AI classifier atlas · stage 5 of 16
The risks and trade-offs the model reasons about while it weighs the request: safety, money, time, trust.
87 labels, plus 8 spellings merged into them. Field names gpt-oss writes when asked to describe its evaluation; not OpenAI internals.
purposeThe model states the purpose behind the query, such as "find a reliable roofer for storm damage", and uses it to decide which businesses fit and what the answer should contain.
query.intent (was evidence.assessment in the provisional inventory: States the person's goal) · also called goal, objective
Term: search-intent
Used in 1 of 100 census runs this week: “identify the most trusted sign company with good reviews”
accepted_entity_typesThe model lists the kinds of business it will accept as answers, for example licensed plumbers and drain specialists for "clogged drain", before screening candidates.
What you can do: State your business category plainly on the homepage and in your structured data, using the same words customers search for.
candidate.eligibility (was evidence.assessment in the provisional inventory: Defines which business kinds may be considered) · also called valid_entity_types, recognized_entities
Term: schema-type
excluded_entity_typesThe model lists the kinds of result it will rule out, such as directories, lead-generation sites or national chains when the person asked for a local pet groomer.
What you can do: Make clear you are the business itself, not a lead broker or directory, with your own address, phone and staff.
candidate.eligibility (was query.scope in the provisional inventory: Hard screen excluding kinds of result) · also called invalid_entity_types, disallowed_entities
strong_corroborationThe model marks a fact about a business as strongly corroborated when several independent sources agree, such as the same address and hours on the site, Business Profile and Yelp.
What you can do: Keep your details identical on your site and every directory listing, and fix any listing that disagrees.
claims.verification · also called high_corroboration, strong_support
Checked by Your directory listings, found and matched
high_verifiabilityThe model rates a claim as easy to confirm from credible sources, such as a contractor license number that appears in the state registry, and gives it more weight.
What you can do: Back claims with something checkable: license numbers, association memberships, named certifications and links to where they can be confirmed.
claims.verification · also called strong_verifiability, high_confirmability
Checked by The claims your site makes: years, prices, refund window
high_weight_sourcesThe model marks sources it trusts most, such as a state licensing board, BBB or a well-known local news outlet, when weighing evidence about a business.
What you can do: Get listed and accurate on authoritative sources for your trade and area, such as licensing boards, BBB and local chambers.
evidence.provenance (was evidence.assessment in the provisional inventory: Classifies source trustworthiness by origin) · also called authoritative_sources, reliable_sources
Checked by Your directory listings, found and matched · Term: authority-weighting
Used in 21 of 100 census runs this week: “Google Business reviews, Yelp ratings, Angi (formerly Angie's List), Better Business Bureau (BBB) accreditation, Industry certifications (e.”
low_verifiabilityThe model rates a claim as hard to confirm from reliable sources, such as "best HVAC in town" with no awards or reviews behind it, and discounts it.
What you can do: Replace vague claims with ones that can be checked, and link to the source, such as the award page or license lookup.
claims.verification · also called weak_verifiability, low_confirmability · same label written as weakly_verifiable
Checked by The claims your site makes: years, prices, refund window
medium_corroborationThe model marks a fact as partly corroborated: some sources agree but others are missing or differ, such as hours on the site matching Google but not Yelp.
What you can do: Find the listings that disagree with your site or are missing and correct or claim them.
claims.verification · also called average_corroboration, medium_support
Checked by Your directory listings, found and matched
medium_verifiabilityThe model rates a claim as only partly checkable, such as years in business stated on the site but not shown on any registry or listing.
What you can do: Add supporting evidence for key claims, such as founding year in your structured data and a link to your business registration.
claims.verification · also called moderate_verifiability, average_confirmability
Checked by The claims your site makes: years, prices, refund window
medium_weight_sourcesThe model marks sources of middling trust, such as general directories or a business's own site, which count but need support from stronger sources.
What you can do: Make your own site's facts specific and consistent so they hold up when compared with stronger sources.
evidence.provenance (was evidence.assessment in the provisional inventory: Classifies source trustworthiness by origin) · also called moderate_sources, average_reliability
Term: authority-weighting
weak_corroborationThe model marks a fact as weakly corroborated when few sources confirm it or they disagree, such as a phone number found only on one old directory.
