What Speed to Lead Means, and Where the Five-Minute Rule Came From

What does speed to lead actually mean?

The time between somebody trying to reach your business and your business responding. Speed to lead is the elapsed interval from a customer’s inbound contact to your reply, measured across the whole path including routing and notification. The term is used slightly more broadly than lead response time, which usually refers only to the final reply.

That breadth is the useful part. Speed to lead is meant to capture everything that happens between submission and response, including the twenty minutes a form spent in a queue and the notification nobody saw. If you want to measure yours properly, that is covered separately in how to measure your real lead response time.

This article is about the number rather than the measurement, and specifically about the one number everyone repeats. The five-minute rule is the most cited and most misquoted statistic in this field, and it comes from a single document published in 2007 that very few of the people citing it have read.

Having read it, the honest summary is that the research is better than its critics assume and considerably narrower than its advocates claim. It is also more candid than its reputation: the authors state what they did not measure, admit they are guessing about the mechanism, and publish a case study that contradicts their own headline. All of that has been lost in fifteen years of retelling.

Where did the five-minute number come from?

From work by Dr. James Oldroyd, then a faculty fellow at MIT’s Sloan School of Management, carried out with the software company InsideSales.com and presented at a MarketingSherpa summit in October 2007. The full report is still publicly available.

It had two parts, and the first one failed. Oldroyd surveyed 495 companies across more than forty industries between June and September 2007, asking marketers and sales managers when they responded to web leads and how well it worked. The survey found nothing statistically significant about timing, for an instructive reason: the companies did not know. On several questions around four in ten respondents answered that they did not know or did not measure.

So they went looking for behavior instead of opinion. The second part analyzed three years of actual call data from six companies, covering more than 15,000 web leads and over 100,000 call attempts. Crucially, that data came from the InsideSales.com system itself, which is what made the analysis possible and also defines its limits.

Before leaving the survey behind, it contained one finding worth more to a small business than the famous multipliers. Companies that could not say how quickly they responded showed a markedly steeper decline in qualification rates than companies that could. Not knowing was itself associated with worse outcomes, over and above being slow. The plain reading is that businesses which never look at this number tend to be the ones with the worst version of it, and the cheapest first move available is simply to find out where you stand.

Those limits deserve stating plainly rather than as a gotcha. Six companies is a small sample. They were customers of a company that sells lead-response software, so they already ran outbound calling operations. And the sponsor had a commercial interest in the finding — the report closes by noting that the study shifted the company’s own positioning. None of that makes the data wrong. It does mean the population studied was B2B inside-sales teams dialing web-form leads in the mid-2000s.

It is worth saying what the report gets right, because the criticism lands harder from someone being fair. It is unusually candid for sponsored research: it says what it did not measure, admits the authors do not know why the effect exists, records that the pattern weakens at the level of individual companies, and prints a case study that contradicts its own headline. Most marketing research of any era does none of those things, and a great deal of what has been written about it since does none of them either.

What they found in that population was dramatic. The odds of reaching a lead called at five minutes rather than thirty differed by a factor of a hundred. The odds of qualifying differed by twenty-one times. Those two numbers are the entire five-minute rule.

What did the study measure, and what did it not?

It measured whether a salesperson reached the person and whether that person entered the sales process. It did not measure whether anybody bought anything. On this the report is unambiguous, stating that “this study did not address close ratios”.

That third figure is the one worth carrying away, and it is almost never quoted. The gap between five and ten minutes is already fivefold on contact odds. Anyone reassuring themselves that calling back within twenty minutes is close enough to five is, by the study’s own numbers, nowhere near it.

Two definitional details matter for interpreting any of this. A “contact” meant a call connecting to a live person for a set duration, and that duration was defined separately by each of the six companies, ranging from two to six minutes. “Qualified” likewise meant whatever each company already used it to mean. These are reasonable choices for real-world data and they mean the multipliers are not measuring one consistent thing across the sample.

