What a Slow First Reply Actually Costs

A little after 6:40 on a Thursday evening in July, a homeowner fills in the contact form on an HVAC company's website. The air conditioning has stopped. She describes what happened, leaves her cell number, and hits send.

The form lands in an inbox two people can see. One of them is in a crawl space finishing a job and will not look at his phone until he is out of it. The other clocked out at five.

She gets a reply at 9:15 the next morning. It is a good reply — polite, specific, offers two windows on Monday. By then she has called two other companies, spoken to one of them at 7:15 the previous evening, and has a technician booked for Friday.

Nothing about that Thursday went wrong. No system failed. Nobody was careless. The company will bill a normal week and close a normal month, and the only trace of any of it is a form submission that got a courteous answer and never became anything.

The inquiry had a price before it had a chance

That form submission was not free. Somebody paid for it — with a click, a directory listing, a service ad, or twenty years of being the company people in that zip code call.

LOCALiQ and WordStream's 2026 benchmarks, drawn from 13,474 search campaigns running between April 2025 and March 2026, put the average cost per lead across all industries at $66.69. Home and home improvement carries the second-highest cost per click of any category they track, at $8.33.

Those numbers move around by market and by trade, and yours will not match them exactly. The point is not the figure. The point is that the money was spent before she typed anything, and it cost the same whether she heard back at 6:45pm or 9:15am. Waiting is not a way of holding onto it. It is a way of throwing it out quietly.

What the research found, and who paid for it

The core finding here is old and it is worth handling honestly.

In October 2007, at a MarketingSherpa conference, James Oldroyd — then a faculty fellow at MIT Sloan — presented a study of actual call data alongside Dave Elkington, the CEO of InsideSales.com. Two results from it have been quoted ever since. The odds of making contact with a lead drop roughly a hundredfold between a five-minute callback and a thirty-minute one. The odds of qualifying that lead drop about twenty-one times over the same half hour. Within the first hour, the odds of qualifying fall by more than six times.

Now the part that usually gets left out.

InsideSales.com sold software designed to solve the problem the study measured. The paper says so directly: the research, in its own words, "caused a significant shift in our corporate positioning." It closes with a product pitch. On the question of why speed matters so much, the authors concede they are guessing. And Oldroyd notes that the pattern only appears clearly once data from several companies is pooled together — meaning any one business looking at its own numbers might not see it at all.

That is a lot of qualification for a statistic people repeat as though it were a law of physics.

It is still worth taking seriously, for two reasons. Oldroyd went on to publish on the same subject with Kristina McElheran and Elkington in Harvard Business Review in March 2011, which put the finding in front of a considerably more skeptical readership. And the mechanism is obvious the moment you say it out loud: a person who has just typed a request is sitting there, phone in hand, thinking about the thing. Forty minutes later she is making dinner. Two days later she has forgotten she filled in the form.

Nineteen years later

Here is the part that should be strange.

In December 2024, Hatch published an analysis of 132,188 speed-to-lead campaigns run by HVAC businesses on its platform. Eighty-eight percent of those users took longer than five minutes to reply. Three percent replied in under a minute. The most common response time was a full day, at 37% of users, with thirty minutes a close second at 33%.

Read the sample again, because it is the whole point. These are not companies with no system. These are companies that had already bought follow-up software and were actively running campaigns inside it. They had the tool. They had, presumably, read the pitch that came with the tool, which would have cited the five-minute research. And the single most common reply time was still a day.

Nineteen years. A finding everyone in the industry can recite. Software bought and installed. Behavior essentially unchanged.

So why is it still true?

It is not a knowledge problem. The five-minute figure has been in circulation since 2007 and turns up in the marketing material of most vendors selling into these trades. It is not really a tooling problem either, or the Hatch numbers would look different.

What it is, in most companies of this size, is that nobody owns the inquiry.

Requests arrive through a website form, the main line, somebody's cell, and a referral that comes as a text message on a Sunday. Four ways in. No single person responsible for all four. Follow-up falls to whoever happens to be free, which on a Thursday evening in July is nobody, and on a Tuesday in February is whoever is closest to a laptop. This is the shape of businesses whose revenue arrives as appointments — trades, clinics, property firms — and it is the same shape whether the job is a furnace replacement or a root canal.

Software does not fix that, because software does not assign responsibility. It sits in the gap where the responsibility should be and waits for a person to operate it. Buy the best follow-up platform on the market, connect it properly, and if there is still no named person accountable for every inquiry from every channel, with a task attached and a due time, then what you have bought is a faster way of doing nothing. The distinction between adopting a tool and designing a system is the whole difference between the two outcomes, and it is decided before anyone logs in.

The tool sat there. So did the inquiry.

The bigger failure is the one nobody measures

Speed might be the second question.

RevenueHero ran an experiment on 1,000 business websites, submitting a demo request to each one and waiting. 63.5% never responded at all. Not slowly. Not badly. Never. The ones that did answer averaged around 29 hours.

That study looked at business software companies rather than home services, so treat the number as a signal about how organizations handle inbound generally, not as a benchmark for your trade. But it points somewhere uncomfortable. Before asking how fast you reply, it is worth establishing whether every inquiry gets a reply at all — because an inquiry nobody ever touches is not a speed problem, and no amount of responding faster to the ones you do see will surface it.

Most companies cannot answer that question about their own business. The web form leaves a record. A missed call at 7pm that nobody returned leaves nothing to reconcile against, so it never appears on any report, and the month closes looking normal.

What this is worth where you are

None of the numbers above are your numbers.

Your cost per inquiry, your close rate on the ones you actually speak to, your real response time on an average day rather than a good one — those produce a figure that means something, and nobody else's benchmarks can produce it for you.

There is a calculator on this site that does the arithmetic. It takes five inputs, runs entirely in your browser, stores nothing, and asks for no email address. The absence of a form is deliberate: a tool that costs you a conversation with a salesperson is not a tool, it is a lead magnet, and the whole argument here is that you should be measuring this yourself.

Worth saying plainly: this is not everybody's problem. A single-location practice taking eight inquiries a month, where the owner knows each caller by name and rings them back from the parking lot, does not need any of this. Neither does a business whose calendar is full for the next six weeks and turning work away. The math only bites when there is enough volume that individual inquiries stop being memorable, and enough margin on each one that losing a share of them costs real money.

If that is you, the number is probably larger than you would guess, and it has been the same number every year.

The Thursday in July is the part worth sitting with. Nothing broke. Nobody was slow by the standards of the business — 9:15 the next morning is a perfectly respectable reply, and on most days it would have been fine.

The inquiry was paid for, it was answered, and it was lost. Only the first two of those show up anywhere.