The Trades Phone Line Report
What happened when we called 3,944 US trades businesses in 30 days. Not a survey. Our own dialling log, aggregated.
Where this data comes from
My team and I made 3,944 outbound calls to US trades businesses (septic, well, plumbing, HVAC, roofing, restoration) over 30 days. Every call was logged automatically: outcome, duration, who answered, what they said. This report is that log, aggregated.
We excluded our own test lines, inbound rings, and 302 Canadian dials, so every number below is a clean US census. Timezones are resolved per number: "8am" means 8am at their local time, every row.
Every number we dialled is the business's publicly listed main line, taken from their Google Business Profile. These are not owner mobiles. We never bought a mobile append and never dialled one. So what follows is not a picture of how easy it is to reach an owner privately. It is a picture of what happens when a stranger calls the number a customer would call, which is the whole point.
Vendors publish consumer surveys. This is what actually happens when the phone rings on the other side.
The morning trap: pickups fall all day, owners rise
Call at 8am and almost half of calls get answered, but it is the office. Call at 5pm and most calls go nowhere, but when someone does pick up, half the time the person answering the main line is the owner. The office went home. The owner is still working.
If you sell to trades: mornings buy volume, late afternoons buy decision makers. Between 8 and 9am roughly 1 in 5 answered calls is the owner. After 3pm it is closer to 2 in 5.
If you run a trades business: your customers hit the same wall. At lunchtime, 42 to 47% of our calls went straight to voicemail. Your busiest quoting window is when your phone is least answered.
The gatekeeper wall
Across 1,438 classified answered calls: 69.6% were a gatekeeper or office staff. 23.4% were the owner or a decision maker. Seven in ten conversations end with someone who cannot say yes.
This is a main-line number by definition, and that is exactly what makes it useful. A business that pays someone to answer the published line is a business whose customers also meet that person first.
The counterintuitive part, from the subset where we hold verified review counts (small sample, n on every row): the more established the company, the more they answer, and the higher the wall gets.
| Google reviews | Dials | Answered | Gatekeeper share of answered |
|---|---|---|---|
| 0 | 21 | 14.3% | 33.3% |
| 10–29 | 37 | 29.7% | 72.7% |
| 30–74 | 50 | 24.0% | 58.3% |
| 75–199 | 61 | 37.7% | 82.6% |
| 200+ | 21 | 57.1% | nearly all |
Small operators do not answer because they are on the tools. Big operators answer, but with a person paid to say no.
What 1,068 trades businesses said when we called
Of 1,474 answered conversations, 1,068 gave a reason.
"No need" plus "too busy" is 70% of all objections. Almost nobody said "too expensive": 15 of 1,068. In this market, price is not the fight. Attention is.
The businesses that say they don't need it are missing calls
"We don't need it" was the single most common thing we heard. It is worth putting that next to what the same phone lines actually do.
Start with the headline the summary understates. 38.7% of calls went to voicemail, but another 21.7% simply rang out. Taken together, 60.4% of all 3,944 dials never reached a human at all.
Then the per-number view. Of 3,482 unique US numbers, 273 were dialled more than once inside the window:
| Cohort | n | Result |
|---|---|---|
| Numbers dialled more than once | 273 | 41.8% both answered us once and missed us another time |
| Gave a "we don't need it" objection | 431 | 45 of them were dialled again at all |
| Of those 45 redialled | 45 | 36 missed at least one later call |
The middle row is the load-bearing one. The same line that picks up for you on Tuesday misses you on Thursday, and that holds across 273 numbers rather than a handful. The "no need" cohort points the same way, but only 45 of those 431 businesses happened to get a second dial, so it is reported as a count and should not be read as a rate.
What happens on the other end is not our data, but it is measured: when a call goes unanswered, 72% of people leave no voicemail and 81% ring the next business (RunClockwork, 2.4 million calls across 1,247 US contractors, 2026).
We deliberately publish no average job value here. We do not hold one we can source, and a roofing number in a plumber's hands is worse than no number. Put your own ticket against one missed job a month and the arithmetic is yours, not ours.
Persistence does not pay on cold numbers
| Attempt on the same number | Dials | Answered |
|---|---|---|
| 1st | 3,482 | 38.8% |
| 2nd | 273 | 28.2% |
| 3rd | 82 | 25.6% |
| 4th+ | 107 | 23.4% |
Every repeat attempt in the same month answers less than the one before. A number that did not pick up is telling you something.
Small print worth knowing
Best days: Monday (40.9%, n=611) and Thursday (41.1%, n=834). Worst weekday: Wednesday (32.1%, n=1,008). Saturday is a dead zone (17.8% on 45 dials).
You get one minute: the median answered conversation lasted 61 seconds.
What we dialled: publicly listed business main lines from Google Business Profile, one number per business. No owner mobile appends were purchased or called, so read every "owner share" figure as "share of people answering the main line who turned out to be the owner".
Method notes: outcomes logged automatically per call; owner share measured only on answered calls we classified (n shown); the review-count table is a labelled subsample where we hold verified data, not the full census.
The uncomfortable conclusion, for us too
I build AI voice receptionists, so here is the honest version of what this data says about my own pitch: trades businesses are not sitting by the phone waiting to buy. 60.4% of our calls never reached a human. The same thing happens to their customers. A homeowner with a flooded basement gets that voicemail once, and the next business on Google gets the job.
That is the whole reason the product exists. But we only learned the shape of the problem by making four thousand calls into it.