What you can do: Make sure your current details appear consistently on your site and the main directories, and remove outdated listings.
claims.verification · also called low_corroboration, weak_support
Checked by Field-by-field audit of your listings
low_weight_sourcesThe model marks sources it gives little weight, such as scraped directories, anonymous blogs or outdated listings, when judging a business.
What you can do: Correct or remove wrong details on low-quality directories, since the model may still read them.
evidence.provenance (was evidence.assessment in the provisional inventory: Classifies source trustworthiness by origin) · also called unreliable_sources, weak_sources
Checked by Field-by-field audit of your listings · Term: authority-weighting
hard_risk_signalsThe model flags serious warning signs about a business, such as being marked closed, an expired license, or many reviews describing no-shows, which can remove it from the shortlist.
What you can do: Fix any listing that marks you closed or moved, keep licenses current and visible, and respond to serious complaints.
evidence.assessment · also called high_risk_signals, serious_risk_indicators
Checked by Anywhere that marks you closed or moved
soft_risk_signalsThe model notes minor concerns, such as a few slow-response reviews or a missing price list, that lower confidence in a business without ruling it out.
What you can do: Address the small gaps: publish prices, response times and replies to mild complaints.
evidence.assessment · also called low_risk_signals, minor_risk_indicators
contradictionThe model notes that sources disagree about a business, such as two different phone numbers or hours, and treats the conflicting fact as uncertain.
What you can do: Find and fix conflicting phones, addresses and hours across your site and listings.
claims.verification (was evidence.assessment in the provisional inventory: Sources disagree about a fact) · also called inconsistency, discrepancy · same label written as contradictions, contradictions_found
Checked by Field-by-field audit of your listings · Term: nap-consistency
risk_handling_ruleThe model writes down how it will treat risky or uncertain candidates, such as "exclude businesses with no license shown" or "mention but caveat ones with mixed reviews".
selection.policy (was evidence.assessment in the provisional inventory: Rule for treating risky candidates in the shortlist) · also called risk_management_rule, risk_mitigation_strategy
community_weight_sourcesThe model marks sources that reflect local opinion, such as Nextdoor threads, Reddit posts or neighborhood Facebook groups, which carry weight for local-service recommendations.
What you can do: Encourage satisfied customers to mention you in local forums, and keep your business name consistent so mentions are attributed correctly.
evidence.provenance (was evidence.assessment in the provisional inventory: Classifies community sources by origin) · also called community_sources, peer_valued_sources
current_signalsThe model weighs evidence that is recent, such as reviews from the last few months or a site updated this year, as showing the business is active now.
What you can do: Keep your site and Business Profile current: recent photos, updated hours and seasonal services, and new reviews.
evidence.freshness (was evidence.assessment in the provisional inventory: About recency of evidence) · also called up-to-date_signals, recent_indicators
Checked by When your page's content last really changed · Term: freshness-boosting
stable_signalsThe model weighs evidence that has stayed consistent over time, such as the same address and steady review ratings for years, as a sign of a reliable business.
What you can do: Avoid needless changes to your business name, phone and address, and keep them the same everywhere.
evidence.freshness (was evidence.assessment in the provisional inventory: About evidence stability over time) · also called consistent_signals, reliable_indicators
Checked by Field-by-field audit of your listings
stale_or_unknown_signalsThe model flags evidence that is old or undated, such as reviews from five years ago or a site with a 2019 copyright, and trusts it less.
What you can do: Update your site's main content and dates, and remove or update stale listings and old promotions.
evidence.freshness · also called obsolete_signals, uncertain_indicators · same label written as stale_or_unknown
Checked by When your page's content last really changed
deeply_localThe model judges a business as rooted in a specific neighbourhood or town, such as a pet store known in one district, which suits queries that name a precise place.
What you can do: Name the neighbourhoods and towns you serve on your site, and mention local landmarks or community ties.
query.geography (was evidence.assessment in the provisional inventory: How locally rooted a business is) · also called hyper_local, very_local
Checked by The places your site names against your Business Profile
locally_presentThe model confirms that a business actually operates in the area asked about, through an address, service-area listing or local reviews, rather than only claiming to serve it.
