Highlights
- The dataset: Over 130,000 AI source citations underpin this analysis, produced by prompts specific to the telehealth industry. The AI responses from four AI models span 12 countries over the course of Q2 2026.
- Reddit makes up 9% of ChatGPT's citations: Its single most-cited source by more than 4x over anything else.
- getmaple.ca is the only universal source: The Canadian telehealth service Maple is the sole domain in all four models' top 20.
- ~84% local-language rate in the Netherlands and Sweden: But only ~38% in Spain.
- 3% overlap for the most divergent pair: Microsoft Copilot and ChatGPT share exactly one top-20 domain, against an 8% average across all pairs.
- 70% of Microsoft Copilot's citations are from commercial domains: Versus 56% for Grok, the widest content-mix gap in the set.
Which sources should you target to get cited as a telehealth brand?
The top-cited domains show how differently the four models source their answers. Each model's five most-cited sources, with their share of that model's total citations:
| Model | Top cited sources | Citation character |
|---|---|---|
| Grok | reddit.com (3.2%), instagram.com (1.4%), sermo.com (1.3%), facebook.com (1.1%), doctorsa.com (1.0%) | Community + review platforms |
| Google AI Overview | getmaple.ca (1.8%), instagram.com (1.7%), google.com (1.5%), ogaei.ca (1.2%), instantconsult.com.au (1.2%) | Provider sites + platform pages |
| Microsoft Copilot | ehalsasverige.se (1.4%), getmaple.ca (0.9%), doctolib.fr (0.9%), ipsosante.fr (0.9%), telmdcare.com (0.8%) | National telehealth providers |
| ChatGPT | reddit.com (8.5%), healthline.com (1.9%), svd.se (1.2%), teldoc.com.au (1.1%), mediquo.com (1.1%) | Reddit + health editorial |
ChatGPT is the most concentrated at the top, Reddit alone is 8.5% of its citations, more than four times its next source, but past Reddit it spreads thinly across telehealth apps and health-editorial sites. Microsoft Copilot is the opposite everywhere: its single most-cited domain is just 1.4%, with citations diffused across hundreds of national provider sites. Grok and Google AI Overview sit in between, anchored by community platforms and provider homepages respectively.
How have AI source rankings changed over time in the telehealth industry?
We split the roughly 16-week window at its midpoint (early May 2026) and ranked the top domains by their share of citations within each half. We read the movement as suggestive rather than conclusive.
The pattern is stark: nearly every global community and review aggregator that anchored the first half, the physician network Sermo, Teladoc, Yelp, Trustpilot, play.google.com, apps.apple.com, fell out of the top 25 in the second half, while their slots filled with national provider sites (midlandhealth.co.uk, medmate.com.au, mediquo.com, doktor.se) and google.com surging to #2. Reddit held #1 in both halves. We observe a localizing, provider-direct shift in the citation mix; we do not claim the underlying models re-ranked their preferences. Track the current picture in the live AI visibility rankings.
What type of content do AI models cite for online healthcare providers?
We classified every cited domain by category. Commercial sites, telehealth-provider homepages and for-profit directories, dominate the category overall, which is unsurprising for a service industry where the "product" is the provider's own booking page. But the mix differs sharply by model.
Every model here is majority-commercial, a structural feature of a category where the entities being recommended are themselves businesses. The insight is in the margins. Grok is the most community-driven: user-generated content is 14.8% of its citations, roughly triple Microsoft Copilot's 2.5%, reflecting its heavy use of Reddit, Instagram, and the physician network Sermo. ChatGPT leans hardest into editorial (9.4%) and also runs UGC-heavy (11.1%), pairing Reddit with health-publisher content like Healthline. Microsoft Copilot is the most purely commercial (70.4%) with almost no UGC.
Do AI models cite local-language content for online healthcare providers?
Using the source's country-code top-level domain (ccTLD) as a proxy for local-language content, we measured the share of citations pointing to a local-ccTLD source in each non-English market. Spanish-language markets were scored against their own national ccTLD: .ar for Argentina, .mx for Mexico, .es for Spain.
Across all non-English markets, 63.2% of citations point to a local-ccTLD source. But the average hides a sharp split: Germanic and Northern-European markets cluster high (65–84%), while the two largest Spanish-speaking markets, Mexico and Spain, sit near 40–49%. The likeliest explanation is structural, not behavioral, the Netherlands (.nl), Sweden (.se), and Italy (.it) have dense national telehealth directories on their local ccTLD, whereas Spanish-language providers frequently publish on .com domains. The ccTLD proxy therefore undercounts genuinely local content in markets that favor generic TLDs, so treat the Spanish figures as a floor, not a true localization rate.
Which AI model relies most on local sources for online healthcare providers?
Holding the market constant, the models differ in how local they go. The heatmap shows each model's local-ccTLD citation rate by language; all cells meet the sample threshold, and Google AI Overview produced too few French citations to report.
Microsoft Copilot and Google AI Overview are the most aggressively local in markets with strong national telehealth directories, both exceed 94% in Dutch and 96% in Swedish, consistent with Microsoft Copilot's overall reliance on national provider sites and Google AI Overview's on provider homepages. Grok is the most globalized, dipping to 40% in Spanish where it reaches for international community and review platforms. ChatGPT lands in a wide band (53–89%), pulled down in Spanish, French, and German by its heavy use of Reddit and other .com sources our method counts as non-local everywhere. The country drives localization more than the model does, but within a country Microsoft Copilot and Google AI Overview reach for local providers while Grok reaches for global platforms.
Should online healthcare providers optimize for each AI model separately?
Almost entirely, yes. We built each model's ranked list of its top-20 most-cited domains and measured how much each pair holds in common, shared domains divided by all the domains the two lists name between them, for all six model pairs. The average overlap is 8.3%, ranging from 2.6% to 17.6%.
