A ServicePulse AI field study · June 2026
We called 1,077 contractors as a real customer.Half never got us to a helpful human.
Roofers, plumbers, HVAC techs, electricians, painters, concrete crews, landscapers, pool companies — called during normal business hours, posing as a customer with a real job to book. Every call recorded where the law allows. 953 recordings on file. 736 calls classified. Everything below is measured from those calls — no estimates, no projections.
What happened on the calls
Read that again: 50% of the time, a ready-to-buy customer could not get a helpful human on the phone. And the most damaging category is the quietest — 13% of the time a person physically answered the phone and the call still died.That's not a missed call. That's a job that called, got a human, and walked away anyway.
It's not equal across the trades
Reached-a-human rate, by trade:
Electricians, pool companies, concrete crews, and painters answer fewer than half their calls. In those trades, the business that simply picks uptakes work directly off the competitor who doesn't.
The failure nobody talks about
When we re-screened 620 answered calls looking for one specific thing — jobs turned away because the shop was booked solid — we found it in about one of every forty answered calls. Real quotes:
- “It would be Monday before I could come.”
- “It will be about a week, week and a half.”
- “Don't have availability until next week.”
One plumber, fully booked, helped our caller find a competitor's number. These are the goodshops — busy because they're good — and being good built the leak. Answering the phone was never the finish line. Captured is the finish line — a name, a number, a slot on the calendar.
What it costs (the honest math)
We didn't open anyone's books, so we won't quote a number we didn't measure. But take a contractor missing even 10 calls a month — modest at a 50% answer rate — at an industry-typical 30–50% close rate and the average job value for their trade, and the model runs to a five-figure monthly leak. That's an estimate, not an audit— deliberately conservative. The point isn't the exact dollar; it's that the calls that don't connect never show up in any report. You can't miss what you never saw ring.
Methodology
- Sample: 1,077 outbound calls to real, operating contractors (sourced from Google Business listings), during normal business hours.
- Method: An AI mystery shopper called each business as a prospective customer with a routine job. No selling, no mid-call disclosure — a normal customer experience.
- Recording: Calls recorded in one-party-consent states only; 953 recordings on file.
- Classification: 736 calls classified by outcome. This report uses only those classified calls.
Trade-level data and methodology detail available on request: hello@servicepulseai.com
Want to know how your phone scores?
We'll run the exact same test on your business line — one real call, recorded — and send you what we heard. Free, no card.
Conducted by ServicePulse AI LLC, Texas — we build AI phone & text answering for home-services contractors. Figures reflect 736 classified calls of 1,077 placed, June 2026.