What you can do: Show a local address or named service areas that match your Business Profile and listings.
candidate.eligibility (was evidence.assessment in the provisional inventory: Screens whether business operates in the area) · also called local_availability, present_locally
Checked by The places your site names against your Business Profile · Term: service-area
regional_brandThe model recognises a business as a known name across a region, such as a landscaping company with several branches in one state, distinct from a single-shop local or a national chain.
What you can do: Describe your branches and the region you cover on one page, and keep each location's details consistent.
entity.identity (was evidence.assessment in the provisional inventory: Regional brand versus local shop or chain) · also called regional_identity, local_brand
Checked by Your directory listings, found and matched
weak_local_fitThe model judges that a candidate fits the place poorly, such as a roofer based two counties away or with no local reviews, and ranks it lower.
What you can do: If you serve the area, say so explicitly with named towns, and gather reviews from customers there.
query.geography (was evidence.assessment in the provisional inventory: Poor fit to the requested place) · also called low_local_relevance, weak_local_connection
Checked by The cities your customers name in reviews
value_fitThe model judges how well a business fits a person looking for value for money, such as a handyman with fair, published rates and good reviews.
What you can do: Publish prices and what they include, and highlight reviews that mention fair pricing.
offer.fit · also called value_alignment, value_relevance · same label written as costo_beneficio_fit
Checked by The claims your site makes: years, prices, refund window
budget_fitThe model judges whether a business fits the person's stated budget, such as "cheap" or "under $500", using prices it can find.
What you can do: Publish prices or ranges on your site, with starting prices for common jobs.
offer.fit · also called cost_fit, financial_alignment · same label written as economic_fit
Checked by The claims your site makes: years, prices, refund window
communityThe model notes signs of community involvement or local reputation, such as sponsoring a youth team or being recommended in a neighbourhood group.
What you can do: Mention your local involvement, sponsorships and community ties on your About page.
reputation.evidence (was evidence.assessment in the provisional inventory: Local reputation and community recommendation) · also called local_community, community_aspect
Used in 1 of 100 census runs this week: “1-5 based on sentiment analysis of forum mentions”
emergency_fitThe model judges how well a candidate suits an urgent need, such as a 24/7 plumber for a burst pipe, based on stated hours, response time and emergency service.
What you can do: State emergency availability, after-hours phone and typical response time, and keep the same hours on all listings.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate handles urgent jobs) · also called crisis_fit, urgent_fit · same label written as emergencia_fit
Checked by Your hours, day by day, everywhere they are printed
corporate_fitThe model judges how well a candidate suits a business client, such as a caterer for office events or a commercial cleaner, rather than a private household.
What you can do: If you serve businesses, create a commercial services page with typical clients, invoicing and contract options.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate serves business clients) · also called business_fit, corporate_alignment
family_fitThe model judges how well a candidate suits families, such as a dentist that sees children or a restaurant with a kids' menu.
What you can do: State family-relevant facts plainly: ages served, kids' options, accessibility and parking.
offer.fit (was evidence.assessment in the provisional inventory: Family amenities and options) · also called family_relevance, family_alignment
verifiable_nowThe model marks facts it can confirm right now from available sources, such as current hours or a live booking page, rather than ones the person would need to check.
What you can do: Keep hours, availability and prices live and accurate on your site and Business Profile.
claims.verification · also called immediate_verification, instant_check
Checked by Your hours, day by day, everywhere they are printed
community_fitThe model judges how well a candidate fits what the local community values, such as a locally owned bakery for someone who wants to support local businesses.
What you can do: Say you are locally owned if true, and describe your community ties and local customers.
offer.fit (was evidence.assessment in the provisional inventory: Match on values like local ownership) · also called social_alignment, community_alignment
eco_friendly_fitThe model judges how well a candidate fits a person seeking eco-friendly service, such as a cleaner using non-toxic products or a solar installer.
What you can do: Describe your eco practices specifically, such as products used and certifications held, rather than just saying "green".
offer.fit (was evidence.assessment in the provisional inventory: Match on eco practices of the offer) · also called sustainability_fit, green_fit
Checked by The claims your site makes: years, prices, refund window
high_end_fitThe model judges how well a candidate fits a high-end request, such as a custom home builder or luxury pool designer, based on portfolio, pricing and clientele.