No pair shares more than six of twenty domains, and the most divergent pair, Microsoft Copilot and ChatGPT, shares exactly one (getmaple.ca). Across all four lists, that Canadian telehealth service is the only domain every model cites in its top 20; it is, functionally, the single point of agreement in an otherwise fractured field. Every overlap is low in absolute terms, so a brand cannot treat "AI visibility" as one target.
How many sources does each AI model cite per answer?
The models differ enormously in how many sources they cite per answer. Grok is by far the most source-hungry; ChatGPT the most economical.
Grok cites roughly 4.7x as many sources per response as ChatGPT. For a provider, that cuts two ways: Grok offers far more citation slots to compete for, but each individual citation carries proportionally less weight; ChatGPT and Microsoft Copilot cite few sources, so earning one of their handful is both harder and more valuable. These averages count only responses that cited at least one source. ChatGPT produced 3,928 responses but cited sources in only 1,167 of them, most ChatGPT answers cited nothing at all.
Context
This analysis draws from Temso's AI visibility monitoring platform, which tracks how brands appear in AI model responses across ChatGPT, Microsoft Copilot, xAI's Grok, and Google AI Overview. The dataset covers online-healthcare-provider (telehealth) prompts across 12 markets (Argentina, Australia, Canada, France, Germany, Italy, Mexico, the Netherlands, Spain, Sweden, the United Kingdom, and the United States) and seven languages, between March 10 and June 29, 2026.
It measures which sources AI models cite, not which telehealth services or treatments are best. A high citation share means a domain is frequently surfaced by a model, not that it is authoritative, accurate, or clinically sound. Citation behavior is a moving target: models update their retrieval and ranking continuously, so a snapshot like this captures a period, not a permanent state.
Methodology
How we measured this
We analyzed every AI response and its cited sources for the online-healthcare-provider category over the observation window: 13,363 responses and over 130,000 cited source citations. A "citation" is a web source a model explicitly surfaced in support of an answer; we count only sources flagged as cited. Cross-model overlap was measured on each model pair's top-20 most-cited domains (shared domains over the union), then averaged. Domain categories (Commercial, UGC, Editorial, Institutional, Reference, Other) were assigned from a global domain registry; 7–12% of citations per model map to uncategorized domains, reported as "Unclassified." Localization is measured by a country-code-domain (ccTLD) proxy, joining each cited source back to its response's market and language, and excludes English-language markets, which sit overwhelmingly on generic domains, so those figures are directional and conservative. Temporal analysis split the observation period at its midpoint (2026-05-04) and compared domain rankings by within-half citation share.
Model coverage varied across the observation window, so temporal movements are reported as relative ranks within each half. All per-model and per-language cells reported here meet the n≈30 minimum.
Frequently asked questions
Do different AI models cite the same sources for telehealth questions?
No, barely. Any two models share only about 8.3% of their top-20 most-cited domains on average, so roughly 92% of the sources one model relies on are absent from another's top tier. The only domain that appears in all four models' top 20 is the Canadian telehealth service Maple (getmaple.ca). There is no single set of "AI-trusted" telehealth sources; there are several distinct ones.
Which sources does each AI model favor for online healthcare providers?
Grok favors global community and review platforms, Reddit, Instagram, the physician network Sermo, and Facebook. ChatGPT is anchored by Reddit, which is 8.5% of all its citations, plus health-editorial sites like Healthline. Microsoft Copilot funnels into national telehealth providers such as ehalsasverige.se, doctolib.fr, and ipsosante.fr, spread thinly across hundreds of sites. Google AI Overview leans on provider homepages, with Maple its single most-cited source.
Does local language matter for telehealth AI visibility?
Yes, but the market matters more than the model. In the Netherlands and Sweden, roughly 84% of citations go to local-ccTLD sources; in Spain and Mexico, only about 38–49% do. Part of that gap is that Spanish-language providers often publish on .com domains, which our method counts as non-local, so those figures are conservative.
Which AI model is most likely to cite a local provider's website?
Microsoft Copilot and Google AI Overview. Both exceed 94% local-ccTLD rates in Dutch and Swedish markets, and Microsoft Copilot's citations are the most heavily commercial (provider and directory sites) overall. If your priority is being cited as a local telehealth provider, those two systems are the most receptive.
Which AI model cites the most sources, and does that help me?
Grok, by a wide margin, about 27.7 cited sources per response, roughly 4.7 times ChatGPT's 5.9. More sources mean more slots to compete for in Grok, but each carries less weight. ChatGPT's and Microsoft Copilot's economy of citations, and the fact that most ChatGPT answers cite nothing at all, makes each of their few slots both harder to win and more valuable.
Is being on Reddit worth it for an online healthcare provider?
For ChatGPT and Grok, meaningfully so. Reddit is ChatGPT's single most-cited domain at 8.5% of all its citations, more than four times its next source, and it is also Grok's top source. User-generated discussion is a real path to AI visibility, but mainly for the models that lean on it; Microsoft Copilot barely cites it, at 2.5% of its citations.
Did the sources AI models cite change over the period?
The citation mix shifted toward national provider sites and Google's own surfaces in the second half, as global community and review platforms like Sermo, Teladoc, Yelp, and Trustpilot dropped out of the top tier. We read it as a suggestive shift in the citation mix, not a proven trend.
What should an online healthcare provider do with this?
Optimize per model, not in general. There is no universal AI-visibility strategy here: presence on national provider directories helps with Microsoft Copilot and Google AI Overview, community and review platforms help with Grok, and a strong Reddit footprint plus health-editorial coverage helps with ChatGPT. Treat each model as a separate channel with its own preferred sources.
