Ask ChatGPT for the best product or tool in your category and it returns a confident shortlist. It reads like judgment, but it is not. It is a summary of the sources ChatGPT pulled at answer time, and once you know which sources those are, the question "why does it recommend them and not me" has a concrete answer.
We classified every source ChatGPT cited across the August 2026 Cited Index, 1,337 citations across every category and both consumer and B2B brands. The mix tells you how the engine thinks:
| Source type | Share of ChatGPT's citations |
|---|---|
| Brand websites | 42% |
| Editorial (independent tech press, mainstream news) | 21% |
| Community (Reddit above all) | 19% |
| Review platforms | 7% |
| Marketplaces | 4% |
| Other | 7% |
What matters most in that table is the number that is not there. At 42%, ChatGPT reads brand websites less than any of the other four AI engines. Gemini sits at 72%. ChatGPT leans harder than any of them on what other people say about a brand, and that changes everything about how you win it.
ChatGPT runs on Reddit and independent press
Look at what ChatGPT actually cites off your site. Its single largest non-brand source, by a wide margin, is reddit.com. After that come independent tech publishers like TechRadar, then mainstream outlets like Times of India, then specialist review press. Marketplaces are almost nothing at 4%.
So when ChatGPT builds a shortlist, it is reading community discussion and independent journalism, and repeating the brands that show up there credibly. This holds across both worlds we track, though the balance shifts: for B2B software it leans on brand sites plus tech-press roundups, and for D2C it leans even harder on editorial and community. Either way, ChatGPT is the engine least satisfied by your own marketing copy.
Why your competitor keeps getting cited
Here is the diagnosis most teams need. If ChatGPT recommends a competitor and skips you, it is rarely because their website is better than yours. It is because they are present on the sources ChatGPT reads and you are not.
Run the check on your own category. Take the prompts your buyers type, ask ChatGPT, and look at the citations underneath the answer. You will almost always find the same shape: your competitor is named in a Reddit thread, included in a "best of" roundup on a tech or lifestyle publisher, and has an extractable page of their own. ChatGPT has three sources pointing at them and zero pointing at you. The model is not biased against you. It simply has more evidence for them.
That is a fixable problem, and knowing it is a sourcing gap rather than a product gap is the whole point. You are not competing on who has the slicker homepage. You are competing on who the internet, and Reddit in particular, actually discusses. Those same sources also decide the adjective ChatGPT attaches to you, which is its own positioning problem worth managing.
How to read your own ChatGPT gap
Want to know how your brand scores on these same metrics?
We'll run 20 prompts across 3 AI platforms and send your report within 24 hours.
Three questions, answerable in an afternoon:
- Are you in the subreddits? Search the communities your buyers use for your brand and your competitors. If they are discussed and you are absent, that is your largest single ChatGPT gap, because Reddit is its largest single source.
- Are you in the roundups? Find the "best [category] in India" articles on the tech and lifestyle publishers ChatGPT cites. If competitors are in them and you are not, that is 21% of the citation base you cannot reach.
- Can ChatGPT extract your own pages? Your site is still 42% of the mix. If your pages read as slogans rather than facts, ChatGPT has nothing clean to lift even when it does reach you.
The tactical fixes for each of those live in the persona playbooks. This piece is the diagnosis: ChatGPT recommends the brands its sources discuss, its biggest source is Reddit, and your competitor is winning because they are in the conversation you are not.
What this does and doesn't tell you
These are market-wide shares, not a verdict on any one answer, and Reddit's weight is not spread evenly: a category with an active subreddit gives ChatGPT a lot to pull, a dead one gives it almost nothing, so check your own communities before you treat Reddit as the universal lever. What the data does establish is where ChatGPT looks, not that a single thread flips a recommendation. It is also one of five engines that read very different webs, which is why we mapped all five side by side. Run your own prompts and read the citations before you commit.
None of that changes the shape of the finding: ChatGPT reads Reddit and independent press more than it reads you, and the brands it names are the ones already being discussed where it looks.
Methodology. 1,337 ChatGPT citations in the August 2026 Cited Index, classified by source-domain type and pooled by citation count across 339 brands in 12 categories. Aggregate and market-wide: no individual brand is named, only public cited-source domains and types, and this is a different measurement system from the per-brand Citation Mix in the Cited dashboard. Citation Source Mix is the sixth of the Cited 8 metrics.
Want to see which sources ChatGPT cites for your brand? Get a free AI Visibility Report across ChatGPT, Gemini, and Perplexity.