Oldroyd himself flagged the biggest caveat in the report: the clear pattern only appears once several companies’ data is pooled, and individual companies vary considerably. A finding that emerges from aggregation and disappears at the level of a single business is exactly the kind that should be applied to your own operation with care.

Why is five minutes not a cliff?

Because it is the smallest interval they looked at. Having found a steep decline across the first few hours, the researchers went back and sliced the first three hours into five-minute segments to see the shape more precisely. Five minutes is the resolution of the analysis, not a boundary anybody detected in the data.

This distinction matters practically, because it changes what the finding tells you to do. If five minutes were a threshold, responding at 4:50 would be very different from responding at 5:10, and hitting the mark would be the objective. Since it is not, the actual finding is that the decay is continuous and brutally steep from the first minute onward, and every minute you remove is worth something.

The decline across hours in the source data is so severe that plotting it legibly is difficult — the first hour towers over everything after it, and by hour three the bars are barely visible. That shape is the finding. Speed matters enormously at the start and hardly at all later, which is why the difference between five and ten minutes exceeds the difference between four hours and eight.

There is a related finding in the same report that nobody quotes and that owners find liberating. Past roughly twenty hours, additional call attempts appeared to make contact and qualification worse rather than better. Whether that generalizes is unclear, and the authors did not claim to know why, but it argues against the instinct to keep dialing an old lead indefinitely.

How did the number get so badly misquoted?

By being absorbed into a different, more famous study. In 2011 Oldroyd published in Harvard Business Review with Kristina McElheran and David Elkington, auditing 2,241 US companies and finding an average first response of 42 hours. That article popularized the topic, and the 2007 multipliers gradually attached themselves to the Harvard name.

The two are separate pieces of research with separate findings. The 100 and 21 times figures are from the 2007 InsideSales work. The 42-hour average, the finding that nearly a quarter of firms never responded at all, and the seven-times-within-an-hour figure are from the 2011 HBR audit. Keeping them apart is the single most common correction needed on this subject.

A second distortion is more consequential. Because the 2007 study measured qualification, and qualification sounds like a step toward a sale, the multipliers get restated as sales figures. “Respond in five minutes and be 21 times more likely to close” appears constantly. The report says it did not examine closing, so that claim is not a summary of the research, it is an extrapolation somebody added later.

A third is the comparator. The finding is five minutes against thirty. It gets rendered as five against ten, five against sixty, or simply as “five minutes” with no comparison at all, which makes it unfalsifiable and useless. If a page cites the multiplier without the thirty-minute reference point, the author has not read the source.

That gives you a quick test for any page repeating this statistic, which is worth having because you will meet it constantly. Does it name the year and the researcher? Does it state the comparison rather than a bare number? Does it distinguish contact from qualification from closing? Does it mention that the data came from one vendor’s platform across six companies? A page failing all four is repeating something it heard, and whatever else it tells you about responding to leads deserves the same scepticism.

Does the finding hold up?

Directionally yes, and for a reason the report itself demonstrates rather than proves. Asked why speed mattered so much, the authors said openly that they were not certain and offered educated guesses: the person is still at their desk, their interest is at its peak, and a fast reply impresses them.

The most convincing explanation is competition, and the report contains an accidental demonstration of it. The company’s own president submitted a mortgage enquiry through a lead provider and received seven calls from seven different businesses. The first came at thirty minutes. The last came three days later. Everyone after the first was competing for somebody who had already spoken to someone else.

That reframes the whole idea usefully. Speed matters less because interest decays in a vacuum and more because you are in a race you did not know you had entered. Your customer contacted three or four businesses at once, and the effect is strongest exactly where that is most common: urgent work, commodity services, anything a person shops around for in one sitting.

