The $842B home services industry has a six-figure revenue leak hiding in every contractor’s voicemail. Built from 2026 industry data, IICRC field standards, and two decades of enterprise operations, here’s the operator’s framework for closing it.
By the team at Pretzel Solutions, with Chad Stell
The United States home services market is on track to grow from $842 billion in 2026 to $989 billion by 2031. It is one of the largest, most fragmented, most underserved-by-technology industries on the planet. Most of it is still being run on whiteboards, paper invoices, and a phone that nobody is in a position to pick up.
In the last 24 months, that has started to change.
Voice AI has crossed a threshold. The best platforms now resolve 90–95% of calls without human escalation while maintaining 99% positive caller sentiment across benchmark testing of more than 347,000 real business calls. The technology is no longer the gating factor. Adoption is.
The bottom line up front: The average home services business is losing $126,000 per year to unanswered calls. The fix now costs a small fraction of that recovered revenue. The only remaining variable is how long the operator waits to deploy it.
This guide draws on two decades of enterprise sales operations experience — including scaling a single site from $10.9M to $26.8M at Conduent (the $4.1B BPO arm of Xerox) serving clients including Apple and BlueCross BlueShield — translated into the operational reality of the home services trades. Chad Stell assembled the Pretzel team specifically to bring that enterprise operational layer into restoration, HVAC, plumbing, and roofing businesses.
1. The leak
Independent research converges on the same uncomfortable number: between 27% and 74% of all calls to home services businesses go unanswered. Large multi-location operators with dedicated dispatch miss 27%. A two-truck plumbing company where the owner is the dispatcher misses 70%+.
Now layer in caller behaviour. Less than 3% of callers leave a voicemail. 80%+ hang up the moment they hear a recording. 85% of callers whose first call goes unanswered don’t call back — they call the next listing. And 78% of home services customers buy from whoever answers first.
This is the first-responder advantage, and in home services it is decisive. The technician you hire is not the one who is best. The technician you hire is the one who picked up.
The math:
- Plumbing standard: $80–$140 lost per missed call
- Plumbing emergency: $135–$200
- HVAC: $120–$320
- Roofing: $700–$875
- Water restoration: $900–$5,000+
The SMB-wide industry average is roughly $126,000 per year in lost revenue from unanswered calls, with each missed call worth about $1,200 in expected lifetime value. If you run a home services business, you almost certainly have a six-figure leak that doesn’t appear on any P&L line.
2. The Stell Standard: why home services breaks the 5-Minute Rule every day
In 2007, the Lead Response Management Study found that a web-submitted lead contacted within 5 minutes is 21 times more likely to be qualified than the same lead contacted at 30 minutes. The first company to make contact has a 238% higher conversion rate than the second.
In home services, the rule compresses. The functional response window for a restoration emergency or HVAC outage is closer to 60 seconds. Not 5 minutes. 60 seconds. The caller isn’t filling out a form between meetings. They’re standing in two inches of basement water at 11pm. They’re cold, they’re scared, and they’re looking at a list of Google results.
This is the operational gap modern voice AI has finally closed. A trade-specific AI answers in under 5 seconds, every call, 24/7.
3. Why generic AI fails in home services
Three reasons.
First, generic AI doesn’t speak the trade. When a homeowner calls at 2am about a sewer backup, the line needs to be picked up by something that knows that’s a Category 3 black water event under IICRC S500 standards, that the mitigation window before mold sets in is 24–48 hours, and how to triage urgency without losing the caller. A generic answering bot reading off a script can’t do that. The caller can tell within 30 seconds.
Second, generic systems don’t handle concurrency. During a hailstorm, a roofing company’s phone can ring 50 times in a single afternoon. Trade-specific platforms hold 50+ simultaneous concurrent calls. Generic SMB tools max out at 1–5.
Third, generic AI can’t route adjusters and homeowners differently. In restoration, roughly 59% of work is insurance-funded. Adjusters are B2B callers — they want claim numbers, ETAs, and moisture logs. Homeowners are B2C callers — they want reassurance and a truck on the way. A trade-specific system runs a branched decision tree from the first 8 words of the call and routes accordingly.
4. The 4-Pillar Pretzel Framework™
Every successful home services AI deployment we’ve seen runs on the same four pillars.
