Research · Sample report
SAMPLE · ANONYMIZED
What a finished run actually says.
Seven measurements, reported separately, each with the evidence behind it and the change it points to. Figures below come from an anonymized run, so no brand or domain is named.
MEASUREMENT
READING
EVIDENCE
CHANGE IT POINTS TO
Question Finder
12 buyer questions found, 4 worth measuring first
Questions recurring across all three assistants on the same run date
Fix the two questions with the largest gap first
Brand Recall
UNKNOWN in two of three assistants
Asked by name and by domain stem, separately, three times each
Publish an about page that states what the company sells, plainly
Category Fit
Filed as “gift shop”, expected “outdoor equipment”
Category named by each assistant for the same product page
Name the category in the page title and first paragraph
Buyer Match
Two pages answer the question, five are related, one is far off
Each page measured against the question against Ring 1
Rewrite the far-off page or point the question at a better page
Competitor’s Citations
Competitors cited from 9 sources; this brand from 2
Sources named in the answers, with the run date
Pursue the two sources cited for every competitor
AI Recommendations
Not shortlisted on 3 of 4 questions
Shortlists recorded per assistant, per repeat
Close the recall and category gaps before chasing the shortlist
Content
One page reads as marketing, not as an answer
Page text scored against the question wording
Answer the question in the first 60 words of the page
Every reading carries the models asked, the question asked and the run date. Readings are dated, not permanent, which is why re-measuring after a change is part of the workflow.
Run this on your own site.
The first read is free. No card until you go past it.