What is an AI citation score and why it matters for your HVAC business
AI citation score is a 0-100 measure of how often your business is named when homeowners ask AI assistants for local-services help. Here's how it's calculated, why it predicts booked jobs better than any other single metric in 2026, and how to read your own score.
Most marketing metrics tell you about activity. Impressions, clicks, sessions, bounce rate, time on page. The AI citation score tells you about a much harder thing — whether your business exists in the answer when a real customer asks the AI for help. It's a binary signal stacked across many trials. Either the AI named you or it didn't. Either the homeowner saw "Acme HVAC" in their ChatGPT response, or they saw "Roto-Rooter, ARS, and Mr. Rooter."
The score is useful because it correlates tightly with actual phone-ringing behavior. Internal benchmarks from contractors we audit show that for every 10-point lift in citation share, booked jobs from new (non-repeat, non-referral) customers rise 14-22% within 60 days, all else equal. That's a much stronger signal than "organic traffic up 8%" — which can mean anything or nothing about phone volume.
How is the score calculated, exactly?
Three inputs feed the score. They roughly map to: how often you appear, where you appear vs the chains, and how reliable the underlying entity signal is.
Input 1 — citation share (50% weight)
Run 30 standardized homeowner-style queries against 4 AI engines (ChatGPT, Claude, Perplexity, Google AI Overviews). Each query is parameterized for your city, state, and service line. The 30 queries cover five intent categories — direct ("best HVAC near me"), comparison ("Roto-Rooter vs local HVAC in {city}"), specific job ("AC company that does ductless installs in {city}"), pre-purchase research ("how much does AC repair cost in {city}"), and trust ("honest HVAC contractor {city}"). Count how many of the 120 prompt-engine combinations name your business. Divide by 120. That percentage is your citation share.
Input 2 — entity strength / NAP consistency (25% weight)
Audit your name, address, and phone number across 12 directories — Google Business Profile, Bing Places, Apple Business Connect, Yelp, Angi, HomeAdvisor, BBB, Yellow Pages, Foursquare, MapQuest, Citysearch, MerchantCircle. Tier-1 directories (Google, Bing, Apple) are weighted ×3. Tier-2 (Yelp, Angi, HomeAdvisor, BBB) are weighted ×2. Tier-3 (the rest) are weighted ×1. A perfect 100 means every field matches byte-for-byte across every directory.
Input 3 — Google Business Profile completeness (15% weight)
Photos, hours, services list, attributes (financing, after-hours, languages, accessibility), Q&A engagement, posts cadence, review velocity, and operator response rate. Score is derived from how many of the available GBP features are populated and how recently they were updated.
Input 4 — lead-flow diagnostic (10% weight)
Page speed (Lighthouse mobile + desktop), schema markup validation, contact-form submission flow, click-to-call functionality, and missed-call test. None of the visibility work matters if the phone rings and nobody picks up.
| Component | Weight | What it measures |
|---|---|---|
| AI citation share | 50% | Mentions in 30 queries × 4 AI engines vs chains and top-3 local rivals |
| Entity strength (NAP) | 25% | Name/Address/Phone consistency across 12 directories |
| Google Business Profile | 15% | Photos, hours, services, attributes, posts, review velocity |
| Lead flow | 10% | Schema, page speed, contact form, click-to-call, missed-call response |
Source: TNova Labs methodology v2.0
Why is citation share weighted heaviest?
Because in 2026, the metric that predicts booked jobs is whether you're named when the AI gives one answer instead of ten links. Forty-five percent of US homeowners now start their local-services search inside an AI assistant (MarketingCode 2026 study), and that share is climbing every quarter. The other three components matter — they're the inputs that determine whether the AI cites you in the first place. But citation share is the outcome that matters to the contractor's revenue.
Two years ago we would have weighted GBP completeness much higher. Today GBP completeness is necessary but no longer sufficient. A perfect Google Business Profile with zero presence in AI answers no longer keeps a phone ringing the way it did in 2024.
