I was running MealCoreAI, a health app with 12,000 active users in India.
I asked ChatGPT to recommend meal planning apps for people managing
diabetes in India. It gave me three tools. None of them were Indian.
MealCoreAI did not appear at all.
That was the moment I knew something was structurally broken.
So I spent two weeks systematically testing every major Indian startup
I could think of across ChatGPT, Gemini, and Perplexity using GeoIQ's
audit tool. The results were worse than I expected.
The results
| Brand | Sector | Score | ChatGPT | Gemini | Perplexity |
|---|---|---|---|---|---|
| Zoho | B2B SaaS | 72/100 | 34 | 22 | 16 |
| Freshdesk | SaaS | 61/100 | 28 | 18 | 15 |
| OYO | Travel | 55/100 | 26 | 16 | 13 |
| Razorpay | Fintech | 54/100 | 24 | 16 | 14 |
| PhonePe | Fintech | 47/100 | 18 | 18 | 11 |
| Groww | Fintech | 38/100 | 16 | 14 | 8 |
| Meesho | E-commerce | 33/100 | 10 | 12 | 11 |
| MealCoreAI | Health | 30/100 | 0 | 0 | 30 |
| Zepto | Q-commerce | 22/100 | 8 | 6 | 8 |
| Unstop | EdTech | 18/100 | 4 | 6 | 8 |
GeoIQ audit, May 25 2026. Total score out of 100.
Average score: 43/100.
Unstop serves 10 million students. It scored 18/100.
Meesho has 150 million users. It scored 33/100.
That gap between product scale and AI visibility is the whole problem.
What the numbers actually mean
Zoho and Freshdesk top the list not because of any secret strategy.
They have been building global citations for 20+ years — G2, Capterra,
Crunchbase, TechCrunch, analyst reports. Their scores reflect the
compounding effect of a long citation history.
MealCoreAI is the most telling result.
- ChatGPT score: 0
- Gemini score: 0
- Perplexity score: 30
What does this mean? Perplexity uses live web search and can find
MealCoreAI. But ChatGPT and Gemini rely on training data — and the
brand launched too recently, has no Crunchbase profile, and no
ProductHunt page. The web knows it exists. The AI models' training
data does not.
Meesho and Zepto are the surprise. Combined they have
over 150 million users and $1B+ in funding. Both score under 35/100.
The reason: their user base is hyperlocal India, their press is
primarily Hindi-language or regional, and their product categories
don't map cleanly to the English query templates that dominate
AI training data.
The three things separating Zoho (72) from Unstop (18)
1. Years of English-language, globally-indexed citations
Zoho has been covered in TechCrunch, Wired, Forbes for two decades.
Every article is a training data citation. Unstop has been covered
primarily in campus newsletters and Hindi startup press — most of
which never made it into GPT-4 or Gemini's training pipelines.
2. Presence on global review platforms
G2, Capterra, ProductHunt, and Crunchbase are crawled heavily by
AI training pipelines. Zoho has thousands of G2 reviews.
Unstop and MealCoreAI have near-zero presence on these platforms.
3. Consistent entity signals
Search for "Zoho" across LinkedIn, Crunchbase, Wikipedia, and G2 —
you get consistent descriptions of the same company. AI systems use
this consistency to build strong entity signals. For newer Indian
brands, descriptions are often inconsistent across platforms,
which fragments the signal.
The five fastest fixes
1. Allow AI crawlers in robots.txt (10 mins)
Visit yourdomain.com/robots.txt right now. Add:
User-agent: GPTBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Claude-Web
Allow: /
User-agent: anthropic-ai
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: Bingbot
Allow: /
If these crawlers are blocked, no other fix matters.
2. Create llms.txt (15 mins)
Put a plain text file at yourdomain.com/llms.txt:
Your Brand Name
One sentence description.
About
2-3 paragraphs about your product and who it serves.
Key Pages
Homepage: Main page
Contact
hello@yourdomain.com
Sitemap
https://yourdomain.com/sitemap.xml
Under 5% of Indian startups have done this. Takes 15 minutes.
3. Create a Crunchbase profile (20 mins)
Go to crunchbase.com/add-new/organization. Fill every field.
This is the single highest-impact citation for ChatGPT visibility.
ChatGPT's training data indexes Crunchbase heavily and it gives your
brand a structured, authoritative identity on the global web.
4. Submit to ProductHunt and G2 (1-2 hours)
Each contributes independent citations that AI training data picks up.
Encourage your existing users to review you on G2 —
3 minutes per reviewer, outsized impact on AI visibility.
5. Get covered in YourStory and Inc42 (2-4 weeks)
For Gemini specifically, Indian publication coverage matters.
YourStory (DR 71) and Inc42 (DR 71) are indexed in Google's
knowledge graph which feeds Gemini's India entity recognition.
The window is still open
Less than 5% of Indian startups have an llms.txt file.
Less than 30% allow all AI crawlers in robots.txt.
Most have no Crunchbase profile, no G2 reviews,
and no Organization schema on their homepage.
The brands that move on this in the next 3-6 months will have
a meaningful head start. AI visibility, like SEO a decade ago,
rewards early movers significantly.
The window is open right now.
Check your own score
I built GeoIQ to track this —
it checks your brand across 6 AI systems in 60 seconds.
Free audit, no signup needed.
Drop your score in the comments —
genuinely curious what Indian devs and founders are getting.
Top comments (0)