AI classifier atlas · stage 3 of 16
What the person wants, read from their words: the job, the urgency, the budget, the trade-offs they are likely to make.
141 labels, plus 36 spellings merged into them. Field names gpt-oss writes when asked to describe its evaluation; not OpenAI internals.
required_answer_behaviorWhat kind of answer the model decides the query needs, such as a direct recommendation, a shortlist with caveats or advice on how to choose, set during query interpretation.
answer.presentation (was query.intent in the provisional inventory: Decides the form the answer takes) · also called response_type, answer_format
financial_riskHow much money the model thinks the asker could lose with a bad choice, such as a roof replacement versus a nail trim. Higher stakes make it stricter about proof of reliability.
What you can do: For costly work, state warranties, written estimates and payment terms on the site.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about financial stakes) · also called financial_concern, economic_risk
geo_classThe kind of location the query names, such as a specific city, a neighbourhood, "near me" or a whole region, which sets how tightly the model filters businesses by place.
What you can do: State your city and service areas clearly and consistently on the site and listings.
query.geography · also called location_type, geographical_category
Checked by The places your site names against your Business Profile
quality_modifierQuality words in the query, such as "best", "top-rated" or "cheap", that the model reads to decide how strongly to weigh reviews, credentials or price.
What you can do: Back quality claims with evidence the model can check: review counts, awards and credentials with sources.
query.intent · also called quality_indicator, quality_level
Used in 2 of 100 census runs this week: “average rating >=4.0, number of reviews >=10, positive comments on professionalism, safety, and workmanship”
likely_decision_stageThe model's guess at where the asker is in choosing: just researching, comparing options or ready to book. For "roofer near me open now" it assumes ready to hire.
What you can do: Offer content for each stage: guides for researchers, comparisons and prices for shoppers, and a clear booking path for those ready to hire.
query.intent · also called decision_phase, decision_making_stage · same label written as decision_stage
trust_riskHow much the model thinks trust matters for the query, such as hiring someone to enter your home or care for a pet. Higher trust risk raises the weight on reviews, licences and background checks.
What you can do: Show background checks, licences, insurance and verifiable reviews on the site.
evidence.assessment (was query.intent in the provisional inventory: trust_risk is a named evidence.assessment example) · also called trust_concern, trust_level
urgency_unknownA flag that the query does not say how urgent the need is, so the model may cover both fast-response and scheduled options, such as for "furnace repair Denver".
What you can do: State response times for urgent and scheduled work separately on your site.
query.intent · also called urgency_status, urgency_level
Checked by Your hours, day by day, everywhere they are printed
decision_reversibilityWhether the model thinks a bad choice can be undone cheaply, such as switching dog walkers, or not, such as a botched roof. Hard-to-reverse choices make it more cautious.
What you can do: For permanent work, state guarantees, warranties and how you handle problems after the job.
evidence.assessment (was query.intent in the provisional inventory: Judges stakes of a bad choice) · also called reversibility_status, decision_flexibility · same label written as reversibility, decision_reversibility_risk
scopeThe breadth the model gives the query, such as one specific service in one town versus a general category across a region, used to decide how many and which businesses fit.
query.scope · also called context_scope, query_range
subjectThe main thing the query is about, as the model identifies it, such as "tree removal" in "who does safe tree removal in Portland".
What you can do: Make each core service the clear subject of its own page, with the service name in the title and main heading.
query.intent · also called main_topic, query_subject
qualifierAn extra condition in the query that narrows the choice, such as "licensed", "24/7" or "pet-friendly", which the model checks candidates against.
What you can do: State each condition you meet explicitly on the site, e.g. "licensed and insured", "open 24/7", in plain text.
query.intent · also called condition, specification · same label written as qualifiers
modifiersWords in the query that change what kind of business fits, such as "cheap", "eco-friendly", "family-owned" or "near me", picked out during interpretation.
What you can do: Use the descriptors that truly apply to you, such as eco-friendly or family-owned, in plain text on the site with supporting detail.
query.intent · also called alterations, descriptors · same label written as modifiers_detected
Used in 4 of 100 census runs this week: “Spanish”
safety_riskHow much physical danger the model sees in the job, such as electrical, gas or tree felling work. Higher safety risk makes it favour licensed, insured and experienced businesses.
What you can do: Show licences, insurance, safety certifications and training on the site, with numbers that can be checked.
evidence.assessment (was query.intent in the provisional inventory: safety_risk is a named evidence.assessment example) · also called safety_concern, safety_level · same label written as safety_health_risk
financialMoney-related aspects the model notes in the query, such as a stated budget, financing needs or cost worries, which affect which businesses it suggests.
What you can do: Publish prices or ranges, payment options and any financing you offer.
query.intent · also called economic_factor, financial_aspect
Checked by The claims your site makes: years, prices, refund window
safetySafety considerations the model picks up in the query, such as a leaning tree near a house or exposed wiring, that steer it toward qualified, insured providers.
What you can do: State safety credentials, insurance and how you handle hazardous jobs on the relevant service pages.
query.context (was query.intent in the provisional inventory: Safety circumstances described in the query) · also called safety_concern, safety_issue
timeTime details the model reads from the query, such as "tonight", "this weekend" or "before winter", which affect how much weight it gives availability.
What you can do: Publish accurate hours and availability, including weekends and emergency service, and keep them consistent everywhere.
query.intent · also called temporal_factor, time_aspect
Checked by Your hours, day by day, everywhere they are printed
trustHow much the query depends on trusting the provider, such as letting a cleaner into the house, which the model uses to weight reviews and credentials.
What you can do: Show reviews, background checks and credentials that can be verified on the site.
evidence.assessment (was query.intent in the provisional inventory: Trust-risk judgment, same as trust_risk) · also called trust_level, trust_factor
Used in 1 of 100 census runs this week: “0.15”
time_riskThe risk that delay makes things worse or a deadline is missed, such as a leaking roof before a storm, which leads the model to favour businesses that can act fast.
What you can do: State response times and same-day availability if you offer them, and keep hours current on all listings.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about delay) · also called temporal_risk, time_concern · same label written as timeline_risk
Checked by Your hours, day by day, everywhere they are printed
budget_sensitivityHow much cost matters to the asker, as the model reads it from words like "affordable" or "cheap", which shifts weight toward businesses with clear, lower prices.
