Pick rate

By the Pickgap team · Updated Oct 15, 2026 · 1 min read

The pick rate is the share of AI answers that name a business, out of all the answers to a set of questions. If a business is named in 30 of 250 answers, its pick rate is 12%. Because AI answers vary, a pick rate should always come with its sample size and a noise range.

How it's calculated

1. Choose the questions customers ask, for example 25 local questions.
2. Ask each question on each AI surface, more than once.
3. Count the answers that name the business.
4. Divide by the total number of answers.

30 answers naming you out of 250 answers = a 12% pick rate.

Why not a rank?

AI assistants don't keep a fixed list. Each answer is generated fresh and names a few businesses, often different ones each time. A rank would change every time you looked. The pick rate measures what's stable: how often you're named over many answers.

The noise range

A pick rate from a small sample is uncertain. Pickgap reports a noise range with every pick rate: the middle 90% of rates we'd expect if the same questions were resampled. "12% ±4 · 30 of 250 answers" means a repeat test would most likely land between 8% and 16%. Changes smaller than the noise aren't counted as real movement.

The gap

Compare your pick rate with the most-picked competitor's. The distance between the two, in points, is your gap: "They're at 64%, you're at 12%: a 52-point gap." See what is a good pick rate? and our method.