Guides
Measuring what buyers ask AI
How to measure brand visibility in AI assistants: prompt-set design, reading tracker numbers, and evidence from the prompt sets we grade.
· 5 min read
Model updates move your AI visibility more than you do
A silent model update can move every prompt in your set at once. Why AI visibility jumps without cause, and how to stop it being read as a trend.
· 6 min read
Does prompt wording change your AI visibility score?
Prompts in category jargon can report 6.6 times more brand mentions than buyer situations. How to check the register of your prompt set.
· 5 min read
How big does a change in AI visibility have to be to be real?
Most month-on-month AI visibility changes sit inside the margin of error. How to work out the smallest change your prompt set can actually see.
· 5 min read
How many prompts does an AI visibility prompt set need?
At least 30, and usually 100 or more. How to size a prompt set from the change you want to detect, the topics you report and the markets you track.
· 6 min read
Why more prompts beat more repeats in AI visibility tracking
Running one prompt 100 times is not 100 observations. How the design effect shrinks your sample, and where to spend a tracking budget instead.
· 5 min read
Branded prompts inflate your AI visibility score
Prompts that name your brand return your brand. Why they flatter AI visibility numbers, how much, and what share of a prompt set they should be.
· 5 min read
How to build an AI visibility prompt set from real buyer language
An eight-step method for writing AI visibility prompts from sales calls, support tickets and search data, so your tracker measures how buyers really ask.
PromptSet · Newsletter
Get new posts as PromptSet, every other Thursday.
How to build prompt sets that reflect real buyers, how to read the numbers trackers report, and anonymised data from the sets we grade. About five minutes, free. What's in it