What you can do: Publish prices or price ranges and any discounts or free estimates in plain text on the site.
query.intent (was offer.fit in the provisional inventory: Asker budget preference, not candidate offer match) · also called budget_concern, cost_sensitivity · same label written as price_sensitivity
Checked by The claims your site makes: years, prices, refund window
Used in 1 of 100 census runs this week: “0.1”
emergency_vs_plannedWhether the model reads the need as an emergency or a planned job, such as a burst pipe versus a bathroom remodel. Emergencies favour availability; planned jobs favour quality and price.
What you can do: Make emergency availability and planned-project services both clear, on separate pages if you offer both.
query.intent · also called emergency_status, planning_type · same label written as emergency_vs_project, emergency_vs_routine, urgencia_vs_planificacion
Checked by Your hours, day by day, everywhere they are printed
reputation_riskHow much the model thinks a poor choice would reflect on the asker, such as a caterer for a corporate event, which raises the weight on reviews and track record.
What you can do: Show testimonials and examples from similar clients and events, and respond to reviews publicly.
evidence.assessment (was reputation.evidence in the provisional inventory: Risk judgment, not review evidence itself) · also called reputational_concern, reputation_level · same label written as reputational_risk, reputacion_risk
property_damage_riskThe chance the job could damage the asker's property, such as tree removal near a roof or a botched plumbing fix, which makes the model favour insured, experienced businesses.
What you can do: State liability insurance and how damage claims are handled on the site.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about property damage) · also called damage_risk, property_risk · same label written as property_risk
urgency_riskThe risk that the need cannot wait, as the model judges it, such as no heat in winter. High urgency risk pushes businesses with fast response ahead.
What you can do: State emergency availability and response times, and keep hours accurate across listings.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about urgency) · also called urgency_level, urgency_concern
Checked by Your hours, day by day, everywhere they are printed
residential_vs_commercialWhether the model reads the job as for a home or a business, such as a homeowner's roof versus a warehouse roof, which changes which providers fit.
What you can do: State whether you serve residential, commercial or both, with examples of each.
query.context (was query.intent in the provisional inventory: Property type background of the request) · also called property_type, context_type · same label written as mode_residential_vs_commercial
service_quality_riskThe risk that the work will be poor, as the model weighs it for the query, which shifts weight toward reviews mentioning workmanship, warranties and portfolio evidence.
What you can do: Show photos of finished work, warranties and reviews that describe the work itself.
evidence.assessment (was service.fit in the provisional inventory: Risk judgment about work quality) · also called quality_risk, service_quality_concern · same label written as quality_risk
budget_vs_luxuryWhether the model reads the asker as wanting a budget option or a high-end one, such as a basic lawn service versus a luxury landscape designer.
What you can do: Make your price tier clear on the site so the model can place you as budget, mid-range or premium.
query.intent (was offer.fit in the provisional inventory: Asker preference for budget versus high-end) · also called cost_vs_quality, luxury_concern · same label written as luxury_vs_budget
Checked by The claims your site makes: years, prices, refund window
event_type_ambiguityA flag that the query does not make clear what kind of event it is for, such as "party catering" that could be a kids' birthday or a corporate function, so the model may cover several.
What you can do: List the event types you serve on the site, each with its own description.
query.intent · also called event_confusion, event_type_uncertainty
operational_riskThe risk that the provider cannot run the job smoothly, such as staffing gaps or equipment problems, which the model weighs when judging reliability.
What you can do: Describe team size, equipment and how you handle scheduling or crew issues.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about provider operations) · also called operational_threat, operational_hazard
pricing_vs_qualityHow the model weighs price against quality for the query, such as a cheaper fence company versus one with stronger reviews and warranties.
What you can do: Publish prices or ranges next to what is included and the warranty, so the trade-off can be judged from your site.
query.intent (was offer.fit in the provisional inventory: Price-quality preference, like budget_vs_quality) · also called cost_quality_tradeoff, value_assessment · same label written as affordability_vs_quality, budget_vs_quality, price_vs_quality
Checked by The claims your site makes: years, prices, refund window
Used in 4 of 100 census runs this week: “Average project cost or price range indicated as competitive; presence of free estimates; multiple positive mentions of reasonable pricing”
scheduling_riskThe risk that a provider cannot fit the asker's timing, such as a booked-up contractor before a move-in date, which the model weighs when recommending.
What you can do: State typical lead times and booking availability, and offer online booking if you can.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about scheduling) · also called time_management_risk, timing_conflict · same label written as schedule_risk, appointment_risk
urgencyHow soon the asker needs help, as the model reads it, such as "now" for a lockout versus "this summer" for a deck. Urgent needs favour available businesses.
What you can do: Keep hours accurate and state same-day or emergency service if you offer it.
query.intent · also called immediacy, time_sensitivity · same label written as urgency_indicator
Checked by Your hours, day by day, everywhere they are printed
Used in 8 of 100 census runs this week: “immediate”
event_typeThe kind of event the model reads from the query, such as a wedding, corporate dinner or kids' party, used to match caterers, venues or rental companies.
What you can do: List the event types you serve with examples and photos for each.
query.intent · also called event_category, type_of_event
family_vs_adventureWhether the model reads the asker as wanting a family-friendly option or an adventurous one, such as a quiet campground versus a rafting outfitter.
What you can do: State who your service suits, such as families with young kids or experienced adventurers, with specifics like age limits.
query.intent · also called family_or_adventure, recreation_preference
geographic_focusThe specific area the model treats as the query's focus, such as a downtown neighbourhood or a single suburb, used to filter businesses by location.
What you can do: Name the neighbourhoods and towns you serve on the site and in your Business Profile.
query.geography · also called location_focus, regional_interest
Checked by The places your site names against your Business Profile
geographyThe place the model ties the query to, such as a city, county or region, used to decide which businesses are local enough to consider.
What you can do: Keep your address and service areas the same on your site, Business Profile and directories.
query.geography · also called location, area
Checked by The places your site names against your Business Profile
Used in 51 of 100 census runs this week: “Austin, TX and surrounding metro area”
geography_precisionHow exact the model needs location to be, such as a precise neighbourhood for a dog walker versus a whole metro for a moving company.