Pillar 1 — The 60-Second Floor. Every call answered in under 60 seconds. No exceptions for storm season, after-hours, or the dispatcher being in a meeting. This single discipline pays for the entire system inside 30 days.
Pillar 2 — Trade-Specific Triage Intelligence. The AI isn’t a voice. It’s a decision tree. Recognise emergency vs. non-emergency in the first 10 seconds. Classify urgency. Route accordingly. A receptionist takes a message. A triage system books the job.
Pillar 3 — Branched Routing Logic. Different callers, different paths. Homeowner with active emergency → reassurance, ETA, dispatch trigger. Adjuster → claim number capture, file ID, office routing. Commercial → property type confirmation, PM routing.
Pillar 4 — Structured Data Loopback. Every call generates structured data — urgency, trade category, job type, decision-maker, location, channel. That data flows into the CRM, attaches to a lead record, fires into marketing attribution, and pulls back out as reporting. Without the loopback, AI is a Band-Aid. With it, it’s a flywheel.
5. The ROI math
Take a mid-sized contractor missing 40 calls/month at a $600 average job value with a 35% close rate on answered calls. Annual missed revenue: $100,800.
Apply a realistic AI recovery rate of 50%. Recovered annual revenue: $50,400. System cost: a single-digit percentage of recovered revenue. Typical payback period: under 30 days.
The question isn’t whether the math works. The question is how long an operator is willing to leave $50,000+ on the table while deciding.
6. The operator’s 5-week implementation playbook
Week 1 — Audit the leak. Pull three numbers from your existing phone system: total inbound calls last 90 days, total unanswered, average job value of converted calls. If you can’t produce these three numbers, you have a measurement problem before you have a technology problem.
Week 2 — Build the triage tree. Document the decision tree before deploying. What’s an emergency? How are adjusters routed? What questions qualify a job in 60 seconds? The AI is only as smart as the tree you give it.
Week 3 — Deploy with parallel routing. Don’t replace your phone system overnight. Run AI in parallel for 30 days. Compare conversion rates.
Week 4 — CRM integration. Connect to ServiceTitan, Housecall Pro, Jobber, or GoHighLevel. If the data isn’t flowing back into your operation, you’re paying for an answering service, not a revenue system.
Week 5+ — Optimise on the data. Review call logs weekly. Refine the triage tree monthly. Compounding gains over 6–12 months come from the optimisation loop, not the initial deployment.
7. The Evolution Mandate: why deployment isn’t the end of the project
There’s a hard truth most vendors won’t say: the voice AI you deploy today is not the voice AI you’ll be running 12 months from now. The underlying models — OpenAI Realtime, Anthropic Claude voice, ElevenLabs Conversational AI, Google Gemini Live — are improving every 90 days.
If a vendor is selling the AI they built 18 months ago, they’re already two generations behind. This is the single biggest reason most home services AI deployments underperform. Operators buy a snapshot. The technology moves. The deployment stays still.
Pretzel operates on a 90-day Evolution Mandate built specifically for this reality. Every quarter, every major voice AI release is benchmarked against the home services test suite. Clients get migrated to the best-performing model. The trade vocabulary layer is refreshed — IICRC updates, R-454B refrigerant transitions, regional code changes. The horizontal AI vendors get broader every quarter. Pretzel gets deeper. That’s the moat, and it widens every 90 days.
The bottom line
The home services industry is going through the same transition every other multi-billion-dollar service industry has gone through in the last 20 years: from labour-bound to operationally-leveraged. The operational layer enterprise companies have used to scale for two decades finally became accessible to a two-truck plumber in Texas or a three-rep restoration company in Florida.
The companies that recognise this and deploy now will compound the advantage for the next decade. The companies that don’t will continue losing the same six-figure leak every year, blaming it on lead quality, market conditions, or the labour shortage.
The leak is real. The fix is here. The only remaining variable is how long it takes the operator to act.
The full operator framework — including the trade-by-trade math across restoration, HVAC, plumbing, and roofing, the FAQs, the glossary, and the data sources — is laid out in the 2026 home services AI operator guide.
Pretzel Solutions
Chad Stell
https://pretzelsolutions.ai/about
2026 home services AI operator guide
