Weekly evolution: how your AI visibility changes week by week
Updated on
The weekly evolution shows how the presence of your brand in the answers of artificial intelligence models moves over time, with one point for each full week and three series that answer different questions, the mention rate, which tells you how often you appear, the total number of mentions, which tells you how much you are named, and the AI Visibility Score, which also weighs the place you take within the answer.
Three weekly series from a real project
Across the nine weeks of the example project the mention rate moved between 30% and 36%, the AI Visibility Score between 29.3 and 34.2, and mentions between 133 and 194 per week, with between 123 and 188 answers analyzed each week.
Mention rate
AI Visibility Score
Mentions
Real data from a production project: 1,503 responses analyzed between August 6 and October 4. Brand names have been replaced with generic labels.
What does each of the three series answer?
The mention rate is the fairest one for comparing weeks with each other, because it calculates in what percentage of that week's answers you appear at least once and is therefore unaffected by whether you add or remove prompts, whereas the total number of mentions is the raw volume and also rises if you add new questions, so it should always be read together with the rate.
The AI Visibility Score adds a nuance the other two do not capture, which is that it does not give the same value to being first as to being tenth, so if this number rises faster than the mention rate it means that, besides naming you more often, the models are placing you earlier within the answer.
How often is it updated and what data is needed?
Each prompt has its own run frequency, which can be daily, every two days, weekly, every two weeks or monthly, and an automatic scheduler checks every hour which ones are due, so the weekly series fills itself in without you having to launch anything by hand, and weeks start on Monday.
At least two weeks of data are needed to draw the evolution, since a single week offers nothing to compare, and only completed answers are taken into account, which prevents a half-finished run from distorting the result of a week.
How do you tell a trend from simple noise?
Variations of one or two points between consecutive weeks are normal, because the models do not answer identically every time, which is why in the example the rate oscillates between 30% and 36% without anything relevant having changed, so a reliable conclusion requires several weeks in the same direction and, if possible, that it coincides with a specific action you have taken.
Noting down the dates of important actions, such as a content campaign, a website improvement or a listing in a directory, and comparing them afterwards with the three series is the simplest way of knowing whether what you do works, which is precisely the reason this view exists.
Frequently asked questions
- Why are two weeks of data needed?
- Because an evolution is a comparison between different moments, and with a single week there is only one point, so the dashboard shows a notice until at least two full weeks with analyzed answers exist.
- What is the difference between the mention rate and the AI Visibility Score?
- The mention rate only tells appearing apart from not appearing, whereas the AI Visibility Score gives more value to high positions through a logarithmic weighting, so two brands with the same rate can have different scores if one of them systematically appears earlier within the answer.
- Why can the total mentions rise without my position improving?
- Because it counts every time you are named, even several times within the same answer, and it also grows if you add new prompts to the project, so it should not be used alone to measure improvement but read together with the mention rate and the AI Visibility Score.