What you can do: State your address and service area boundaries precisely, such as listing neighbourhoods or ZIP codes served.
query.geography · also called location_accuracy, geographic_detail
Checked by The places your site names against your Business Profile
health_riskHealth risks the model sees in the job, such as mould removal, pest treatment chemicals or pet medication, that steer it toward certified, careful providers.
What you can do: State certifications, products used and safety precautions on the relevant service pages.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about health hazards) · also called health_threat, wellness_risk
insuranceWhether insurance matters for the query, such as the provider's liability insurance or whether the asker's insurance covers the job, which the model weighs when choosing.
What you can do: State your liability insurance and whether you work with homeowners' insurance claims.
credentials.status · also called coverage, risk_protection
Used in 5 of 100 census runs this week: “serves Denver proper and surrounding metro area”
legal_riskThe risk of legal trouble from a poor choice, such as unpermitted electrical work or an unlicensed contractor, which makes the model favour licensed, permitted providers.
What you can do: Show licence numbers and say you pull required permits, so compliance can be checked.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment about legal exposure) · also called legal_threat, compliance_risk
local_vs_regionalWhether the model reads the need as served by a nearby local business or a regional provider, such as a neighbourhood groomer versus a statewide moving company.
What you can do: State plainly whether you are a local business or serve a wider region, with the areas listed.
query.geography (was query.intent in the provisional inventory: local_vs_regional is a named query.geography example) · also called local_or_regional, area_scope
Checked by The places your site names against your Business Profile
overdiagnosis_or_upsell_riskThe risk that a provider recommends unneeded work, such as a mechanic or HVAC company upselling a full replacement, which the model weighs using reviews and pricing transparency.
What you can do: Publish diagnostic fees, typical prices and a policy on second opinions or repair-first recommendations.
evidence.assessment (was query.intent in the provisional inventory: Named evidence.assessment example) · also called diagnosis_risk, upsell_concern
Checked by The claims your site makes: years, prices, refund window
product_categoryThe kind of product the query is about, such as dog food or garden tools, which the model uses to match stores that actually stock it.
What you can do: List the product categories and brands you carry on the site, in text rather than images.
service.fit (was query.intent in the provisional inventory: Matches stores that actually stock the product) · also called product_type, category
seasonal_availabilityWhether a service or product is only available in some seasons, such as snow removal or pool opening, which the model weighs against when the asker needs it.
What you can do: State seasonal hours and services clearly, and update them each season on the site and listings.
evidence.freshness (was query.intent in the provisional inventory: Seasonal availability is a named freshness example) · also called seasonal_factors, availability_timing
Checked by Your hours, day by day, everywhere they are printed
service_reliability_riskThe risk that a provider shows up late, cancels or does uneven work, which the model weighs using review patterns for a recurring service like lawn care.
What you can do: Show reviews mentioning reliability and state your scheduling and rescheduling policy.
evidence.assessment (was service.fit in the provisional inventory: Risk judgment about provider reliability) · also called service_consistency_risk, dependability_threat · same label written as service_failure_risk
style_preferenceThe style the asker wants, such as modern, rustic or traditional, which the model uses to match designers, landscapers or remodelers.
What you can do: Show portfolio photos with style labels, such as modern kitchens or rustic decks, on the site.
query.intent · also called design_preference, aesthetic_choice
Used in 2 of 100 census runs this week: “Helpful, neutral, and concise with actionable details.”
time_horizonThe timeframe the model reads for the need, such as this week or over the next year, which affects whether availability or long-term value weighs more.
What you can do: State lead times and how far ahead you book.
query.intent · also called time_frame, planning_horizon
accreditation_requirementCertifications or accreditations the model thinks the job needs, such as a licensed electrician or a certified arborist, which candidates must show to be recommended.
What you can do: List licences and certifications with numbers and issuing bodies, so they can be checked.
credentials.status (was query.intent in the provisional inventory: About required certifications and licences) · also called certification_need, accreditation_standards
Used in 3 of 100 census runs this week: “NATE, ACCA”
action_requiredWhether the asker needs to take an action now, such as booking or calling, which shapes whether the model gives a direct pick with contact details or general advice.
What you can do: Make phone, booking and contact options clear and consistent on the site.
action.handoff (was query.intent in the provisional inventory: What the person should do next) · also called required_action, response_needed
Checked by Name, address and phone in one block on your page
Used in 3 of 100 census runs this week: “encourage contacting shops for current quotes”
address_confusionA flag that a business's address is unclear or conflicting, such as different street addresses on the site and directories, which makes the model less sure the business is where it claims.
What you can do: Use one exact address everywhere and correct listings that show an old or wrong one.
presence.consistency (was entity.identity in the provisional inventory: Conflicting addresses across listings is NAP consistency) · also called location_confusion, address_clarity
Checked by Field-by-field audit of your listings · Term: nap-consistency
agent_independenceHow much the model can decide on its own versus deferring to the asker, such as picking one plumber or listing options and asking for preferences.
selection.policy (was query.intent in the provisional inventory: Whether to pick one or list options) · also called autonomy_level, independence_assessment
amenities_focusWhich features or amenities the asker emphasises, such as a fenced dog run at a kennel or Wi-Fi at an RV park, which the model checks candidates against.
What you can do: List amenities in plain text on the site, with photos and specifics.
offer.fit (was query.intent in the provisional inventory: Checks candidate amenities against needs) · also called features_priority, service_emphasis · same label written as amenities_importance
Used in 4 of 100 census runs this week: “Includes key wedding services (catering, ceremony space, parking, accommodations).”
assurance_incluseWhether guarantees or assurances are part of the offer, such as a workmanship warranty or satisfaction guarantee, which the model weighs as a sign of confidence.
What you can do: State guarantees and warranties with their terms in plain text on the site.
offer.fit (was query.intent in the provisional inventory: Guarantees are part of the candidate offer) · also called confidence_level, guarantee_inclusion
Checked by The claims your site makes: years, prices, refund window
beginner_learning_curveHow easy the model thinks a service or product will be for a beginner, such as a first-time dog owner choosing a trainer, used to favour beginner-friendly options.
