Short poll
AI visibility poll: results
Results will appear here once enough responses are in.
Method
How results are published
The site stores a running count per option and a random, answer-free receipt for each submission so an accidental resend within 24 hours is counted once. Receipts are deleted on a schedule, normally within 25 hours; after that a resend would count as a new submission. Answers are never linked to an address or account, so this is one count per submission, not one per person. A short-lived, in-memory rate limit per connection expires after a one-hour window and is removed by periodic cleanup, normally within about 70 minutes while the service is running (a restart clears it; a paused or delayed process can delay cleanup). It is not written anywhere; the web server and proxy keep ordinary request logs that may include IP addresses, as for any page.
A new snapshot is published only when at least 20 new responses have arrived since the last one, every option that changed has changed by at least three responses, and at least six hours have passed. Percentages are whole numbers and the total is rounded down to the nearest ten. This reduces what any two published snapshots can reveal to groups of three or more responses; it is batching and suppression, which lowers inference risk but cannot guarantee anonymity.
Earlier figures
Results declared as of 28 October 2026
What did an earlier WhatsApp group poll report?
In an informal WhatsApp group poll, 220 responses were reported. The counts below are the only figures shared with us, and they are self-reported observations, not measurements.
- 90 responses: big brands appear in AI answers.
- 70 responses: their own brand appears, but not everywhere.
- 30 responses: they do not know how to get their business into AI answers.
- 30 responses: they are happy with early AI references or traffic.
How much weight should you give that poll?
Not much on its own. These are self-reported observations from an informal WhatsApp group poll. The original question, poll date, sampling method and whether responses represent unique people have not been verified. The findings do not represent the wider market or establish what caused visibility or traffic.
What the poll is useful for is the question it raises for your own brand: do AI assistants name you, and if so, everywhere or only sometimes? The ChatGPT guide is how to find out with your own prompts rather than someone else’s impressions.