How fast to reply to a new inquiry: what the studies show

Here is an invented but familiar case. A web inquiry reaches a plumber in Leeds at 9:12 p.m. on a Tuesday: a leaking tank, a phone number, nothing else. He sees it at 7 the next morning and calls. The customer answers, politely, and says a firm already came out at 10 p.m. the night before.

Nobody did anything wrong. The plumber was asleep. What cost him the customer was the gap between 9:12 p.m. and 7 a.m.

How much is that story worth, and what should an owner do about it? Most advice answers with a number: 5 minutes. The number has a pedigree, but it is thinner than the posts quoting it suggest. We went back to what the original studies measured. This piece says where we could and could not read them, and it ends with a reply policy that does not depend on any single statistic.

The 2007 study and the 2011 article get blended into one

When someone says "studies show" about reply speed, they are usually blending two pieces of research.

The 2007 study. In October 2007, the company now called InsideSales (its press release is headed with an earlier name, XANT) presented research at a MarketingSherpa conference, done in conjunction with Dr. James Oldroyd, whom the release describes as a Sloan School of Management fellow at MIT. The company's own write-up states that contacting a lead within 5 minutes "yields 21 times more qualifications" than waiting 30 minutes.^[1] The same page says the analysis drew on "tens of thousands of calls" and does not give the number of companies or leads.

Secondary summaries add detail: six companies, more than 15,000 leads, and more than 100,000 call attempts, with contact odds about 100 times higher at 5 minutes than at 30.^[2][3] We could not find that detail on a primary page, so treat the exact counts as reported rather than confirmed.

Two details deserve attention. The 100x and 21x figures compare 5 minutes with 30, not with a day. And one secondary summary notes that the study did not measure close rates at all.^[2] What it measured was whether a salesperson reached a person, and whether that person counted as qualified under the company's own definition. Neither is revenue.

The shorthand "MIT study" overstates it. One summary describes the work as vendor platform data analyzed by an academic, not an MIT publication.^[2] That is no scandal. It is a reason to write "InsideSales.com and Dr. Oldroyd" instead of "MIT".

The 2011 article. In March 2011, Harvard Business Review published "The Short Life of Online Sales Leads" by James Oldroyd, Kristina McElheran, and David Elkington.^[4] The HBR page lists Elkington as chairman and CEO of InsideSales.com, the company whose CEO presented the 2007 research,^[1][4] so this is a follow-up with a shared lineage, not independent confirmation.

Our reading has a gap here. The HBR page is behind a paywall, and we did not get around it: we saw the header, the authors, and the date. Every number below comes from two secondary summaries, and none could be checked against the article text.

The summaries describe two pieces of work. In the first, researchers submitted a web inquiry to 2,241 US companies and timed the replies. The second was a dataset of about 1.25 million leads at 42 companies (29 consumer-facing and 13 business-to-business), according to one summary.^[2]

As reported, the test found that 37% of companies replied within an hour, 23% never replied, and the average first response among those who did reply came after about 42 hours.^[2][3] One summary notes that 42 is often misquoted as 47.^[2] As reported, the lead dataset showed that firms that tried to contact a lead within an hour were roughly seven times as likely to qualify it as firms that tried even an hour later.^[2]

What the numbers can carry

One finding deserves more attention than the multipliers: most companies were slow, and many never replied. By the reported figures, a firm that replied within an hour was among the fastest 37% in that test, and the 23% that never replied were not competing at all.^[2][3] For a small business that is more useful than "5 minutes".

Now the limits, for anyone planning to build a process on these numbers.

Records of what firms did, not an experiment. Neither summary describes anyone assigning reply speed at random, and we found nothing in them saying the multipliers were adjusted for lead quality or sales skill. That is our reading, not a finding: take them as associations. Fast responders may be better organized in general. They may also get hotter leads, because a person who filled in a form while holding their phone is simply easier to reach.