What you can do: Say if you welcome beginners and explain what first-timers can expect.
service.fit (was query.intent in the provisional inventory: Whether the service suits a beginner) · also called learning_difficulty, user_familiarity
brand_vs_modelWhile reading the query, the model decides whether the person wants a brand or company ("a Trane dealer") or a specific product or model ("a Trane XR16 install"), which changes whether it shortlists businesses or product sellers.
What you can do: Name the brands you carry and the specific models you install or service on your service pages, so you match both brand-level and model-level queries.
query.intent · also called brand_identity, product_specificity
budget_levelThe model estimates how much the person expects to spend (cheap, mid-range, high-end) from words like "affordable" or "premium", then favours businesses whose stated pricing fits that range.
What you can do: Publish price ranges or starting prices on your site, for example "drain cleaning from $129", so the model can place you in a budget tier.
query.intent (was offer.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called financial_range, cost_level
Checked by The claims your site makes: years, prices, refund window
budget_unknownThe model notes that the query gives no budget signal, so it cannot rank by price and may hedge by listing options across price tiers.
What you can do: State what drives your price and give typical ranges, so the model can still describe you accurately when the person names no budget.
query.uncertainty (was offer.fit in the provisional inventory: Notes what the query leaves unknown) · also called financial_uncertainty, cost_unknown
building_typeThe model works out what kind of building the job involves (house, condo, apartment block, warehouse, historic home), because a gutter installer for a ranch house differs from one for a three-storey commercial building.
What you can do: List the building types you work on, such as single-family homes, multifamily, retail or historic properties, on your service pages.
query.context (was query.intent in the provisional inventory: Background situation of the request) · also called structure_category, building_classification
buyer_vs_sellerThe model decides whether the person is buying or selling, for example a home buyer looking for an agent versus an owner looking for a listing agent, and screens businesses by which side they serve.
What you can do: Say plainly whether you represent buyers, sellers or both, with a separate page for each if you serve both.
query.intent · also called transaction_roles, market_participants
comfort_riskThe model weighs how likely the person is to end up physically uncomfortable or unhappy with the result, such as a too-firm mattress, a cold house after a bad furnace install, or a cramped campsite.
What you can do: Describe comfort-related details concretely: return or adjustment policies, temperature guarantees, site sizes, or what follow-up visits are included.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called satisfaction_risk, user_comfort
commission_transparencyThe model judges whether a business that earns commissions or fees, such as a real estate agent, insurance broker or travel agent, makes those fees clear, and treats hidden fee structures as a trust risk.
What you can do: State your commission rate or fee structure on your site in plain numbers, including who pays it and when.
offer.fit (was query.intent in the provisional inventory: About candidate's fee and pricing transparency) · also called fee_clarity, commission_visibility
Checked by The claims your site makes: years, prices, refund window
Used in 1 of 100 census runs this week: “true”
complete_overhaul_vs_minor_updateThe model decides whether the person wants a full renovation or a small refresh, for example gutting a kitchen versus replacing cabinet doors, and matches them to businesses that take on that size of job.
What you can do: Show both large and small projects in your portfolio, and say whether you take small jobs or have a minimum project size.
query.intent · also called change_scope, update_type · same label written as full_remodel_vs_minor_update
compliance_riskThe model weighs whether the job carries legal or code exposure, such as permits for an electrical panel upgrade or licensing for asbestos removal, and prefers businesses that show they handle it.
What you can do: State your license numbers, which permits you pull, and the codes or regulations you work to, on the relevant service pages.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called regulatory_risk, legal_compliance · same label written as regulatory_risk
corporate_vs_socialThe model decides whether the event or service is for a business setting or a personal one, such as a caterer for a company offsite versus a birthday party, and filters vendors by that fit.
What you can do: Have separate pages or sections for corporate and private clients, with examples of each kind of event or job.
query.context (was query.intent in the provisional inventory: Background situation of the request) · also called business_vs_community, corporate_social
cosmetic_vs_generalThe model decides whether the request is about appearance or overall function, for example cosmetic dentistry versus general dental care, or a driveway resurfacing versus a full replacement.
What you can do: List cosmetic and general services separately and say which you specialise in, so the model can match the right kind of request.
query.intent · also called surface_vs_substantive, aesthetic_vs_functional
design_modifierThe model picks up words in the query that set a design requirement, such as "modern", "farmhouse" or "minimalist", and uses them to filter designers, remodelers or landscapers by style.
What you can do: Name the styles you work in on your site and label portfolio photos by style, for example "modern kitchen remodel".
query.intent · also called design_influence, design_factor
design_vs_costThe model judges whether the person cares more about design quality or keeping cost down, for example a custom landscape plan versus the cheapest sod install, and leans toward businesses that fit that balance.
What you can do: Offer and describe clear tiers, such as a design package and a basic package, with what each includes and costs.
query.intent · also called design_budget, cost_vs_aesthetics
design_vs_functionThe model decides whether looks or practical performance matter more in the request, for example a decorative fence versus a secure dog-proof fence, and weights candidates on that.
What you can do: Describe both the look and the practical performance of your work: materials, durability, and what problem each option solves.
query.intent · also called aesthetics_vs_utilitarian, design_functionality
design_vs_regulatoryThe model weighs where the person's design wishes may clash with rules, such as HOA limits on fence height or historic-district restrictions on windows, and favours businesses that show they handle those limits.
What you can do: Mention experience with HOA approvals, permits and historic-district rules, with examples of projects that met them.
query.context (was query.intent in the provisional inventory: Regulatory constraints surrounding the person's situation) · also called design_compliance, regulatory_constraints
documentation_requirementsThe model notes what paperwork the job needs, such as inspection reports, permits, insurance certificates or loan documents, and looks for businesses that say they provide or handle it.
What you can do: List the documents you provide or need, for example signed estimates, permit copies, certificates of insurance, or what to bring to an appointment.
query.scope (was query.intent in the provisional inventory: Paperwork the job requires defines its scope) · also called paperwork_needs, record_requirements
documentation_varianceThe model notes that paperwork requirements differ between providers, for example lenders asking for different income documents, and may explain those differences rather than assume one standard.
