Confidence confidence
At shortlisting, how sure the model is that a candidate is a good pick, given how much it could verify, such as high for a roofer with consistent listings and reviews.
What you can do: Keep name, address, phone and hours identical across your site and listings so the model can verify you easily.
evaluation.scoring (was selection.policy in the provisional inventory: Confidence level is a score, not a shortlist rule) · also called certainty, assurance
Checked by What AI answers say your phone, address and hours are · Term: confidence-score
Justification justification
At shortlisting, the reasons the model writes for picking or dropping a candidate, such as "licensed, 20 years local, strong reviews" for a plumber.
What you can do: State the facts that would justify choosing you, like license, years local, service area and guarantees, on your site in plain text.
answer.justification (was selection.policy in the provisional inventory: Reasons for picking or dropping a candidate) · also called rationale, explanation
Risk Note risk_note
At shortlisting, a note on a concern about a candidate, such as unverified insurance or a recent complaint pattern, that the model keeps even if it still lists the business.
What you can do: Close the gaps that cause such notes: show insurance and license, answer complaints, and fix outdated listing details.
evidence.assessment (was selection.policy in the provisional inventory: Flags a risk judged from evidence) · also called risk_assessment, warning
Checked by Field-by-field audit of your listings
Runner Up runner_up
At shortlisting, a candidate the model rates as strong but just behind the top pick, such as the second electrician offered in an answer.
What you can do: Make the facts that set you apart easy to find, such as specialties, response time and warranties.
evaluation.scoring (was selection.policy in the provisional inventory: Rank position outcome, not a shortlist rule) · also called second_choice, alternative
Winner winner
At shortlisting, the candidate the model picks as the best overall match for the query, such as the one fence company it names first for Denver.
What you can do: Cover the basics the model checks: matching service, clear service area, verifiable license and consistent contact details.
evaluation.scoring (was selection.policy in the provisional inventory: Rank position outcome, not a shortlist rule) · also called top_choice, selected
Checked by Pages AI cites for your trade and city, and which you are missing from
Used in 2 of 100 census runs this week: “Name, why it stands out (ratings, experience, etc.) with contact info”
Reason reason
At shortlisting, one specific factor behind a choice, such as "offers 24/7 emergency service" or "located in the city asked about".
What you can do: State your distinguishing facts as plain, quotable sentences on the site.
answer.justification (was selection.policy in the provisional inventory: A reason behind a choice) · also called factor, consideration
Recommendation Status recommendation_status
At shortlisting, whether the model will recommend a candidate, list it with caveats, or leave it out.
What you can do: Remove reasons for exclusion: outdated contact details, closed flags on listings, and unverifiable claims.
selection.policy · also called endorsement, suggestion_status
Checked by Anywhere that marks you closed or moved
Used in 1 of 100 census runs this week: “Provide the name, contact info, website, and a short justification referencing the scores/criteria.”
Community Quote Candidates community_quote_candidates
At shortlisting, candidates the model can back with a quote or summary from customer reviews or forums, such as a review praising a tree service's cleanup.
What you can do: Ask customers for detailed reviews that name the service and location; these are easier for the model to quote.
selection.policy · also called community_feedback, peer_reviews
Official Source Backup official_source_backup
At shortlisting, backup candidates the model can support with official sources, such as a licensing board, BBB profile or the business's own site.
What you can do: Keep your official site, licensing board entry and BBB profile accurate and consistent with each other.
selection.policy · also called verified_sources, trusted_references
Checked by Your directory listings, found and matched · Term: authority-weighting
Backup Options backup_options
At shortlisting, other candidates the model keeps in reserve if the top pick is booked, closed or unsuitable.
What you can do: Show availability and booking options clearly so the model knows you can take the job.
selection.policy · also called alternatives, substitutes
Used in 3 of 100 census runs this week: “briefly list 2‑3 other excellent options”
Bigger Project Option bigger_project_option
At shortlisting, a candidate the model offers for a larger job than the typical one, such as a remodeler who also takes full additions or a caterer for 500 guests.
What you can do: State the largest jobs you take, with examples, crew size and capacity.
selection.policy · also called expansion_candidate, large_scale_option
Caution caution
At shortlisting, a flag that the user should be careful with a candidate, such as mixed reviews or a license that could not be confirmed.
What you can do: Check that your license, insurance and contact details are verifiable, and answer negative reviews.
evidence.assessment (was selection.policy in the provisional inventory: Caution flag judged from evidence) · also called warning, alert
Used in 1 of 100 census runs this week: “Reminder to verify credentials and discuss fees.”
Best Budget Alternative best_budget_alternative
At shortlisting, the candidate the model names as the best pick for a tight budget, such as a lower-cost handyman with decent reviews.
What you can do: If you compete on price, say so with specific prices or ranges and what they include.
selection.policy (was offer.fit in the provisional inventory: Shortlist slot, like other best/runner-up picks) · also called affordable_choice, cost_effective_option
Checked by The claims your site makes: years, prices, refund window
Best Overall Official Source Choice best_overall_official_source_choice
At shortlisting, the top candidate whose credentials the model can confirm through official sources, such as a state license lookup or the business's own site.
What you can do: Show license numbers and link to your licensing board entry from your site.
selection.policy · also called top_verified_choice, official_recommendation
Best Quality Runner Up best_quality_runner_up
At shortlisting, the runner-up the model offers for users who put quality first, such as a pricier but highly rated roofer.
What you can do: Document quality with specifics: certifications, materials used, warranties and detailed reviews.
selection.policy · also called quality_alternative, second_best