TL;DR: SEO gets you ranked, AEO gets you quoted as the answer, and GEO builds the wider authority that makes AI engines recommend you. We wrote 15 buyer-style prompts and are running them across ChatGPT, Gemini, Claude and Perplexity on our own brand, before changing anything. Here is the method, what we are changing, and how we will judge it.
Everyone selling AI search visibility should be able to show their own numbers. So I am measuring mine, before changing anything.
What is the difference between GEO, SEO and AEO?
Quick answer: SEO is the foundation: content and technical work that gets pages ranked in classic search. AEO (answer engine optimisation) makes content look like an answer: query-matched headings, concise definitions, schema. GEO (generative engine optimisation) is the broader authority layer that makes AI systems cite and recommend your brand across many sources.
One ecosystem, three jobs. They overlap, and the fix order matters more than the acronym. The longer comparison is on GEO vs SEO vs AEO: A Practical Comparison.
How are we testing it?
Quick answer: We wrote 15 prompts a buyer might ask: 5 on fractional CMOs in India, 4 on marketing diagnostics and the Hexagram, 4 on AI search consulting, and 2 on the brand itself. Each runs on ChatGPT, Gemini, Claude and Perplexity, and we record whether ADG was mentioned, whether it was linked, and who else appeared.
The first run, on 1 October 2026, was not clean enough to publish figures from. Two engines had account memory switched on, which flatters a brand the account has already talked about, and one engine could not be read at all. A baseline you cannot trust is worse than no baseline, because every later number gets compared against it. So the baseline is being re-run under clean conditions, and the numbers will be published from that run.
What are we recording?
Quick answer: For every prompt and engine pair: mentioned or not, linked or not, position in any list, which competitors were named, and whether the description of ADG was accurate. We also check for name collisions with similarly named firms.
The prompts are grouped on purpose. A brand can do well when someone asks about it by name and still be invisible on the questions that bring in new clients. Grouping by theme separates "do the engines know who we are" from "do the engines recommend us when a buyer has a problem". Only the second one pays.
What does LinkedIn's own research say about getting cited?
Quick answer: LinkedIn's 2026 B2B guide, drawing on Meltwater, Semrush, Ahrefs and Profound data, says individual profiles drive about 75 percent of LinkedIn citations, long-form articles about 60 percent, and that median time to a first citation is about 7 days, with 90 percent of cited pages arriving within 37 days.
Treat this with caution: LinkedIn has an interest in saying LinkedIn matters, and the figures are the guide's, not ours. Its own closing advice is the right one: test on your own data.
What are we changing, and how will we know it worked?
Quick answer: Five changes: one consistent name and company descriptor everywhere, schema that ties our profiles to our site, long-form articles on the queries buyers ask, real expert bylines, and a monthly re-run of the same 15 prompts. We will judge results at day 30 after the clean baseline, not before.
- One canonical name and one descriptor, "ADG Advisory, Bengaluru, India", on every surface, to separate us from similarly named firms.
- Organisation and person schema linking our site, LinkedIn, Substack and social profiles.
- Long-form articles on cost, comparison and diagnostic queries, the ones that bring in new clients.
- Content published under a real, named expert, not an anonymous brand.
- The same 15 prompts, same conditions, re-run at day 30, with results published.
Checklist: measure your own AI search visibility
- Write 10 to 15 prompts a real buyer would type, grouped by theme.
- Run each on ChatGPT, Gemini, Claude and Perplexity, logged out or with memory off.
- Record: mentioned or not, linked or not, position, competitors named, sentiment.
- Check for name collisions with similarly named companies.
- Fix entity consistency before publishing more content.
- Re-run on the same prompts after 30 days. Do not judge at week three.
If you want to see what the engines say about you today, Is Your Business Invisible in ChatGPT and Perplexity? is a ten-minute version of the same check.
A visibility claim without a baseline is a guess. We would rather publish the baseline late than publish a guess.
The Hexagram Diagnostic scores The Signal pillar, which covers SEO, AEO and GEO, alongside the other five pillars of your marketing architecture. It is free and takes 8 minutes.
