/16 in two weeks. CC BY 4.0.
How Google indexation built up
Cumulative pages indexed by Google (authoritative Search Console status). Bing & Brave are left off deliberately: both are a confirmed zero, and plotting a flat zero line beside Google would blur “indexed” with “never even crawled.” Their status is in the panel below.
CrawledPages=0), homepage included. Not a block: robots.txt is open
and IndexNow was sent — Bing just hasn't crawled a new, low-authority domain. So H1 has
nothing to compare yet.site:
checks undercount — directional only, but consistent with Bing: the site hasn't reached
it either.Per-page detail — Google (the only engine with trustworthy per-page data)
| Page | Grp | Days to Google index | |
|---|---|---|---|
| GPTBot | B | 0* | |
| AI share of voice | B | 0* | |
| Answer engine | B | 1 | |
| IndexNow | A | 2 | |
| GEO (generative engine optimization) | A | 3 | |
| Google AI Overviews | A | 3 | |
| Google AI Mode | B | 3 | |
| RAG (retrieval-augmented generation) | B | 3 | |
| Canonical URL | A | 3 | |
| AEO (answer engine optimization) | B | 4 | |
| llms.txt | B | ▢ not indexed — “unknown to Google” | |
| OAI-SearchBot | A | ▢ not indexed | |
| ClaudeBot | A | ▢ not indexed | |
| PerplexityBot | B | ▢ not indexed | |
| Schema markup | A | ▢ not indexed | |
| Crawl budget | A | ▢ not indexed | |
* Indexed on the first check day — almost certainly crawled shortly before the formal launch, so treat “0” as a floor. Grp A = submitted via IndexNow, B = sitemap only (the IndexNow test is measured on Bing, not Google).
What we can say
- Google is fast but partial: median 3 days, yet 38% still aren't indexed after 11 days.
- Bing never crawled the site — 0 of 16, homepage included (Webmaster confirms). IndexNow ≠ a crawl for a new domain.
- So the site is citable by Google's AI but invisible to ChatGPT, which reads Bing.
What we can't say yet
- H1 (IndexNow vs sitemap) — Bing crawled neither group, so there's nothing to compare yet.
- Whether the 6 stalled Google pages ever index.
- Group A indexed slower on Google than B — noise (IndexNow doesn't touch Google).
What's next
- Keep the fixed daily checker running — it'll catch Bing's first crawl (and answer H1 via discovery dates).
- Treat authority / off-site links as the real unlock for Bing — not more submissions.
- Upgrade to a final verdict at day 28.
Pre-registered hypotheses — status
The 16 pages under test
Each unit is a genuinely useful AI-SEO glossary term, published in a single deploy and left up as a reference. Group A was pinged to Bing via IndexNow; Group B was left to the sitemap.
Group A · IndexNow 8
Interim readout · n = 16 pages, single new site — directional, not definitive ·
Google = Search Console “indexed” verdict · Bing = Webmaster crawl status (checker fixed
2 Jul; GetCrawlStats confirms 0 crawled) · Brave = site: best-effort ·
Data 21 Jun – 2 Jul 2026 · Pre-registered Jun 20, 2026.