What you can do: Show high-end projects, materials and typical budgets in a portfolio on your site.
offer.fit (was evidence.assessment in the provisional inventory: Match on high-end tier) · also called luxury_alignment, premium_quality · same label written as premium_fit
luxury_fitThe model judges how well a candidate fits a request for exclusivity or prestige, such as a private chef or luxury car service, beyond plain quality.
What you can do: Describe exclusive features such as private service, limited bookings or premium materials, with examples.
offer.fit · also called exclusive_fit, high_status_fit
overall_quality_fitThe model gives an overall judgement of how good a candidate is, combining reviews, credentials and evidence of work, regardless of price.
What you can do: Gather reviews, show credentials and publish examples of completed work on your site.
evidence.assessment · also called quality_assessment, quality_alignment
Used in 2 of 100 census runs this week: “Average rating ≥ 4.0 stars on major review platforms, Minimum 30 patient reviews, Positive remarks about clean facilities, gentle staff, and”
pet_friendly_fitThe model judges how well a candidate suits someone with pets, such as a pet-friendly hotel or a carpet cleaner using pet-safe products.
What you can do: State pet policies or pet-safe practices plainly on your site and listings.
offer.fit (was evidence.assessment in the provisional inventory: Pet-friendly amenities or products) · also called pet_compatibility, animal_friendly_fit
risk_avoidance_fitThe model judges how well a candidate suits a person who wants the safest choice, favouring licensed, insured businesses with long track records and few complaints.
What you can do: Show license, insurance, years in business and guarantees prominently on your site.
evidence.assessment · also called risk_management_fit, safety_fit
Checked by The claims your site makes: years, prices, refund window
verifiableThe model judges whether a claim about a business can be confirmed at all from sources it can reach, such as a stated license or founding year.
What you can do: Support claims with evidence the model can reach, such as registry links and consistent structured data.
claims.verification · also called checkable, confirmable
Checked by The claims your site makes: years, prices, refund window
wedding_fitThe model judges how well a candidate suits a wedding, such as a florist, venue or photographer with wedding portfolios and packages.
What you can do: If you do weddings, have a dedicated page with packages, prices and a portfolio of weddings.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate does weddings) · also called bridal_fit, nuptial_fit
airport_transfer_fitThe model judges how well a candidate suits airport transfers, such as a car service that covers the named airport, offers flight tracking and lists fares.
What you can do: State which airports you serve, fares or rates and booking options on a dedicated page.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate does airport transfers) · also called transport_fit, travel_transfer_fit
Checked by The places your site names against your Business Profile
bali_inspired_fitThe model judges how well a candidate matches a Bali-inspired style request, such as a landscaper or spa offering tropical or Balinese design.
What you can do: If you offer this style, name it and show examples on your site.
offer.fit (was evidence.assessment in the provisional inventory: Match on design style) · also called tropical_fit, island_style_fit
boutique_fitThe model judges how well a candidate fits a request for a small, specialised business, such as a boutique hotel or independent bridal shop, rather than a chain.
What you can do: Describe what makes you small and specialised, such as owner-run service or curated stock.
offer.fit (was evidence.assessment in the provisional inventory: Match on boutique style of offer) · also called specialty_fit, unique_fit
buyer_fitThe model judges how well a candidate fits a person buying, for example a real estate agent who represents buyers rather than sellers.
What you can do: State plainly which client side you serve, with examples, on your service page.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate serves the buyer side) · also called customer_fit, client_fit
compliance_fitThe model judges whether a candidate meets required regulations or standards, such as a licensed electrician or a daycare with state certification.
What you can do: List licenses, permits and certifications with numbers and issuing bodies on your site.
credentials.status (was evidence.assessment in the provisional inventory: About licences and certifications) · also called regulatory_fit, standards_fit
conditionally_verifiableThe model marks a claim as confirmable only under some condition, such as a price that holds only for certain sizes or a license that must be checked in a state registry.
What you can do: State the conditions behind your claims plainly, such as which jobs a price applies to.
claims.verification · also called conditional_check, partial_verification · same label written as verifiable_with_limits
Checked by The claims your site makes: years, prices, refund window
consulta_detallada_fitThe model judges how well a candidate suits a person wanting a detailed consultation, such as a lawyer or designer offering in-depth first meetings.