It is old. The 2007 dataset covers roughly 2004 to 2007, before smartphones were everywhere.^[3] Reply behavior has probably changed. The summaries cite newer vendor-run work, such as a 2017 Drift audit of 433 B2B websites^[3] and a 2021 study of 5.7 million inbound leads by the company now called InsideSales.^[2] We have not read either, and we cannot say how they treat messaging apps.

It measures phone contact. The 2007 figures are about telephone call attempts and the share that ended in a real conversation.^[2][3] If your inquiries arrive by form and you reply by email or message, the mechanism may differ. A message waits; a phone call is a window that may shut.

The numbers drift when repeated. A trade magazine summarizing the work cited "a 90 percent increase in contact rate when responding within five minutes".^[5] That is a different claim from 100 times, with no source given. Meanwhile the oft-seen line that "78% of customers buy from the company that responds first" is, according to one analysis, traceable to no primary source.^[2] We do not use it for that reason.

None of this makes the core idea wrong. It makes the specific multipliers unsafe to put in a business case.

A fast reply that says nothing

Both studies measure the first attempt. Neither tells you what a fast, useless reply is worth.

An instant auto-acknowledgment ("Thanks, we'll be in touch") has not made contact with anyone. It does not tell a person whether their question is understood, whether a slot exists, or what happens next. Stop there and you have bought the look of speed. The point of the research, read generously, is that a person's problem gets engaged with while it is still hot. A reply that does that in 15 minutes can beat a hollow one in 15 seconds.

So the useful distinction is not fast versus slow. It is acknowledged versus engaged.

A reply-time policy that rests on no statistic

What follows is our judgment, not a research finding, because the evidence above cannot justify more.

1. Pick two clocks, not one. One clock is acknowledged: the inquirer knows the message arrived and what happens next. The other is engaged: someone, or something, has answered the question that was asked. Set a target for each. For acknowledgment, a target in minutes is reasonable. For engagement, set a number you can keep, and write it where the team can see it.

2. Split business hours from after hours. During hours, route to a named person with a short target, such as 15 minutes, and a fallback if they are away. After hours, do not pretend someone is there. Say when you will reply ("We open at 8. You will hear from us by 9") and keep to it. A promise you keep beats a speed you cannot sustain.

3. Automate the parts that need no judgment. Confirmation, the three questions you always ask, booking a slot from a calendar, a short answer to "do you cover my postal code". These are safe to automate as long as the system says it is automatic.

4. Hand anything with money, risk, or emotion in it to a person. A complaint, a price negotiation, an unusual job, a customer who is upset. An automatic reply there is faster and worse.

5. Measure the second clock. Most businesses can tell you when a lead came in. Few can tell you when it was first answered properly. Pull a month of inquiries and look at the median and the worst decile. Do this before you buy anything. If your median is already under an hour, the best money may not be in speed at all.

What we are not sure about

We do not know the right target for your business. A boiler repair, a wedding venue, and a software subscription differ in how long a person will wait. Evidence from sectors like yours, if it exists, would beat any number here. We would also like to read the HBR article's methods section in full, and a reader with access should check our summary against it. If we have a figure wrong, the error is most likely in those secondary numbers.

Three things we are fairly confident of. In that 2011 test, replying within the hour put a business among the fastest 37%, as the secondary sources report it.^[2][3] Replying the next morning after a night of silence is a gamble, and the plumber lost it. And the exact number of minutes matters less than whether the first reply is a real one.

Measure first, build second

Measuring needs a spreadsheet and a month of inquiries, and the five steps above work without us. If the median turns out to be the problem, Gonzo Digital builds a lead-response system for this gap: a first answer aimed at a minute or two (a design target, not measured live), in the customer's language, from an AI that says it is an AI, with a handoff to you for anything that needs a person. It is a fixed-price build, and we will talk through whether it suits your volume before you commit. Either way, you are welcome to talk to us.