How AI assistants choose the brands they name.

The engine counts how often each brand's name appears across the pages it read. Perplexity says so in its own answers, in the same words again and again. Three examples, from real estate, home decor and skincare:

The engine is describing its own method. It counts how often a name appears across the pages it read. Being on those pages matters more than how good any one page is. That is why the same two or three brands hold a category's answers on every engine. It is also why nobody can buy their way in.

The engines will read a brand's own site as a source and still recommend someone else. Vaaree's domain was the second most cited source in the home decor answers we ran, above ikea.com. Vaaree was still named in 16 of 30 answers, and Pepperfry in 23.

Pet care went the same way. Several answers cited supertails.com and then recommended Royal Canin. The counts for both categories are in the research report: how AI assistants name Indian brands.

The payroll study shows the same thing at category scale. Waggex and Trilliant Software both had pages the engines read as sources. Neither was named to the buyer in any of the 159 answers. A citation is the engine using your page to write about your competitor.

In the payroll study, five companies were on Google's first page on 28 August 2026 for one of the forty questions. None of them was named in any of the 159 answers. The query each one ranks for is beside its name.

The same brand can be strong on one engine and weak on another. The engines read different kinds of pages. Across the HRMS answers we ran, ChatGPT cited vendor pages most, greythr.com and keka.com in roughly two thirds of its answers. Perplexity cited listicles and LinkedIn posts instead. HROne was named in 13 of 15 Perplexity answers and 6 of 15 ChatGPT answers, for the same fifteen questions.

The four-engine payroll study found the same spread the other way round. HROne was named 22 times by Google AI Overviews and 13 times by Perplexity. ChatGPT named it 12 times and Gemini 5, out of forty questions each.

The weak engine is the one that reads the page type your category has not published. When a brand scores the same on both engines, the engine is not the problem. The brand is simply not on the sources either one reads.

Whatever you claim about yourself has to be written down in a plain sentence somewhere. The engine copies what it reads. It does not guess.

So the engine gives a claim to whoever wrote it down most clearly. It answers is-it-any-good from the harshest review profile it can find, not the biggest one. And if you publish no rupee price, it takes a price from somewhere else.

If you invent your own words for what you sell, the engine files them under your brand name. It never connects them to the words buyers actually type. The worked examples behind all four points, with the counts, are in the research report: how AI assistants name Indian brands.

• Count first. Run the buyer's questions and find out which sources the engines read for your category. Those pages, not your homepage, decide the shortlist.
• Get onto the recurring sources. The action differs by vertical: user reviews and Reddit for D2C, clinical-grade content for health, directory and regulator listings for real estate.
• State what you are in one plain sentence, on a page the crawler can reach. Use the words buyers use, and put a rupee price beside it.
• Consolidate the review profiles the engines quote from, especially the small angry ones.
• Measure per engine. A brand can be fine on ChatGPT and invisible on Perplexity, and the fix is different for each.