What you can do: Describe your consultation process, length and cost on your site.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate offers detailed consultations) · also called detailed_consult_fit, in-depth_fit
deep_steam_fitThe model judges how well a candidate suits a request for deep steam cleaning, such as a carpet cleaner using hot-water extraction equipment.
What you can do: Name your cleaning methods and equipment specifically on your service pages.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate does deep steam cleaning) · also called intensive_clean_fit, steam_clean_fit
direct_matchThe model marks a candidate as matching every stated criterion, such as a licensed, local, 24/7 locksmith for a lockout query.
What you can do: Make sure your key facts, services, area and hours are stated plainly so the model can confirm a full match.
service.fit (was evidence.assessment in the provisional inventory: Candidate matches all requested criteria) · also called exact_fit, perfect_match
easy_to_verifyThe model rates a fact as easy to confirm, such as a phone number shown the same on the site and every listing.
What you can do: Keep your facts consistent across your site and listings so they are easy to confirm.
claims.verification (was evidence.assessment in the provisional inventory: Ease of confirming a fact across sources) · also called simple_verification, quick_check
Checked by Field-by-field audit of your listings
eco_low_water_fast_dry_fitThe model judges how well a candidate suits a person wanting low-water, fast-drying cleaning, such as a carpet cleaner that promises dry carpets in a few hours.
What you can do: State drying times and cleaning methods specifically on your site.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate offers low-water fast-dry cleaning) · also called water_efficient_fit, quick_dry_fit
emergency_fix_fitThe model judges how well a candidate suits an urgent repair, such as fixing a broken garage door the same day.
What you can do: State same-day repair availability and response times on your site.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate does urgent repairs) · also called urgent_repair_fit, quick_fix_fit
Checked by Your hours, day by day, everywhere they are printed
event_fitThe model judges how well a candidate suits a specific event, such as a caterer for a corporate party or a DJ for a birthday.
What you can do: List the events you handle, with examples and packages.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate handles the event) · also called occasion_fit, event_compatibility
event_planning_fitThe model judges how well a candidate suits event planning needs, such as a planner who handles vendors, timelines and venues.
What you can do: Describe your planning services and what is included, with examples.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate offers event planning) · also called planning_fit, event_management_fit
exceptionThe model notes a case that breaks its usual criteria, such as a business outside the area but known as the only specialist, and decides whether to keep it.
selection.policy (was evidence.assessment in the provisional inventory: Decides keeping exceptions to shortlist criteria) · also called anomaly, outlier
excluded_or_deprioritized_entity_typesThe model lists kinds of result it will drop or rank lower, such as directories or national chains for a local pet groomer query.
What you can do: Make clear you are the business itself, with your own address, phone and services.
candidate.eligibility (was query.scope in the provisional inventory: Screens out or deprioritizes kinds of result) · also called deprioritized_types, excluded_entities
exclusive_fitThe model judges that a candidate is uniquely suited to a specific need, such as the only local shop that repairs antique clocks.
What you can do: Name your specialities explicitly, especially rare ones, on your site.
service.fit (was evidence.assessment in the provisional inventory: Unique ability to do the specific job) · also called unique_fit, specific_fit
executive_fitThe model judges how well a candidate suits executive or senior clients, such as an executive coach or a car service for business travelers.
What you can do: If you serve executives, describe that service and typical clients.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate serves executive clients) · also called strategic_fit, leadership_fit
future_availability_fitThe model judges whether a candidate can take the job at the time the person needs, such as a contractor booking for next spring.
What you can do: State your booking lead time and availability on your site.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate can take the job when needed) · also called future_access_fit, availability_fit
hardThe model marks a requirement or case as hard to satisfy, such as finding a same-day appliance repair on a holiday.
evidence.assessment · also called challenging, difficult
hard_to_verifyThe model rates a claim as hard to confirm, such as an award with no link or a license number that cannot be found.
What you can do: Link claims to their sources, such as award pages and license lookups.
claims.verification (was evidence.assessment in the provisional inventory: Difficulty confirming a claim) · also called difficult_to_confirm, verification_challenge
Checked by The claims your site makes: years, prices, refund window
independent_fitThe model judges how well a candidate suits someone wanting an independent business, such as a locally owned hardware store rather than a chain.
