Home / Resources / AI Visibility Study

We asked ChatGPT about 50 Indian brands. 34 were invisible.

We took 50 well-known Indian mid-market brands — companies with real revenue, real customers, often decades of trading history — and asked the AI engines the questions their buyers actually ask. The results should worry every marketing head in the country.

How we tested

For each brand we built a small panel of buyer-intent prompts and ran them across ChatGPT, Gemini and Perplexity:

• Category prompts — "best category companies in India", "recommend a category supplier for a mid-size business"
• Direct prompts — "what does brand do?", "is brand good?"
• Comparison prompts — "brand vs competitor"

A brand scored visible if it appeared unprompted in category answers, and accurate if direct answers described its offering, scale and location without material error.

What we found

• 34 of 50 brands never appeared in category answers on any engine. Their buyers asking AI for recommendations were being sent — every single time — to competitors.
• Of the 16 that did appear, 9 were described with material errors: outdated product lines, wrong locations, defunct pricing, or confusion with similarly-named companies.
• Only 7 of 50 were both visible and accurately described. That is a 14% pass rate among companies that dominate their categories offline.
• The engines agreed with each other far more than they agreed with reality — they cite the same handful of sources, so one stale directory listing propagates everywhere.

What the invisible 34 had in common

1. No entity footprint. Little or no presence on the reference sites and trade directories the engines actually cite — however strong their own website was.
2. Brochure-ware content. Sites full of "welcome to our company" prose, with nothing quotable: no data, no direct answers, no comparisons.
3. JavaScript-locked pages. Several sites rendered their entire product catalogue client-side, invisible to retrieval crawlers.
4. Blocked AI crawlers. Four brands blocked GPTBot outright — usually a developer's default, not a decision anyone made.

What to do about it

The fix is systematic, not magic — entity clean-up, answer-shaped content, citable data and corroboration in the sources engines trust. We've written the full method in our Complete Guide to Generative Engine Optimization, including the 90-day playbook we run for clients.