
- Zoho Payroll
- RazorpayX Payroll
- greytHR
- Keka
Issue 01 · Payroll and HRMS software · India · 28 August 2026
Forty questions, four engines: ChatGPT, Perplexity, Gemini and Google AI Overviews. 160 runs, 159 answers. Run by hand in the consumer apps from India, logged in as an ordinary user, every answer read and tallied by a person.
01 · The questions
02 · Four engines, four answers
Question 1 to every engine. Brands in brass are the ones it named.
Which payroll software is best for a 50-person company in India?


Re-captured 29 August 2026
03 · The leaderboard
Times named, out of every possible answer in the view. The unprompted view removes the six questions that name a brand in the question itself.
04 · Engines differ
ChatGPT gives a tight five-name list. AI Overviews has a long tail. A brand can be strong on one and missing on another, so you have to work on each one.
HROne, times named per engine
05 · Named zero times
Each of these companies is on Google's first page for the query beside its name. None of them was named in any of the 159 answers.
0of 160
06 · Cited is not recommended
Engines cite vendor blogs and comparison pages as sources. Waggex and Trilliant pages were used to write answers that then recommended somebody else. Being crawled is not being chosen.
Read by the engine
Named to the buyer
07 · What this costs
OpenAI says India passed 100 million weekly ChatGPT users in February 2026. That is one assistant, in one country, in a single week, and it is the company’s own count, published by them and not independently audited.
OpenAI, “Introducing OpenAI for India”, 18 February 2026Pew tracked a month of real browsing for 900 people. On Google pages that carried an AI summary, 8% of visits ended in a click on a search result. On pages without one, 15% did. The answer is where the search stops.
Pew Research Center, 22 July 2025, United States, 900 tracked panellists, March 2025Ahrefs matched four million URLs cited by Google’s AI Overviews against the search pages for the same queries. 37.9% of them appeared in the first ten blocks of that page, ads, snippets and answer boxes counted alongside organic listings. Ranking and being cited have become two different jobs.
Ahrefs, 2 March 2026, 863,000 search pages, 4 million cited URLsWhat we would do for a zero-named brand
Open the doors. Check that every engine's crawler (GPTBot, PerplexityBot, Google-Extended, ClaudeBot) can reach and read the site, and fix what blocks it: robots rules, client-side rendering, pages behind scripts.
Structure what is already published. Mark up the pricing, comparison and FAQ pages with the schema each engine parses (Organization, Product, Offer in rupees, FAQ), so the facts on them can be quoted.
Add the files engines check first: an llms.txt, a clean sitemap, and one consistent set of entity facts (name, founded, headquarters, pricing) wherever the brand is described.
Brief, do not write. For each of the 40 questions, a page-level brief: which page answers it, what it must state, in what order. The team writes; we check it before it goes live.
Fix the third-party record. The listings and threads the engines cite for this category (G2, Capterra, SoftwareSuggest, Reddit) are audited for missing, wrong or rupee-less entries and corrected.
Re-run the 40 questions every month, per engine, and fix what moved.
Fifteen minutes. We pick ten of your buyers' questions, run them live, and show you who gets named.
Or write to [email protected]