What you can do: State that you are independently owned, if true.
entity.identity (was evidence.assessment in the provisional inventory: Independent versus chain ownership) · also called self_sufficient_fit, autonomous_fit
large_home_or_tile_combo_fitThe model judges how well a candidate suits a large home or a mixed carpet-and-tile job, such as a cleaner who handles both in one visit.
What you can do: Say which home sizes and surfaces you handle, and offer combined packages.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate handles large or mixed-surface jobs) · also called large_space_fit, combo_fit
local_fitThe model judges how well a candidate fits the place asked about, such as a roofer based in the named town with local reviews.
What you can do: Name the towns you serve, matching your Business Profile.
query.geography (was evidence.assessment in the provisional inventory: Fit to the requested place) · also called regional_fit, contextual_fit
Checked by The places your site names against your Business Profile · Term: service-area
local_trusted_fitThe model judges whether a candidate is both local and trusted, such as a long-established neighbourhood electrician with strong local reviews.
What you can do: Show years in your area, local reviews and community ties.
reputation.evidence (was evidence.assessment in the provisional inventory: Local trust and reviews) · also called trusted_local_fit, community_fit
Checked by The cities your customers name in reviews
mismatchThe model notes that a candidate does not fit what was asked, such as a residential-only cleaner for an office cleaning query.
What you can do: State what you do and do not do, so the model does not mismatch you.
service.fit (was evidence.assessment in the provisional inventory: Candidate does not do the job asked) · also called discrepancy, incompatibility
modern_style_fitThe model judges how well a candidate fits a request for modern style, such as an interior designer with contemporary portfolios.
What you can do: Show recent projects in modern styles with clear labels.
offer.fit (was evidence.assessment in the provisional inventory: Match on design style) · also called contemporary_fit, current_style_fit
Used in 1 of 100 census runs this week: “0.15”
needs_user_verificationThe model marks a fact the person should confirm themselves, such as current prices or availability, before relying on the recommendation.
What you can do: Publish current facts plainly so the model needs fewer caveats.
action.handoff (was claims.verification in the provisional inventory: Tells person what to confirm themselves) · also called user_confirmation_needed, verification_required
Term: hedged-answer
non_verifiable_superlativesThe model flags claims like "best in town" or "#1 roofer" that no source backs up, and discounts them.
What you can do: Replace superlatives with evidence: awards with links, review counts, years in business.
claims.verification · also called unverifiable_claims, unsupported_superlatives
Checked by The claims your site makes: years, prices, refund window
not_acceptableThe model marks a candidate as failing a hard requirement, such as no license where one is required, and drops it.
What you can do: Make sure required licenses and insurance are shown.
candidate.eligibility (was evidence.assessment in the provisional inventory: Fails a hard requirement and is dropped) · also called unacceptable, non_compliant
not_directly_verifiableThe model rates a claim as confirmable only indirectly, such as through reviews mentioning a service rather than the business stating it.
What you can do: State services and facts directly on your site.
claims.verification · also called indirect_verification, verification_issue
Checked by The claims your site makes: years, prices, refund window
partialThe model marks a candidate or answer as only partly meeting needs or partly supported, such as a business with hours but no services listed.
What you can do: Fill gaps on your site and listings: services, hours, area, prices.
evidence.assessment · also called incomplete, fragmentary
partial_matchThe model marks a candidate as matching some criteria but not all, such as a local plumber without 24/7 service for an emergency query.
What you can do: State all your key facts so the model can see a full match where there is one.
service.fit (was evidence.assessment in the provisional inventory: Candidate matches only some criteria) · also called incomplete_fit, near_match
pet_odor_fitThe model judges how well a candidate suits pet odor problems, such as a carpet cleaner offering enzyme treatments.
What you can do: Describe pet odor treatments specifically on your site.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate treats pet odor) · also called pet_smell_fit, odor_control_fit
planned_policy_fitThe model judges how well a candidate suits someone planning a policy purchase, such as an insurance agent for a future home policy.
What you can do: Describe the policies you offer and how quotes work.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate offers the policy) · also called policy_alignment_fit, policy_compatibility
planned_project_fitThe model judges how well a candidate suits a planned project, such as a contractor for a kitchen remodel next spring.
