Know where you rank.
Eight buyer questions, each asked five times per report. See supported recommendations and observed list positions, not just mentions.
Pickentra tracks how your business appears across eight consistent local buyer questions, with five independent answers per question in every complete report—showing recommendations, mentions, citations, competitors, and the actions most likely to improve your visibility.
Our latest Batch benchmark uses OpenAI GPT-6 Sol (gpt-6-sol). Each real report records the actual model and processing mode used; Standard or historical runs may use a different model, including chat-latest.
Creating an account requests access. Reports require manual approval, completed setup and enabled services. Batch processing is not instant.
"Which HVAC companies in Boise can repair a furnace?"
Observations with evidence. No guaranteed rankings, leads or sales.
Eight buyer questions, each asked five times per report. See supported recommendations and observed list positions, not just mentions.
Competitors observed in the answers, linked to the questions that named them.
The websites and directories each answer linked to, and whether your own domain was one.
Specific things to check and test, each tied to a stored answer. Not guaranteed fixes.
Up to 20 public pages of your site are read, then you confirm the details.
Neutral, repeatable questions reflecting local purchase intent, then frozen.
Five fresh answers per question: 40 tests per complete report, with recommendations, citations, competitors and evidence-backed actions.
A separate AI classification reviews each saved answer. Recommendation Rate is supported recommendations divided by valid completed buyer tests. First-Listed, Citation and Mention Rates stay separate. We do not award points for accuracy or next steps.
Illustrative counts for the fictional Bluebird Heating & Air in Boise, ID. Not a customer result.
Worked example: 20 supported recommendations in 40 valid completed tests = 50%. Eight approved buyer questions, five independent AI answers per question: 40 tests per complete report. Failed tests are excluded; uncertain classifications leave the affected rate unavailable. The latest report's eight actual questions define the dashboard comparison. Matching questions count across scans, even if previously labeled rotating or run with another model. Changed questions remain visible separately as unstable. The latest scan score and pooled matching-question rate are shown separately.
Mentions
25 / 40 tests
Answers that named the business anywhere.
Supported recommendations
20 / 40 tests
Answers that suggest considering or contacting this provider.
Verified own-domain citations
10 / 40 tests
Answers that linked to its own website, checked.
The sample report walks through one fictional week: answers, ranks, citations, score and recommendations.
Creating an account requests access. Reports require manual approval, completed setup and enabled services. Batch processing is not instant.