Glenn Gabe, who runs GSQi and has spent years doing forensic, page-by-page analysis of Google core updates, has been tracking one site since June 22, 2026, under a title that tells you the whole setup: 'Surging in ChatGPT, Dead in Google.' A YMYL property with essentially zero organic visibility in Google search, and a steady, substantial stream of citations inside ChatGPT answers. On August 3, 2026, Gabe posted an update. Bing deindexed the site.
The case Glenn Gabe has been tracking
Gabe's original post, published June 22, 2026, documented something genuinely strange for anyone used to thinking of 'search visibility' as one number. The site in question had no meaningful presence in Google's organic results, the kind of visibility collapse that would normally read as a core-update casualty or a manual-action situation. And yet the same site was showing up, repeatedly, as a cited source inside ChatGPT answers on topics in its niche. Two numbers that are supposed to move together, Google visibility and AI-answer visibility, were pointing in completely opposite directions for the same domain.
The explanation, once you look at the mechanics of how ChatGPT actually retrieves web content, isn't mysterious. ChatGPT doesn't run its own independent crawl-and-rank pipeline the way Google does. For a meaningful share of live web retrieval, it draws on Bing's index and search infrastructure, a well-documented architectural fact of how Microsoft's investment in OpenAI has shaped ChatGPT's plumbing. A site that Google has effectively dropped from its results can still be alive and well in Bing's index, and if it's alive in Bing's index, it can still surface as a ChatGPT citation, completely independent of whatever happened to it in Google.
How ChatGPT citations happen while a site is invisible on Google
This matters beyond one case study because it's a clean illustration of something our own anatomy of an AI citation research has argued for months: getting cited by an AI engine is not the same thing as ranking well in classic search, and the two can diverge sharply depending on which index and which retrieval mechanism a given engine actually leans on. ChatGPT citing you says something about Bing's relationship to your content. Google ranking you says something about Google's relationship to your content. Those are two separate systems, evaluated separately, and a site can pass one test while failing the other completely.
| GOOGLE ORGANIC SEARCH | BING INDEX (FEEDING CHATGPT RETRIEVAL) | |
|---|---|---|
| This site's status, per Gabe's tracking | Effectively zero visibility, since before Jun 22 | Indexed and regularly cited, until Aug 3 |
| What determines inclusion | Google's own crawl, quality, and ranking systems | Bing's own crawl and index inclusion criteria |
| Downstream effect | No organic traffic from Google | Steady ChatGPT citation volume, independent of Google |
| As of Aug 3, 2026 | Unchanged (still near-zero) | Deindexed — citation volume now at risk |
August 3: the deindex
Gabe's August 3 update is short and direct: Bing deindexed the site. He's now tracking the downstream effect on the site's ChatGPT citation volume, the same way he originally tracked the citation volume itself, methodically and over time rather than off a single snapshot. That follow-through matters. A lot of GEO case studies stop at the interesting anomaly (a Google-dead site thriving in ChatGPT) without sticking around long enough to see what happens when the underlying infrastructure that anomaly depended on changes. Gabe's still watching, and the honest answer to 'what happens to ChatGPT citations when Bing deindexes the source' is a live experiment right now, not a settled result.
“"Surging in ChatGPT, Dead in Google" — Glenn Gabe's tracking title for the case, originally published Jun 22, 2026, updated Aug 3, 2026: "Bing deindexed the site."”
What we can say without waiting for Gabe's next update is the structural point, which doesn't depend on exactly how much the citation volume drops. A site whose AI-citation visibility was entirely downstream of a single third-party index has no visibility into that index's health, no warning before a deindexing event, and no recourse once it happens beyond the same reconsideration and technical-fix process any site uses to try to get back into any index. The citation volume wasn't earned independence from Google's judgment. It was borrowed from Bing's, and Bing can revoke it exactly as unilaterally as Google can.
That concentration figure, reported in our own prior analysis of citation concentration, is about Copilot specifically, not ChatGPT, but it's directly relevant context here: Bing's index and Microsoft's Copilot ecosystem already show AI citation volume clustering hard around a small number of mega-authority domains. A mid-sized YMYL site building meaningful citation volume through that same Bing-fed pipeline was already an outlier worth watching before the deindex, not a stable, typical case. Outlier positions are exactly the ones most exposed when the underlying index shifts, because they were never protected by the kind of broad, redundant authority signal that keeps an Amazon or a Wikipedia visible no matter what a single index does.
Why single-index ChatGPT citations are a structural risk
We've made a version of this argument before about relying on any single third-party channel for AI-citation volume, most directly in our coverage of Reddit's citation collapse after Google cut its bulk search-access deal with Reddit. That was a commercial-relationship risk: two companies' business arrangement changed, and every site whose GEO strategy leaned on Reddit's citation volume took the hit as collateral damage. This is the same shape of risk, one layer down the stack. It isn't a business deal that can end. It's a search index a site never had any relationship with in the first place, doing routine index maintenance that happens to erase the specific pages a GEO strategy was quietly depending on.
What to check on your own citation dependencies
The practical question this case study raises for anyone running a serious generative engine optimization program is simple to ask and uncomfortable to answer honestly: how much of your current AI-citation volume can you actually trace back to a specific index, and would you know if that index dropped you tomorrow? Most teams can answer the first half of that question for Google, because Search Console makes Google's view of a site legible. Far fewer have equivalent visibility into Bing's index status, and fewer still are actively monitoring for a deindex event the way Gabe has been doing here as a deliberate research exercise rather than routine operations.
That's a real gap worth closing. Bing Webmaster Tools exists, is free, and gives a direct read on whether Bing's crawler can reach and index your content at all, the same category of first-party signal Search Console provides for Google, just less commonly checked. If any meaningful share of your team's reporting shows AI-citation numbers moving independently of your Google Search Console numbers, that's not necessarily good news framed as GEO success. It can just as easily be a Bing-index dependency nobody has verified is stable, sitting quietly until an event exactly like this one shows up in the data.
None of this argues against citation volume that happens to route through Bing, or for treating Bing's index as untrustworthy. It argues for knowing which index is actually doing the work behind a number you're reporting as a win, the same discipline we've pushed for separating grounding, citation, and mention as distinct trackable events rather than blending everything into one dashboard figure. A citation number with an unexamined single point of failure underneath it isn't a stable asset. It's a liability wearing a good quarter.
We'll be watching for Gabe's follow-up on how far the site's ChatGPT citation volume actually falls, and how long recovery takes if Bing reinstates it. Either outcome is useful data. A sharp, sustained drop confirms the dependency was real and total. A partial or short-lived dip would suggest ChatGPT's retrieval leans on more than Bing's live index alone for a site with an established citation history, which would be its own useful finding about how sticky AI-citation volume can be once it's established. Either way, the case is a live reminder that a metric moving in the right direction isn't the same thing as a metric standing on solid ground.
See where you are cited today
A free snapshot audit of your rankings and AI citations before we ever talk.
Josh leads work at the intersection of SEO and generative engines at Something Inc., helping B2B brands get ranked and cited across every major AI engine.