What you can do: Describe your project types, timelines and booking lead times.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate does the planned project) · also called project_alignment_fit, future_project_fit
private_party_fitThe model judges how well a candidate suits a private party, such as a caterer or venue for a small private event.
What you can do: List private events you handle, with sizes and packages.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate handles private parties) · also called private_event_fit, exclusive_fit
project_fitThe model judges how well a candidate fits the specific project described, such as a fence company for a 200-foot cedar fence.
What you can do: Describe project types and sizes you handle, with examples.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate fits the described project) · also called project_compatibility, task_fit · same label written as projet_fit
renter_fitThe model judges how well a candidate suits a renter, such as a mover for apartments or a cleaner offering move-out cleaning.
What you can do: Mention renter-friendly services such as move-out cleaning or landlord-approved work.
service.fit (was evidence.assessment in the provisional inventory: Whether candidate serves renters) · also called tenant_fit, lease_fit
reputacion_y_verificacion_fitThe Spanish-language label for reputation and verification fit: the model judges a candidate's reputation and how well its claims can be checked.
What you can do: Gather reviews and support claims with checkable evidence, such as license numbers.
reputation.evidence · also called reputation_fit, credibility_fit
Checked by The claims your site makes: years, prices, refund window
seller_fitHow well a business fits a query from someone selling, not buying, such as a homeowner asking which realtor or consignment shop to list with. The model weighs listing terms, commission and seller-side services.
What you can do: If you serve sellers, give them their own page: fees or commission, what you handle, typical time to sale.
offer.fit (was evidence.assessment in the provisional inventory: Judges listing terms and commission, an offer match.) · also called vendor_fit, merchant_fit
standard_fitWhether a candidate suits an ordinary version of the request, with no special need such as emergency, luxury or budget. A tree service doing routine trimming at normal prices scores well here.
What you can do: Describe your everyday core services plainly, with typical price ranges and turnaround, not only premium or emergency offers.
service.fit (was evidence.assessment in the provisional inventory: About doing the ordinary job asked, not evidence risk.) · also called typical_fit, baseline_fit
sustainability_fitHow well a business matches a query asking for an eco-friendly option, such as green cleaning or solar-ready roofing. The model looks for named practices, products or certifications rather than vague green wording.
What you can do: Name the specific products, methods and certifications you use (e.g. EPA Safer Choice cleaners) on a dedicated page.
offer.fit (was evidence.assessment in the provisional inventory: Eco practices are an offer attribute matching the query.) · also called eco_fit, green_fit
transparent_pricing_fitWhether a candidate publishes prices or clear ranges the model can quote, such as a junk removal company listing cost per truckload. Businesses with "call for quote" only score lower when the user wants upfront pricing.
What you can do: Publish prices or honest ranges for common jobs, what is included and what changes the price.
offer.fit · also called clear_pricing, open_cost
Checked by The claims your site makes: years, prices, refund window · Term: offer
urgence_fitA label (French urgence, urgency) for how well a candidate fits an urgent request, such as a locksmith for a lockout tonight. The model checks for stated response times and after-hours availability.
What you can do: State your emergency availability, hours and typical response time on the site and in your Business Profile.
service.fit (was evidence.assessment in the provisional inventory: Checks response times and availability for the urgent job.) · also called priority_fit, time_sensitive
Checked by Your hours, day by day, everywhere they are printed
urgent_same_day_fitWhether a candidate can plausibly do the job the same day, as for "same day water heater replacement". The model looks for explicit same-day promises, hours and stocked inventory rather than generic "fast service".
What you can do: If you offer same-day service, say so plainly with cut-off times, service area and any extra fee.
service.fit (was evidence.assessment in the provisional inventory: Whether the candidate can do the job same day.) · also called immediate_fit, same_day_response
Checked by Your hours, day by day, everywhere they are printed
vintage_restore_fitHow well a candidate fits a request to restore something old, such as a vintage car, antique furniture or a historic window. The model looks for named restoration work, era expertise and examples.
What you can do: Show restoration projects with before-and-after photos, the era or make, and methods used.
service.fit (was evidence.assessment in the provisional inventory: About offering restoration work, a service match.) · also called heritage_fit, restoration_fit