A fragmented, labor-starved industry PE is consolidating fast
The trades are the hottest roll-up in private equity right now. Apex Service Partners (Apollo-backed) has scaled to 107 brands and ~$1.3B revenue, closing roughly 60 add-ons in 2025 alone (PipelineOn); Champions Group sold to Blackstone in 2026 at ~$2.5B and 18.5x EBITDA (CT Acquisitions). The model rests on multiple arbitrage — pay 17-20x at platform level, source add-ons at 5-8x. But arbitrage alone is a treadmill. The platforms that win get operationally better, and the binding constraints are labor and demand capture. That’s where AI earns its keep. Not chatbots. Cash.
Play 1 — Never miss a call: AI booking and missed-call recovery
This is the single most defensible play because the leak is enormous and the fix is cheap. 27% of calls to home-services businesses go unanswered, less than 3% of voicemail callers leave a message, and ~85% never call back (Invoca). A missed call is a lost job. AI voice agents now book those calls 24/7 — Newo.ai converted 64% of calls to booked appointments vs. 60% for top human CSRs in a head-to-head (Newo.ai). Avoca raised $125M+ at a $1B valuation in 2026 — capital is voting on this play (PR Newswire); ServiceTitan now bundles it natively (ServiceTitan). For a 20-location platform, recovering a few jobs a day per site is seven figures of incremental revenue at near-zero marginal cost.
"What's our true call-answer rate by location, including after-hours and overflow — and what booking rate are we getting on recovered calls? If you can't answer in numbers, we're leaking jobs we already paid marketing to generate."
Play 2 — Dispatch and routing: more revenue from the same techs
Labor is the constraint, so the highest-leverage move is making each existing tech more productive. The average field tech loses 40%+ of the workday to travel and idle time (AEX); AI routing reclaims it. The smarter version isn’t shortest-route — it’s value-aware assignment: ServiceTitan’s Dispatch Pro predicts a job’s value and matches the right tech by skill and location (ServiceTitan). A senior tech on a high-ticket replacement earns far more than on an $89 tune-up. For a roll-up, standardizing dispatch across acquired brands is where “synergy” stops being a slide.
Play 3 — First-time-fix: the silent margin killer
Every repeat visit is a free truck roll you eat. Industry-average first-time-fix is ~75%; top quartile hits 88%+ (ServicePower), and an unresolved call needs 1.6 additional dispatches at $200-300 per truck roll (AEX). Parts availability drives 51% of first-visit failures; skills mismatch ~25%. AI helps by predicting the likely fix and required parts before the truck rolls. This is pure cost-out, no demand assumptions required — a 10-point FTF improvement drops straight to EBITDA.
Play 4 — Pricing and quote optimization: the fastest path to margin
Most acquired shops underprice and quote inconsistently. Moving to structured flat-rate and good/better/best pricing is the highest-margin lever in the toolkit — a tested 3-tier vs. 2-tier presentation lifted average project value 34% at the same conversion rate (FieldCamp). Be skeptical of vendor figures, but the direction is well-established. For a roll-up, standardizing a pricebook across 50 brands is a margin event by itself; AI makes it adaptive.
Play 5 — Membership and retention: manufacturing the recurring revenue PE pays for
This is the play that directly re-rates the asset. Shops with 50%+ revenue in maintenance contracts and 90%+ renewal trade at 8-10x EBITDA; moving a $1M-EBITDA business from 30% to 50% recurring can add $500K-$2M in enterprise value (Main Street Wealth). Membership books retain in the mid-90s% versus 65-70% for emergency-only operators (RepuClinic). AI’s role: predict which one-time customers are membership-likely and which members are about to churn, then trigger the save before they lapse. Lead with this in an IC memo — it converts an operating improvement directly into multiple expansion.
"What percentage of revenue is under membership, and what's our renewal rate? What's the plan to get from X% to 50%, and do we have a churn-prediction trigger — or are we finding out customers left when the renewal doesn't hit?"
What kills these initiatives
- No clean data. AI dispatch/pricing/churn models need consistent job, parts, and CRM data. Newly acquired shops run different software or paper.
- Treating it as a tech project, not an ops change. AI call-booking fails if dispatchers don’t trust it or the capacity calendar is wrong.
- Vendor-spin ROI baked into the model. Pilot one brand against a control, then roll out.
- Integration drag in roll-ups. 100 brands on 100 configurations means the AI never sees a standardized pricebook. Consolidate the platform first, then layer AI.
- Pricing backlash. Aggressive dynamic pricing can spike margin and tank reviews. Watch CSAT alongside ticket size.
- No owner of the number. If no one’s accountable for “answer rate” or “revenue per tech” post-close, the tool gets bought and never moves EBITDA.
Sources
| Source | What it told us | Confidence |
|---|---|---|
| PipelineOn | Named platforms (Apex, Wrench), multiple-arbitrage math | STRONG |
| CT Acquisitions | 27+ active PE HVAC platforms; Blackstone/Champions 18.5x | STRONG |
| Invoca | 27% calls unanswered, under 3% leave voicemail | MEDIUM |
| Newo.ai | AI 64% vs human 60% booking head-to-head | VENDOR |
| PR Newswire — Avoca | $125M raise at $1B valuation | STRONG |
| ServiceTitan — Contact Center | AI Voice Agents, Second Chance Leads recovery | VENDOR |
| ServiceTitan — Dispatch Pro | Value-prediction + skill/location matching | VENDOR |
| ServicePower | 75% avg, 88%+ top quartile FTF | MEDIUM |
| AEX | Parts = 51% of FTF failures; 1.6 extra dispatches | MEDIUM |
| FieldCamp | 3-tier presentation +34% project value | VENDOR |
| Main Street Wealth | 50%+ recurring = 8-10x EBITDA; +$500K-$2M EV | MEDIUM |
| RepuClinic | Membership mid-90s% retention vs 65-70% | MEDIUM |