What's a good score for an independent HVAC contractor?
Distribution across the contractors we've audited:
Source: TNova Labs
Half of the contractors we audit score under 20. A score above 60 is rare. A score above 80 means you've out-cited the chains in your metro for your service line, which we have measured in exactly one of the 43 audits to date — and that contractor had spent four years building a press-relations and content moat that the chains haven't matched locally.
How fast can the score move?
First measurable lift: 4-8 weeks after entity strength work is deployed (NAP cleanup, schema markup, GBP completion, directory coverage). The lift comes as AI engines refresh their citation graphs and pick up the cleaned-up signals.
Compounding lift: 6-12 months as press mentions, news coverage, and city-specific deep content accumulate. The longer you maintain the entity work, the more the AI engines treat your business as a recognized local entity.
Plateau: 18-24 months at a score that reflects your metro's competitive density. In a small metro with two competitive independents, scores in the 60-80 range are achievable. In Houston or Phoenix with hundreds of qualifying contractors and three established chain presences, a score above 50 is exceptional.
How does the score relate to revenue?
Loosely but predictably. Across audited contractors, every 10-point lift in AI Visibility score correlates with 14-22% lift in net-new booked jobs from non-repeat, non-referral sources within 60 days. Revenue lift trails by 30-60 more days as new customers complete their first job and the lifetime-value tail begins.
We don't claim causation in either direction — there are too many confounding variables (seasonality, weather events, competitor moves). What we claim is that contractors who invest in moving their citation score up tend to also see their booked-jobs number move, and the rough proportionality holds across enough audits to be a useful planning heuristic.
How do you measure your own score?
Manually: pick 10 of the 30 standardized queries (or write your own city-and-service-specific versions), submit each to ChatGPT, Claude, and Google's AI Overviews, count how many name your business. Divide by 30. That gives you a rough citation share. Combining it with NAP and GBP audits is more involved but doable with a spreadsheet and a few hours.
Automated: we run the full audit at /visibility-audit. Free, 24-hour turnaround, written report. The link is below. We use the same methodology described above.
How is AI citation score different from organic SEO?
Organic SEO measures how high you rank in search engine results pages. Citation score measures how often you're named when AI assistants give one direct answer to a homeowner's question. The two are correlated — strong organic SEO usually means decent entity strength — but they diverge in 2026. You can rank #1 organically and have 0% citation share if your entity layer (NAP, schema, directories) is weak.
Does paying for ads improve your citation score?
Not directly. AI engines don't index paid placements as identity signals. Paid ads can drive review volume and brand recognition over time, which indirectly helps citation share, but the lift is slow and indirect. The fastest dollar moves on citation score are spent on technical entity work, not paid clicks.
Can I get a perfect 100?
Almost never. A perfect 100 would require dominating AI citations against the national chains and every local competitor across all 30 query intents — which is structurally rare in any competitive metro. A score above 70 is the practical ceiling we plan against. Most contractors should target 50-60 within 12 months and treat anything above as a stretch goal.
Are the AI engines stable enough that the score is meaningful?
Stable enough at 4-week intervals. AI engines re-rank constantly within a day, but the underlying citation graph (the data they pull from to compose answers) refreshes on a slower cadence. We've validated that the same 30 queries against the same 4 engines produce results within ±3 points across consecutive 4-week measurements for the same business — tight enough to detect real change.
Does the score account for paid AI placements (sponsored answers)?
Currently no, because no major AI engine has launched fully sponsored business placements as of May 2026 — though Google AI Overviews has begun testing sponsored citations in a few US verticals. When sponsored placements roll out broadly, we will add a separate sub-score for paid-vs-organic AI mentions so the metric stays interpretable.
See where you show up when customers search.
Free 24-hour Visibility Audit. Yours to keep — whether you hire us or not.