What you can do: Spell out exactly what documents you require, so the model can describe your requirements rather than a generic average.
comparison.context (was query.intent in the provisional inventory: Differences between providers) · also called documentation_differences, record_variability
eco_friendly_priorityThe model registers that the person wants an environmentally friendly option, such as organic lawn care or low-VOC paint, and moves businesses that state green practices up the shortlist.
What you can do: State specific green practices on your site: products used, certifications held, recycling or disposal methods. Avoid vague "eco" claims without detail.
query.intent · also called sustainability_focus, green_priority · same label written as sustainability_focus
eligibility_criteria_ambiguityThe model notes that it is unclear who qualifies, for example for a loan, a rebate program or a clinic's new-patient list, and may hedge or tell the person to confirm eligibility directly.
What you can do: Publish your eligibility rules plainly: minimum credit score, service area, insurance accepted, or age and patient requirements.
query.uncertainty (was query.intent in the provisional inventory: Notes what the query leaves unknown) · also called criteria_uncertainty, qualification_ambiguity
event_failure_riskThe model weighs how likely the event is to go wrong because of the vendor, such as a caterer not showing or a DJ's equipment failing, and favours vendors with evidence of reliability.
What you can do: Show a track record of events delivered, backup plans for equipment or staff, and reviews that mention reliability.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called event_risk, failure_probability · same label written as event_success_risk
event_reputation_riskThe model weighs how much a poor vendor would embarrass the host, for example at a wedding or a client-facing corporate dinner, and screens more strictly when the person's own reputation is at stake.
What you can do: Show high-profile or formal events you have handled, with client testimonials, and name any venues you regularly work with.
evidence.assessment (was reputation.evidence in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called reputation_risk, event_image
event_type_varianceThe model notes which kind of event the query is about, such as a wedding, a corporate retreat or a kid's birthday, since one vendor may suit some types and not others.
What you can do: List the event types you serve on separate pages or sections, with examples of each.
query.context (was query.intent in the provisional inventory: Event type is background of the request) · also called event diversity, event classification · same label written as event_type_variation
exclusivity_modifierThe model picks up words like "private", "VIP" or "members-only" in the query and favours businesses that show they offer exclusive or limited-access service.
What you can do: If you offer private bookings, buyouts or limited-availability service, state it explicitly with what it includes.
query.intent · also called exclusivity indicator, exclusive tag
facility_quality_ambiguityThe model notes that it cannot tell how good a facility is, such as a boarding kennel, gym or RV park, because the sources say little about its condition, and may hedge or tell the person to visit first.
What you can do: Post current photos and concrete facility details: size, equipment, cleaning schedule, recent renovations, and any inspection results.
evidence.assessment (was query.intent in the provisional inventory: Judging facility quality from thin evidence) · also called facility quality uncertainty, facility standard ambiguity
family_vs_budget_vs_petThe model weighs which of three priorities leads the request, family-friendliness, low cost or pet-friendliness, for example in choosing a campground or rental, and ranks options on the one that dominates.
What you can do: State your family amenities, prices and pet policy on the site, so you can be matched on whichever one the person cares most about.
query.intent (was offer.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called family budget pet, priority assessment
family_vs_pet_friendlyThe model decides whether the person needs a family-friendly or a pet-friendly option, such as a hotel, campground or restaurant patio, and filters out places that do not state that fit.
What you can do: Publish your pet policy (fees, size limits, rules) and your family amenities as clear, specific facts on the site.
query.intent · also called family or pet friendly, user preference distinction
financial_productThe model identifies which financial product the query is about, such as a mortgage, auto loan, small-business loan or insurance policy, and screens providers by whether they offer it.
What you can do: Give each product you offer its own page with rates, terms and eligibility, so the model can match you to that product.
service.fit (was query.intent in the provisional inventory: Identifies the product type to match providers) · also called financial service, financial offering
franchise_vs_localThe model decides whether the person prefers a national franchise or a locally owned business, for example a chain lawn service versus an owner-run one, and weights candidates by that ownership signal.
What you can do: State your ownership plainly: locally owned since a given year, or which franchise you belong to, on your About page and listings.
query.intent · also called franchise or local, business type distinction · same label written as independent_vs_chain
in_person_needThe model decides whether the job needs someone on site or in an office, such as a roof inspection, or can be done remotely, such as a virtual design consult, which changes how much location matters.
What you can do: Say which services are in person, which are remote or virtual, and where you meet clients.
query.intent · also called physical presence requirement, in-person necessity
Checked by The places your site names against your Business Profile
Used in 1 of 100 census runs this week: “fence installation/repair services”
independent_local_mode_probabilityThe model estimates how likely the person is to want an independent, locally owned business rather than a chain, from cues like "local" or "family-owned", and how much to weight that in ranking.
What you can do: If you are independent and local, say so plainly with your city and how long you have operated there.
query.intent · also called local preference likelihood, independent option probability
informational_vs_vendor_choiceThe model decides whether the person wants to learn something ("how much does a roof replacement cost") or pick a provider ("best roofer in Tulsa"), which decides whether it names businesses at all.
What you can do: Publish helpful explainer pages, such as cost guides, alongside clear service pages, so you can be cited in both kinds of answer.
query.intent · also called information vs choice, decision-making context
insurance_help_neededThe model notes the person may need help with an insurance claim, for example after storm damage to a roof or a car accident, and favours businesses that say they work with insurers.
What you can do: State whether you work with insurance claims, which insurers you deal with, and what help you give with paperwork and adjusters.
query.intent (was credentials.status in the provisional inventory: The person's need, not a credential check) · also called insurance assistance requirement, insurance support needed
interpretation_of_thidsThe model records how it read an ambiguous word or phrase in the query, so later stages search for the right meaning. The key name appears to be a garbled spelling in the model's own output.
What you can do: Use the common names customers use for your services alongside trade terms, so an ambiguous phrase still maps to you.
query.intent · also called term interpretation, concept understanding
Used in 7 of 100 census runs this week: “User wants a recommendation for a moving company in Austin, TX, likely for residential or small business relocation.”
itinerary_flexibilityThe model weighs how fixed the person's travel plans are, which matters when choosing tour operators, rentals or transfer services that differ on rebooking and cancellation.
