AI Visibility Index

The AI Visibility Index is a public ranking Rixe publishes for a sector in one country: how often AI engines name each brand when people ask the sector's typical questions. Anyone can read it, without an account. It is built to survive being checked: the sample sizes and confidence intervals are printed, and a brand measured too few times is listed but not ranked.

Read this page if your brand appears in an index, or if you want to quote one.

#Where to find the indexes

Published indexes are listed at rixe.ai/endeks. Each index has its own public page, with a Download as PDF button.

Indexes are created and published by Rixe. They are not built from customer data and customers cannot create their own.

#How an index is made

• One country, one sector. An index is about one country's brands: a bank index for Türkiye and one for Saudi Arabia are two different indexes, not one index in two languages. It is written in that market's language.
• A hand-picked brand list. The brands are chosen by hand. The list does not claim to represent the whole sector.
• Standard questions that name no brand. The same set of questions is put to several AI engines. No question names a brand, so every mention is the engine's own choice.
• Independent measurement. The answers come from queries made for the index, not from any customer's account.
• Monthly editions. Every published index is measured again on the 2nd of each month (06:00 UTC), with the same brands and questions. The previous edition becomes history, which gives the Change column. Up to twelve monthly editions are kept.

#Reading an index page

#Summary

The top of the page names the period, the number of Questions, AI engines, Brands and Answers measured, and the Leader: the brand named most often, with its rate. If other brands are statistically level with the leader, it says so.

#Highlights

A few sentences the data supports, for example who leads, how big the gap to second place is, which brands were never named, and how many brands appear on only one engine.

#Ranking

Two rules keep the ranking honest:

• Ties are shown as ties. Where two brands' confidence intervals overlap, the difference between them is not statistically significant, and the table says the brand is statistically tied with the other. A place in the ranking does not mean it is ahead of a brand it is tied with.
• Too few measurements, no rank. A brand with fewer than 20 measurements is listed but not ranked, marked insufficient sample. A high rate from a handful of answers can be chance.

#Visibility by engine

A heat map: each cell says how many of every 100 answers from that engine named the brand. Darker means more often. A dash means that engine gave no answers in this measurement.

#Who wins each question

For each question, the three brands named most often.

#Methodology

The method section is generated from the run itself, never typed by hand: the period, the number of questions, the engines asked, the total measurements, the ranking threshold and how the brand list was chosen.

#Is this your brand?

Each brand row has a Your brand? link, and the page ends with an Is this your brand? box. These lead to the brand's full scorecard — which engines name it, which questions rivals win, and how far it is from the leader. The scorecard is shown only to people signed in with a confirmed e-mail address on the brand's own domain. See Scorecards and newsletter.

#Common questions

My brand is missing from an index. The brand list of each index is chosen by hand by Rixe and does not claim to cover the whole sector.

Why is a brand with a higher rate shown as tied with a lower one? Their confidence intervals overlap, so the data cannot say which one is really ahead.

How is the index different from my own score in Rixe? The index uses a fixed, public set of questions measured once a month. Your workspace is measured daily on your own questions and competitors — see Visibility score.

#Related pages

• Scorecards and newsletter
• Visibility score
• How measurement works