How long a lead lives: what is actually known about response speed

Where “100 times” comes from

The claim that replying in 5 minutes instead of 30 raises your odds 100-fold is almost always credited to MIT, Harvard or Kellogg. Its source is different: the 2007 Lead Response Management report, copyright InsideSales.com. It was presented by Dave Elkington, head of InsideSales, and James Oldroyd at a MarketingSherpa conference on 16 October 2007 (report PDF).

The report has two parts. The first is a survey of 495 companies labelled Kellogg. It gave no significant answer on when to respond, and the authors say so (p. 19). The second is InsideSales’ own dialer data: 6 companies, three years, more than 15,000 leads and more than 100,000 call attempts (p. 20). Oldroyd was a visiting fellow at MIT Sloan at the time, but the report is not an MIT publication.

What was measured: the odds of reaching a person by phone drop 100-fold if you call at 30 minutes rather than 5. The odds of qualifying the lead drop 21-fold. Between 5 and 10 minutes the drops are 5-fold and 4-fold (pp. 21, 27). Sales were not measured — the authors say so themselves (p. 20). Each company defined “qualified” in its own way. On the chart on p. 26 the values reach 2×10¹⁹: these are model outputs, not observed rates. The report was never peer-reviewed.

And a detail nobody retells. In the FranklinCovey case on p. 29, calling back shoppers who abandoned their cart within the first 24 hours worked worst of all.

What HBR actually says

The second popular source is “The Short Life of Online Sales Leads” in Harvard Business Review, March 2011. Its authors are the same Oldroyd, Kristina McElheran and Dave Elkington of InsideSales (HBR).

The piece contains two different datasets, and they are often merged. The first is an audit of 2,241 US companies, each sent one test lead through its website. 37% replied within an hour, another 16% within a day, 24% after more than a day, and 23% never replied. The 42-hour average only counts those that replied within 30 days.

The second dataset is 1.25 million leads at 29 B2C and 13 B2B companies. Companies that tried to contact a customer within the first hour were nearly 7 times as likely to qualify the lead as those that tried an hour later, and more than 60 times as likely as those that waited a day or more. In HBR, “qualify” means having a conversation with a decision-maker. Not closing a sale.

HBR is a business magazine, not a peer-reviewed journal. The method and source of the second dataset are not described.

How companies actually respond

There have been many mystery-shopper checks since, and they show a similar picture.

• InsideSales, 2014. Test leads sent through the websites of 9,538 companies in 2013. 47% never replied. The median time to the first call was 3 hours 8 minutes, the mean 61 hours. 121 companies, about 1.3%, called within 5 minutes (report PDF). The leads were only sent between 8:00 and 17:00 — no nights, no weekends.
• Drift, 2017. 433 B2B software-as-a-service companies. 7% replied within 5 minutes; 55% did not reply within five business days (Drift’s original post). Drift sells website chat, which is worth keeping in mind.
• Pied Piper, 2026. Enquiries through the websites of 3,290 US car dealers. 51% gave a perfect response — fast, by email or text, plus a phone call. 14% did not personally respond to the customer (Business Wire).
• Yandex Real Estate, 2026. 219,226 first calls to the sales offices of Russian property developers from February to May 2026, plus 382 mystery-shopper calls. 14% of calls went unanswered. Missed calls were returned in 11% of cases, in Moscow in about 5%. Almost 27% of call-back requests never reached a conversation with a manager (summary of the study on Rambler).

The peer-reviewed study: speed is a signal

In 2026 Management Science published “Speed Is a Signal” by Einav Hart, Eric VanEpps, Ovul Sezer and On Amir (UC San Diego release on EurekAlert). The setting is different — hiring freelancers, not sales leads. But the method is stronger: 11.6 million conversations on Fiverr plus experiments with more than 8,600 participants.

In the Fiverr data, a one-hour delay in replying was associated with a 46% lower chance of being hired, a full day with roughly 90% lower. The effect held even when the client could see ratings and the text of the reply. The title is the finding: response speed by itself tells the recipient something about the sender.

The key conclusion for anyone automating sales: according to the authors, fast replies lost their advantage when the recipient believed the reply had been generated automatically or by AI. An instant “Thank you, your request has been received” is not a signal. The signal is a fast reply to the actual question, with a fast, real person behind it.

Pied Piper sees the same risk. When AI could not cope and handed a question to staff, the customer was twice as likely to get no personal reply. More on that in our article on car dealerships and the OEM CRM.

Figures that travel without a source

• “35–50% of sales go to whoever responds first.” The earliest trace is an InsideSales blog post of 19 July 2010: it says sales studies consistently show this, but names none. The figure is often credited to the 2012 Google and CEB report “Digital Evolution in B2B Marketing”, but it is not in it.
• “78% buy from whoever responds first.” Usually credited to a company called Lead Connect. We could not find a report with a method.
• “Harvard, MIT and Kellogg found 100× and 21×.” These are InsideSales dialer data from the 2007 report. HBR’s figures are different — 7 and 60 times.
• “7 times more likely to close.” 7 times more likely to qualify, 100 times more likely to reach someone. None of these sources measured closed deals.

What follows for a business

1. Measure your own numbers, not market averages. Export a month of first-response times and break them down by hour and day of the week. The average hides nights and weekends, which is where the hours pile up.
2. Count a reply to the question as the first response. An auto-confirmation does not count — both the Hart et al. study and the customer’s common sense say so.
3. Treat nights and weekends as a separate problem. InsideSales sent leads only during working hours and still got a median of over three hours.
4. Make the handoff to a human a metric of its own. A bot that replied fast and passed the question into a void makes things worse. You need to know how many minutes it took a manager to pick the lead up.
5. Fix one metric before a pilot starts. For example, the share of leads answered on substance within 15 minutes. How to choose the process and not fool yourself about the result, we covered in our article on “95% of AI pilots”.

What we don’t know

We found no peer-reviewed study that directly links lead response speed in B2B to closed deals. HBR and the 2007 report are US data from 2007–2011, and both are tied to InsideSales, which sold software for exactly these calls. We know “Speed Is a Signal” from the university release: the journal page did not open for us. It is also about hiring freelancers, not sales.

We read the Yandex Real Estate study in a summary, not the original. We have not seen Drift’s 2018 report, often quoted alongside the 2017 one: the original page is gone and the figures in retellings disagree. We found no measurements of response speed among companies in Uzbekistan. We have no numbers of our own yet.

What a fast, substantive reply looks like in practice — with qualification, booking and handover to a manager — is shown on Sales automation with AI.

If you want to know how long a lead lives at your company, describe your process in the questionnaire on our home page or message us on Telegram. The breakdown of one process takes 48 hours and is free.