What you can do: Publish your cancellation, rescheduling and change policies with exact time windows and fees.
query.intent · also called travel flexibility, itinerary adaptability
Checked by The claims your site makes: years, prices, refund window
lender_type_ambiguityThe model notes it is unclear which kind of lender fits, such as a bank, credit union or online lender, and may lay out the differences instead of naming one.
What you can do: State what kind of lender you are and what you offer that the other kinds do not, such as speed, rates or local decisions.
query.uncertainty (was query.intent in the provisional inventory: Notes what the query leaves unknown) · also called lender type uncertainty, lender classification ambiguity
local_verificationThe model flags facts it cannot confirm from sources and that the person should check locally, such as current hours, whether a shop still operates, or a license, before relying on the recommendation.
What you can do: Keep name, address, phone and hours identical on your site and every listing, so the model finds the facts confirmed rather than flagging them.
claims.verification · also called local confirmation, verification requirement
Checked by Field-by-field audit of your listings · Term: nap
Used in 1 of 100 census runs this week: “0.1”
locality_vs_specializationThe model weighs whether being nearby or being a specialist matters more, for example a nearby general vet versus an exotic-animal vet across town.
What you can do: Name your specialties clearly and the area you serve, so you can win on whichever factor the query weights.
query.intent (was service.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called local vs specialized, service focus distinction
location_distanceThe model weighs how much distance from the person matters, for example strongly for an emergency locksmith and less for a wedding photographer, and favours businesses it can place close by.
What you can do: Give a full address or the specific towns and neighbourhoods you serve, matching your Business Profile.
query.geography (was query.intent in the provisional inventory: About distance or locality of the request) · also called distance relevance, location proximity
Checked by The places your site names against your Business Profile
Used in 12 of 100 census runs this week: “Boise, ID”
location_flexibilityThe model judges how willing the person is to travel or accept a provider from outside their area, which decides how wide a radius it searches.
query.geography (was query.intent in the provisional inventory: About distance or locality of the request) · also called location adaptability, geographic flexibility
Checked by The places your site names against your Business Profile
luxury_levelThe model estimates how upscale an option the person wants, from cues like "high-end" or "boutique", and matches businesses whose pricing, materials and presentation fit that level.
What you can do: Describe your premium offerings concretely: materials, brands, service level and price ranges, rather than just calling yourself luxury.
query.intent (was offer.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called luxury standard, luxury classification
luxury_vs_valueThe model weighs whether the person wants the best experience regardless of price or the most for their money, for example a high-end spa versus a well-reviewed mid-priced one.
What you can do: Explain what is included at each price point, so the model can judge you on value as well as on quality.
query.intent (was offer.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called luxury versus value, value assessment
marketing_mode_probabilityThe model estimates how likely the query is about marketing services, such as signage, print or ad campaigns, rather than another reading, and how much to weight that reading when choosing vendors.
What you can do: If you sell marketing services, list each service with examples and the industries you serve.
query.intent · also called marketing strategy likelihood, marketing approach probability
medical_riskThe model weighs whether the request carries health risk, for example a med spa treatment, a dental procedure or pet surgery, and screens more strictly for licensed, credentialed providers.
What you can do: List licenses, board certifications and the credentials of the practitioners who do each procedure, with names.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called health risk assessment, medical safety consideration
menu_customization_ambiguityThe model notes it cannot tell whether a caterer or restaurant can adapt its menu, for example for allergies, vegan or kosher needs, and may hedge or tell the person to ask.
What you can do: State which dietary needs you accommodate and how custom menus work, with examples.
offer.fit (was query.intent in the provisional inventory: Whether candidate's menu offer can adapt) · also called menu flexibility uncertainty, customization ambiguity
method_preference_unknownThe model notes the person has not said which method they want, for example steam versus dry carpet cleaning, or traditional versus clear braces, so it may present options instead of one pick.
What you can do: Describe each method you offer and when it is the right choice, so you fit whichever method the person settles on.
query.uncertainty (was query.intent in the provisional inventory: Notes what the query leaves unknown) · also called method uncertainty, preference ambiguity
mobile_service_requirementThe model decides whether the person needs the business to come to them, such as mobile dog grooming, mobile auto glass or an on-site mechanic, and filters out businesses that only serve at a shop.
What you can do: State clearly that you offer mobile or on-site service and list the areas you travel to.
service.fit · also called mobile service necessity, mobile requirement
Checked by The places your site names against your Business Profile
multi_service_unknownThe model notes it is unclear whether the person needs one service or several, for example roof repair plus gutters, and may favour businesses that cover a wider range or say so.
What you can do: List all services you offer, including combined jobs and bundles, so a multi-service request still matches you.
query.uncertainty (was service.fit in the provisional inventory: Notes what the query leaves unknown) · also called service_uncertainty, multi_service_confusion
nib_size_standardThe model checks which standard size the query refers to, seen for pen nibs (fine, medium, broad), so it can recommend a shop that stocks that size.
What you can do: List the sizes and specifications you stock on product or category pages.
query.scope (was query.intent in the provisional inventory: Product specification bounding the request) · also called size_standard, nib_size
overall_vs_valueThe model decides whether the person wants the best overall option or the best value for money, which changes whether it picks a top-rated premium provider or a strong mid-priced one.
What you can do: Give prices or ranges and the evidence of quality (reviews, warranties) together, so you can be judged on either.
query.intent (was offer.fit in the provisional inventory: Person's stated preference, not candidate offer matching) · also called quality_vs_value, overall_quality
overnight_vs_extendedThe model decides whether the person needs a one-night or short service, such as overnight pet boarding, or a longer stay, and filters businesses by the durations they offer.
What you can do: State the stay lengths you offer and the rates for each, such as nightly, weekly and monthly.
query.intent · also called stay_duration, service_duration
partner_typeThe model identifies what kind of partner the person is looking for, such as a supplier, a subcontractor, an agency or a co-host, and screens businesses by whether they work in that role.
What you can do: Say who you work with and in what role, for example "we subcontract for general contractors".
query.intent · also called partnership_type, collaborator_type
patient_age_groupThe model notes the patient's age group, such as children, adults or seniors, and favours providers like pediatric dentists or geriatric care practices that state they serve that age.
