How do I get my SaaS cited by ChatGPT and Perplexity

Why isn’t my SaaS showing up in AI answers?

Because the engine builds its answer from a handful of sources it already trusts, and your site is probably not one of them. An AI assistant asked “best X for Y” does not rank every vendor in the category; it retrieves a few pages, names three to five products, and stops. Absence is the default state, not a penalty.

We measured how common that absence is. Across 300 brand-answer pairs — 30 companies, 10 answers each — the average brand was named in 46% of the answers about its own category. Five of the 30 companies were invisible: never named and never cited, in any of the ten answers about the category they sell into. A sixth was named zero times and cited once.

None of those five are small: they are funded vendors with sales teams, review-site profiles and years of content. They lost the answer anyway, because nothing the engines read that day described them as a candidate for the question being asked.

How often does an established brand actually get named?

Less than you would expect: 46% on average, and no category rewarded more than three vendors. In every one of the five categories a leader appeared in 8 to 10 of the 10 answers while three or four competitors sat at 0 to 3. There is no comfortable middle — brands are either in the recommendation set or effectively absent from it.

In two of the five categories the gap is wide enough to call winner-take-most. In real estate CRM, Follow Up Boss was named in 10 of 10 answers and the next closest vendor in 6. In staffing ATS, Bullhorn was also named in 10 of 10, with the runner-up at 7 and the median vendor at 3. Those two are not just ahead; they are the answer, and every other vendor is a footnote the buyer may never read.

The useful read for a challenger: the tail is long and cheap to beat. Moving from 0 of 10 to 4 of 10 is a different project from unseating a category leader, and it is the one that changes your pipeline.

Is being mentioned the same as being cited?

No, and the difference decides which fix you need. A mention is the engine naming your product inside the prose. A citation is the engine linking your domain as a source for what it just said. They come apart constantly.

The clearest example in the audit is dental practice management. Dentrix was named in 9 of 10 answers but cited as a source in only 2 — the engines knew the brand from everyone else’s writing and never needed its site. Curve Dental ran the opposite way in the same ten answers: named in 7, cited in 9. Its own pages were doing the explaining, so the model reached for them even when recommending someone else.

A negative gap and a positive gap are different problems. Dentrix does not need more brand awareness; it needs pages the engines can source. Denticon has the opposite job: its material is good enough to quote, and it still gets named in fewer than half the answers.

Which sources do ChatGPT and Perplexity actually cite?

Different ones, which is why a single content plan underperforms on both. Across the 50 answers the two engines returned 706 cited URLs from 313 distinct domains, and the overlap between their favourites is thin.

• ChatGPT leaned on vendor sites: 46% of its 215 cited URLs pointed at one of the six audited vendors’ own domains — pricing pages, product pages, docs — and it cited Reddit zero times.
• Perplexity leaned on everyone else: only 11% of its 491 cited URLs were vendor-owned. Its single most-cited domain was reddit.com, ahead of every vendor and every review site.
• Review sites and niche comparison blogs appear in both, with G2 the most cited of them, alongside a small rotating set of category-specific roundups and a few mainstream tech outlets.

Two practical consequences. Your own marketing pages are worth writing — ChatGPT reads them, and they are the cheapest source you fully control. And they are close to irrelevant on Perplexity, where the citation you need is a Reddit thread, a review profile or a roundup post written by someone who is not you.

What do I actually do about it?

Five things, in this order, because each one makes the next cheaper:

• Ship Organization schema on every page, with the same legal name, description, logo and sameAs profiles everywhere. Entity consistency is what lets a model connect your Reddit mention, your G2 profile and your site to one company instead of three.
• Publish an llms.txt at your root: a plain-text summary of what you sell, to whom, at what price, with links to the pages that prove it. It costs an afternoon and removes any ambiguity about how you should be described.
• Make your pages extractable. Answer the question in the first two sentences under each heading, keep the heading phrased the way a buyer would ask it, put pricing and specs in tables and lists, and stop hiding the substance behind a demo request. Models quote what they can lift cleanly.
• Earn third-party mentions where the engines already look: review-site profiles kept current, category roundups that list you by name, and honest participation in the forums your buyers read. On Perplexity this is nearly the whole game.
• Measure mentions and citations separately, per engine. They are different failures with different fixes, and an average across engines hides both.

Nothing on that list is a growth hack, and none of it moves in a week. What it does is make you a legible candidate for a question you are currently absent from.

How do I know it’s working?

By re-running the same prompts on a schedule and watching two numbers move: the share of answers that name you, and the share that cite your domain. Rankings do not tell you this — a page can rank first in Google and never be sourced by an assistant answering the same question.

Pick 20 to 50 prompts a real buyer would type, run them against each engine on the same day of the month, and log every named vendor and every cited URL. The first run is your baseline; the third tells you whether anything you shipped mattered.

How this was measured

On 4 September 2026 we ran 5 buyer prompts per category against ChatGPT and Perplexity — 10 answers per category — across 5 B2B SaaS categories, with 6 vendors tracked in each: 30 companies in total, 300 brand-answer pairs. For every answer we recorded which vendors were named, which were cited by URL, and every source the engine returned: 706 cited URLs across 313 distinct domains. The prompts were written the way a buyer types them, not the way a marketer would.

One caveat worth stating, because it moves the numbers: the third prompt in every category is a head-to-head that names three of the six vendors outright. Those three get a mention rate that is partly handed to them by the question. The leaders below still lead on the four prompts that name nobody — but a vendor’s score is not purely earned, and any audit that hides its prompts is hiding that.

Here are all 25, verbatim.