So the honest generalization is narrower than the slogan. When somebody has actively asked several businesses for help with something they want now, answer first. When somebody has quietly abandoned a purchase, or is early in considering something discretionary, immediate pursuit can read as pressure. Knowing which situation your inquiries represent is worth more than any multiplier.

Whether a 2007 finding still describes 2026 is a fair question, and the answer cuts both ways. One of the study’s stated guesses was presence — the person had filled in a form at a desk and was sitting next to a phone. That is much less true now, when the same enquiry arrives from a phone on a sofa and the person may not answer an unknown number at all. Against that, people now expect replies faster than they did in 2007, not slower, and comparison shopping across several businesses in one sitting has become easier rather than harder. The mechanism has shifted from being near a phone toward being still engaged in the search, and the practical conclusion survives the change intact.

Why is five minutes the wrong target for a small business?

Because it was derived from teams whose entire job was dialing web leads, and it assumes somebody sitting at a phone with nothing more urgent to do. A tradesperson under a sink cannot hit it, knows they cannot hit it, and reasonably concludes the whole idea is not for them. That conclusion is the real cost of the way this statistic is promoted.

The way out is to stop treating speed to lead as one number. Split it into two clocks that behave differently and can be hit by different means. Time to acknowledgement is how long before the customer knows they have reached a real business that is dealing with them. Time to human contact is how long before somebody who can actually help them speaks to them.

Five minutes belongs to the first clock, and automation holds it easily. An hour is a realistic target for the second, and most small businesses can manage it with better routing alone. A customer who receives a useful answer within minutes and a call within the hour has had an excellent experience, and no part of it required anybody to answer a phone while on a ladder.

Splitting them also makes the race winnable. In the mortgage example, the first competitor called at thirty minutes. A business whose website answered the same person in ninety seconds was first, comfortably, without anybody being available to dial. Being first no longer requires being fastest with a phone.

One caution on the acknowledgement clock: it only counts if it does something for the customer. An automated “we have received your enquiry” stops no clock worth measuring, because the person has learned nothing and can do nothing with it. The acknowledgement has to answer something or ask something.

A useful one does three things in a couple of sentences. It answers whatever the person actually asked, or says honestly that it cannot and who will. It tells them what happens next with a time attached, because “someone will call you before five today” is a commitment and “we will be in touch shortly” is noise. And it asks one thing that makes the eventual human conversation shorter. Judged that way, a good acknowledgement is not a holding message at all — it is the first half of the conversation, and the call that follows is the second.

What should you actually aim for?

An answer within minutes and a human within the hour, measured separately, with the second target relaxed if your work is considered rather than urgent. Set both against your own data rather than a 2007 benchmark, and judge whether closing the gap is worth the cost using your lead value.

The arithmetic is quick. Multiply the leads you would plausibly recover by your lead value — the calculation is in how to calculate what one lead is worth — and compare it against what faster response would cost you. Routing fixes are usually free. Coverage costs either wages or software.

Where the gap sits in the acknowledgement clock, particularly outside working hours, this is what Quibora’s managed AI assistant and lead capture does. Quibora answers visitor questions from the business’s own website content within seconds at any hour, asks qualifying questions, captures contact details, and emails the lead with an AI-drafted follow-up ready to send. Plans run from $99 to $399 per month with a one-time setup fee.

Three limits worth stating, given what this article has argued. It holds the first clock, not the second — a human still has to call, and if you do not, a fast acknowledgement makes the eventual silence worse rather than better. It works on website inquiries and does nothing for your phone. And for considered, non-urgent purchases the value of speed is genuinely lower, so the case for paying for it weakens.

Above all, treat the five-minute rule as what it is: a real finding from a narrow study, pointing in a direction that is almost certainly right, attached to a number that was never meant to be a pass mark. Faster is better, the first minutes matter most, and the version of this advice that tells you to hit exactly five minutes or fail has read the headline rather than the research. Knowing that is worth more than the rule itself, because it lets you improve something real instead of abandoning a target you were never going to reach.

Common questions