We logged 551 cold call outcomes by hand. Only 18.5 percent reached a live human, and one number in eight was dead.
Most cold calling advice is written from memory. This is written from a call log. Every number comes from outcome notes a caller typed after each attempt on a list of 4,818 US home services businesses.
See the full results tableinto the CRM
by hand
conversation
dead on arrival
Every outcome, counted
Six categories, 551 attempts, nothing excluded.
| Outcome | What the caller wrote | Attempts | Share |
|---|---|---|---|
| Voicemail or message left | Voicemail, left a message, machine picked up | 299 | 54.3% |
| No answer | Rang out, nobody picked up | 81 | 14.7% |
| Dead number | Not assigned, disconnected, wrong number, out of service | 69 | 12.5% |
| Reached a human who refused | Not interested, hung up, do not call | 64 | 11.6% |
| Callback promised | Call back later, send information, speak to the owner | 35 | 6.4% |
| Business closed | Company no longer trading | 3 | 0.5% |
| All logged attempts | 551 | 100.0% | |
| Of which reached a live human (refused, callback, or closure confirmed by a person) | 102 | 18.5% | |
If you are paying for calls, here is what you are actually buying
Owners judge a calling program on meetings booked. These six numbers explain what happens long before anyone books one.
An eighth of your budget dials nothing
69 of 551 attempts hit a number that was disconnected, reassigned or wrong. Across 4,818 rows that projects to roughly 600 unusable records, each paid for twice, once when the list is bought and again when a caller proves it dead.
Voicemail is the majority experience
More than half of all attempts ended at an answering machine. A program that treats voicemail as a wasted dial has written off most of what you pay for.
Roughly five dials per conversation
102 live conversations across 551 attempts. A quoted connect rate above 30 percent on cold small business numbers means a warmer list, repeated attempts, or different counting.
Most people who answer say no, and that is normal
Of the 102 people reached, 64 declined. A refusal is a qualified removal, not a failed call. The expensive outcome is a number nobody reaches.
A third of conversations went somewhere
35 of the 102 people reached asked for a callback, information, or a time to speak to the owner. That is the real output of a first pass, and it becomes a meeting only if somebody dials again.
Closed businesses are rare, bad data is not
Only 3 companies had actually gone out of business. Twenty three times more numbers were dead than companies were, so this is a data problem, not a shrinking market.
Three fixes that follow directly from this data
Each exists because a number above made the old approach indefensible.
Verify the list before anyone dials it
At 12.5 percent dead, a caller working 4,818 rows spends hundreds of dials discovering what a line validation pass finds in an afternoon. Validate first, drop what fails, and the same paid hour buys more conversations.
The test: if a provider cannot tell you the dead number rate on the list they sold you, they never measured it.Treat voicemail as a channel, not a failure
Voicemail is 54.3 percent of everything that happened. Leaving no message, or a different improvised one each time, throws away the largest point of contact you have. Script it, keep it under twenty seconds, and send an email the same day.
The test: ask to hear the voicemail script. If it is not written down, it does not exist.Plan and fund attempts two, three and four
Our data is mostly first attempts, and first attempts are where refusals and voicemails live. The 35 callback promises hold the remaining value, and each needs a second call at the time somebody named.
The test: count how many numbers on your list were dialed more than once. In most programs we review it is close to zero.How to use these numbers
- 1Remove 12.5 percent of your list until you have measured your own dead rate.
- 2Expect one live conversation per five and a half first attempts.
- 3Expect about a third of those to produce a callback.
- 4Model meetings from the callback pool, not from a first pass.
How we collected this, and where it stops being reliable
A study is worth citing only if it says what it cannot prove.
How we collected this
- The list
- 4,818 US home services businesses (HVAC, plumbing, heating and air, contractors) loaded with phone numbers into a CRM we run for a client campaign.
- The period
- July to September 2026, in US business hours.
- The caller
- One caller, working the list in sequence from a single line.
- The records
- 551 outcome notes, typed into the CRM record right after each attempt. Nothing generated automatically or back filled later.
- The grouping
- Free text notes grouped by keyword, so not in service and number disconnected both count as dead. Each attempt landed in one category.
- The arithmetic
- Each category divided by 551. The live human figure sums refusals, callbacks and confirmed closures, since each needed a person to answer.
Honest limits, stated plainly
One client, one vertical. One campaign in US home services. Businesses that answer their own phones behave differently from enterprises with switchboards, and another industry could move every number here.
One caller. Connect rates move with the hour, the day and the person. Another caller would not produce the same figures.
Mostly first attempts. In most cases a number was dialed once, which is why this is not a booking rate study. Bookings arrive on attempts two to four, and those are not in this data.
Human wording varies. Notes were written by a person between calls, so phrasing varies and the grouping is a judgment call at the edges. A note reading no answer could have been a ring out or an unrecognized voicemail.
No revenue outcome. This measures what happened on the phone. It claims nothing about deals closed, cost per appointment or return on spend.
Straight answers about this data
Want to know what these numbers look like on your list?
Tell us your market and what you sell. We come back with an honest read on your list quality and where your follow up leaks. No charge, no obligation.
Get My Free Plan and Price