Use comparable samples
Keep prompt intent, market, language, provider mix, and date range stable when comparing two periods.
Free measurement tool
What share of AI recommendations is yours?
Use observed answer counts to create a transparent baseline for AI share of voice. Mention rate, recommendation rate, and competitive share answer different questions.
Your sample
Enter observed answer counts
Use one consistent sample. Do not mix branded and category prompts, or providers and countries, without labeling the segments.
Total answers in the same prompt, provider, market, and language sample.
Count an answer once, even if it repeats the brand name.
Use a stricter classification than a casual mention or list inclusion.
Sum the comparable recommendation placements for the competitors you track.
Observed result
Three numbers worth keeping separate
Mention rate
34%
34 of 100 answers
Recommendation rate
18%
18 of 100 answers
Recommendation share
30%
18 of 60 tracked placements
How to read this
- High mention, low recommendation: the brand is known but not winning the buyer’s decision criteria.
- Low mention, high recommendation: the sample may be too small or overly branded; inspect prompt coverage.
- Low recommendation share: inspect competitor rationale and the sources cited before rewriting pages.
This calculator is a directional planning aid. It does not correct for prompt difficulty, answer length, provider weighting, or duplicate placements.
Classify the answer
A casual mention is not the same as a shortlist placement or a recommendation. Write the classification rule before counting.
Inspect the reason
A percentage shows the gap. The answer, competitor, and cited source explain what to fix next.
Next diagnostic
Turn the baseline into an improvement loop
Once you have a baseline, map the prompts where competitors win, inspect their source patterns, and update the smallest useful content or evidence gap.