How to measure whether ChatGPT and Perplexity mention your store (honestly)
AI answers change from one run to the next. A method for measuring mentions and citations that gives you a trend you can trust — not a vanity number.
September 1, 2026 · 3 min read · measurement · share of voice · Perplexity · ChatGPT
"Are we on ChatGPT?" is the question every merchant asks and almost nobody can answer properly. Here's a method that works — and the two traps that make most "AI visibility scores" meaningless.
Trap 1: one answer is not data
Ask ChatGPT the same question twice and you'll get two different answers. Ask through the API and you'll get a third. A screenshot of one good answer proves nothing; a screenshot of one bad answer proves nothing either. Only repeated measurement over time means anything.
Trap 2: measuring the wrong question
If you test "best perfume store in Saudi Arabia," you're measuring a generic brand race you probably can't win yet. The questions that matter are the ones your actual customers ask: specific, situational, often in dialect. "وين أشتري عود كمبودي أصلي بالرياض؟" tells you more than any generic query.
The method
1. Build a prompt panel. 5 to 30 questions in three groups:
- Brand — "what is [store]? is it trustworthy?"
- Category / problem — "best [category] for [situation] in [city]"
- Alternatives — "stores like [competitor]"
Write them the way customers talk, in the languages your customers use.
2. Run them on a schedule. Same prompts, same platforms (Perplexity and ChatGPT to start), weekly or monthly. Record for each answer:
- Was the brand mentioned? (fuzzy match — Arabic and Latin spellings)
- Was your domain cited as a source?
- What position — first, later, absent?
- Which competitors appeared?
3. Watch the trend, not the number. Mention rate 20% → 35% over three months is a signal. 35% on one Tuesday is noise.
4. Connect it to changes. Fixed your robots.txt in week 3? Rewrote category intros in week 6? The trend line tells you what moved the needle.
What a good result looks like
Not "we're #1 on ChatGPT." Something like: "In Saudi-dialect category questions, we went from absent to mentioned in 40% of Perplexity answers and cited in 25%, while competitor X's share fell." That's a sentence a business owner can act on.
Doing it yourself vs. letting a tool do it
You can run this by hand with a spreadsheet and patience. It gets tedious after the second week, which is exactly when consistency starts to matter. NIMSEO's visibility check automates the panel, the scheduling and the detection, and shows the trend in your dashboard — with an honesty label on every result, because we'd rather tell you it's an approximation than pretend it's a ranking.