What you can do: State the age groups you treat and any age-specific services, such as pediatric or senior care.
query.context (was query.intent in the provisional inventory: Patient age is background about the asker) · also called age_category, patient_demographics
pet_issue_unknownThe model notes it is unclear what problem the pet has, such as a behaviour issue or a medical one, so it may suggest a general vet or trainer rather than a specialist.
What you can do: Describe the specific problems you handle, for example anxiety, allergies or post-surgery care, so the model can match unclear requests.
query.uncertainty (was query.intent in the provisional inventory: Notes what the query leaves unknown) · also called pet_problem_uncertainty, pet_issue_confusion
physical_vs_digitalThe model decides whether the person wants a physical product or service, such as printed signs or an in-store visit, or a digital one, such as an online course or remote design work.
What you can do: Say whether you deliver physical goods, digital products or both, and how each is delivered.
query.intent · also called tangible_vs_intangible, physical_digital_distinction
planned_policy_mode_probabilityThe model estimates how likely the query is about planning a policy ahead of time, such as shopping for insurance before a renewal, versus urgent cover, and how much to weight that reading.
query.intent · also called policy_mode_likelihood, policy_mode_probability
production_mode_probabilityThe model estimates how likely the query is about production work, such as printing a run of yard signs or producing a video, rather than design or consulting, and how much to weight that reading.
query.intent · also called production_mode_likelihood, production_mode_estimate
project_sizeThe model estimates how big the job is, for example a single window versus a whole-house replacement, and matches businesses that take on that size.
What you can do: State your typical project sizes and any minimums, with examples of small and large jobs.
query.scope (was query.intent in the provisional inventory: Defines boundaries or size of the requested job) · also called project_scope, project_scale
ranking_focusThe model decides which criteria should lead the ranking for this query, such as speed for an emergency plumber or craftsmanship for a cabinetmaker, before it scores candidates.
What you can do: Make the facts behind common ranking criteria easy to find: response time, years in business, licenses, warranties and reviews.
evaluation.scoring (was query.intent in the provisional inventory: Decides ranking criteria weights) · also called ranking_priority, ranking_criteria · same label written as top_ranking_criteria
Used in 15 of 100 census runs this week: “Why these were chosen (high ratings, insurance, services).”
rapid_booking_needsThe model detects that the person wants to book soon, such as a same-day cleaning or a groomer this week, and favours businesses that show availability, online booking or quick response.
What you can do: Show online booking or clear availability, state same-day options if you have them, and keep hours correct everywhere.
query.intent · also called urgent_booking, immediate_booking_needs
Checked by Your hours, day by day, everywhere they are printed
recommendation_modifierThe model picks up words that change what kind of recommendation to give, such as "top-rated", "cheapest" or "trusted", and adjusts how it ranks the shortlist.
query.intent · also called recommendation_adjuster, suggestion_modifier
regional_vs_generalThe model decides whether the answer should be specific to a place, such as Denver building codes or local providers, or general advice that applies anywhere.
What you can do: Mention local specifics on your pages, such as city regulations, climate issues and neighbourhoods, so you are a strong source for regional answers.
query.geography (was query.intent in the provisional inventory: About distance or locality of the request) · also called local_vs_global, regional_general_distinction
regulatory_mode_probabilityThe model estimates how likely the query is mainly about meeting a rule or requirement, such as a required inspection or certification, and how much to weight compliance when choosing providers.
What you can do: If you do compliance work, list the specific inspections, certifications or regulations you handle.
query.intent · also called regulatory_likelihood, compliance_mode_probability
satisfaction_riskThe model weighs the chance the person is unhappy with the result, using review sentiment and complaint patterns, and favours businesses with consistent positive feedback.
What you can do: Ask satisfied customers for reviews, publish a satisfaction guarantee if you have one, and answer complaints.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called customer_satisfaction_risk, satisfaction_assessment
scaleThe model estimates the size of what is being asked for, such as one room versus a whole building, or ten guests versus five hundred, to match businesses of the right capacity.
What you can do: State your capacity: crew size, maximum guest count, or largest project handled.
query.scope (was query.intent in the provisional inventory: Defines boundaries or size of the requested job) · also called magnitude, scope
Used in 2 of 100 census runs this week: “0-100 per criterion, summed to a total score”
seasonal_vs_year_roundThe model decides whether the need is seasonal, such as snow removal or a summer campground, or year-round, and checks whether the business operates in the relevant season.
What you can do: State your season and hours by season, and keep seasonal closures updated on your site and listings.
evidence.freshness (was query.intent in the provisional inventory: Seasonal availability of the business) · also called seasonal_year_round_distinction, seasonality
Checked by Your hours, day by day, everywhere they are printed
service_delay_riskThe model weighs the chance the business is slow to start or finish, using complaints about no-shows or long waits, and lowers candidates with those patterns.
What you can do: Publish realistic lead times and response times, and reply to delay complaints with what changed.
evidence.assessment (was service.fit in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called delay_risk, service_timeliness_risk
service_disruption_riskThe model weighs the chance service is interrupted partway, such as a pool service missing weeks or an internet installer leaving a job unfinished, and favours businesses with steady track records.
What you can do: Describe staffing, backup coverage and what happens if a visit is missed.
evidence.assessment (was service.fit in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called disruption_risk, service_interruption_risk
service_nicheThe model identifies the specific niche a business serves, such as slate roofing or exotic-pet grooming, and uses it to match narrow queries to specialists.
What you can do: Name your niche explicitly in page titles and headings, and give it its own page.
service.fit · also called market niche, service specialization
service_riskThe model sums up what could go wrong with this kind of service, such as damage, delays or upselling, which sets how strictly it screens candidates.
What you can do: Address common risks directly on your site: insurance, warranties, upfront pricing and timelines.
evidence.assessment (was service.fit in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called risk assessment, service risk analysis
service_scope_ambiguityThe model notes it is unclear exactly what a business does or what the job includes, such as whether a "handyman" does electrical work, and may hedge or rank that business lower.
What you can do: List what you do and do not do on your services page, and keep that list the same on your listings.
service.fit · also called scope uncertainty, service clarity risk
size_of_eventThe model estimates how many guests or how big the event is, and matches caterers, venues or rental companies whose capacity fits.
What you can do: State minimum and maximum guest counts or capacity on your site.
query.scope (was query.intent in the provisional inventory: Defines boundaries or size of the requested job) · also called event scale, magnitude assessment
specializationThe model judges how specialised a business is in what the query asks for, such as an orthodontist versus a general dentist, and usually ranks a clear specialist higher for specific needs.
What you can do: State your specialties and any certifications that prove them, with a page for each.
service.fit · also called expertise level, field specialization · same label written as specialization_level
Used in 1 of 100 census runs this week: “Residential, new construction, and relocation services”
specific_areaThe model pins down the narrow sub-area of a field the query is about, such as family law within law, or tile within flooring, to search for providers in that sub-area.
What you can do: Name your sub-areas plainly in headings and service pages.
query.scope (was query.intent in the provisional inventory: Narrows the sub-area of the request) · also called niche area, focused domain
style_preference_riskThe model weighs the chance a business's style will not suit the person, such as a traditional photographer for a couple wanting candid shots, and favours those whose portfolio shows the requested style.
What you can do: Show your portfolio labelled by style, and name the styles you do and do not work in.
offer.fit (was query.intent in the provisional inventory: Whether candidate's style matches) · also called style mismatch risk, preference alignment risk
style_vs_availabilityThe model weighs the person's style preference against what is actually available locally and when, such as a specific florist style during wedding season.
What you can do: State your booking lead times and current availability alongside your style portfolio.
offer.fit (was query.intent in the provisional inventory: Style match against available offers) · also called style availability assessment, preference vs supply
time_modifierThe model picks up time words in the query, such as "tonight", "this weekend" or "before winter", that limit which businesses are useful.
What you can do: Keep hours and seasonal availability accurate on your site and listings.
query.intent · also called timing factors, time influences
Checked by Your hours, day by day, everywhere they are printed
time_specificityThe model judges how precisely the person gave a time, such as "Saturday at 10" versus "sometime soon", which decides whether it must check exact availability.
What you can do: Keep day-by-day hours consistent across your site, Business Profile and directories.
query.intent · also called specific time frame, temporal clarity
Checked by Your hours, day by day, everywhere they are printed
timingThe model works out when the service is needed, such as a date for an event or a season for planting, and checks candidates' availability against it.
What you can do: Keep hours, booking windows and seasonal schedules current on your site and listings.
query.intent · also called event timing, scheduling assessment
Checked by Your hours, day by day, everywhere they are printed
travel_inconvenience_riskThe model weighs how much hassle getting to or using the business involves, such as a remote campground, poor parking or long transfer times, and lowers options that add travel friction.
What you can do: State directions, parking, distance to landmarks or airports, and any shuttle or transfer service.
evidence.assessment (was query.intent in the provisional inventory: Risk judgment from evidence, not the person's intent) · also called travel risk, inconvenience assessment
type_de_vehiculeWhile reading the query, the model notes what kind of vehicle is involved (car, truck, RV, motorcycle, boat), because a mobile mechanic, tow service or detailer that handles sedans may not handle a 30-foot RV.
What you can do: List the vehicle types and sizes you service, such as cars, light trucks, RVs or fleets, on your service pages.
query.context (was query.intent in the provisional inventory: Vehicle type is background of the request) · also called vehicle type, transport category
usability_riskWhile interpreting the query, the model weighs whether the result could be hard for the person to use, such as a complicated smart thermostat a contractor installs or a booking process that needs several phone calls.
What you can do: Explain how customers use what you install or deliver, and offer simple booking with clear steps on your site.
evidence.assessment (was query.intent in the provisional inventory: Judges usability risk of the result) · also called usability concerns, user-friendliness risk
user_experience_levelThe model estimates how much the person already knows, for example a first-time homebuyer versus a landlord with ten units, and adjusts which businesses and how much explanation it gives.
What you can do: Have pages for first-time customers that explain the process plainly, alongside detail for experienced buyers or trade clients.
query.context (was query.intent in the provisional inventory: Describes the asker's knowledge level) · also called experience assessment, user skill level
value_mode_probabilityThe model estimates how likely it is that the person is shopping for good value rather than lowest price or top tier, which shifts it toward businesses that show what is included for the price.
What you can do: Show what each price includes, for example "lawn service from $45: mow, edge, blow", so value is visible.
query.intent (was offer.fit in the provisional inventory: Estimates the person's shopping mode, a preference) · also called value perception risk, value likelihood
Checked by The claims your site makes: years, prices, refund window
value_vs_premiumThe model judges whether the person wants solid value or is willing to pay for a premium provider, such as a standard carpet cleaner versus a high-end rug specialist.
What you can do: State plainly where you sit, with prices and what extra a premium option adds, on your services or pricing page.
query.intent (was offer.fit in the provisional inventory: Person's value-versus-premium preference) · also called value comparison, premium assessment
Checked by The claims your site makes: years, prices, refund window
vendor_mode_probabilityThe model estimates how likely it is that the person wants to hire a vendor directly, rather than get advice or do it themselves, which decides whether it names businesses at all.
What you can do: Make it easy to see you can be hired: services, service area, phone and a quote or booking path on the main pages.
query.intent · also called vendor engagement risk, vendor likelihood
warranty_needThe model judges whether the person will care about a warranty or guarantee, as with a roof replacement or a used car, and favours businesses that state their warranty terms.
What you can do: Publish your warranty or guarantee terms with length and coverage, for example "10-year workmanship warranty on roofs".
query.intent · also called warranty assessment, guarantee need
Checked by The claims your site makes: years, prices, refund window
workmanship_riskWhile reading the query, the model weighs the risk of poor workmanship, such as leaky tile work or a bad paint job, and leans toward businesses with evidence of careful work.
What you can do: Show photos of finished jobs, name your process and standards, and link to reviews that mention the quality of the work.
evidence.assessment (was query.intent in the provisional inventory: Judges workmanship risk from evidence) · also called quality risk, craftsmanship assessment
Also listed in this stage, defined under another: service_class → service, audience_type → audience, service_type → service
← Sourcing · Query Complexity →