The five-minute lead rule: what it rests on, and what 326 Calgary sites do

One document. “Lead Response Management”, presented at a MarketingSherpa summit on October 16, 2007, and copyrighted on every page after the cover by InsideSales.com, a sales-software vendor. The report calls its second part “the MIT Study”, but it is a vendor document, not an MIT or Harvard publication. The lead researcher, James Oldroyd, was a faculty fellow at MIT’s Sloan School when the second part was done.

The report contains three separate studies, which is the first thing that gets lost. A survey of 495 companies run through Kellogg between June and September 2007, self-reported by sales and marketing managers. A data study of six companies, three years, more than 15,000 leads and over 100,000 call attempts. And a case study with FranklinCovey on abandoned carts. The famous numbers come from the middle one, and the report states where its data came from: “We used the data from the InsideSales.com system.” Six of the vendor’s own customers.

Here are the headline claims, quoted from the summary on page 21:

“The odds of calling to contact a lead decrease by over 10 times in the 1st hour. The odds of calling to qualify a lead decrease by over 6 times in the 1st hour.” And: “The odds of contacting a lead if called in 5 minutes versus 30 minutes drop 100 times. The odds of qualifying a lead if called in 5 minutes versus 30 minutes drop 21 times.”

That is the rule, in its original words. Note what is being compared: five minutes against 30 minutes, and contacting or qualifying a lead, not closing a sale.

This is the part that matters most, and it takes one table to see.

Every chart supporting the decay claims carries a title in the same form: “Initial Dials to Leads that become Contacted”, or “Initial Dials to Leads that become Qualified”. The report never says whether these bars are counts or rates, and the same title sits over day-of-week bars between 1.59 and 2.38 and a time-of-day axis that runs to 350 million. The hourly contact chart on page 25 plots these values, redrawn here at the same scale.

Divide the first by the second. 12,700 divided by 1,190 is 10.7. That is the “over 10 times”. The qualify chart on the same page runs 10,384 then 1,680, and 10,384 divided by 1,680 is 6.2, which is the “over 6 times”.

So the multiples match ratios of one bucket’s value to the next bucket’s value. To turn a count into an odds or a rate you need a denominator: how many dials were attempted in each time bucket. The report publishes no denominator anywhere. We searched the full text for one.

This matters. If the bars are counts, and most calls to a new lead are placed soon after it arrives, the first hour would be large and the ninth hour tiny whether or not speed changes any individual lead’s chance of being reached. Without a denominator a reader cannot rule that out, and a falling bar is not evidence, on its own, that the fifth minute beats the thirtieth.

Two more things we found by reading rather than citing. There is a chart on page 26 whose y-axis runs to 2 times 10 to the 19th, from a dataset the report describes as “over one hundred thousand call attempts”. Values in the quintillions cannot be counts from that data. And although the section twice calls its results “statistically significant”, it reports no inferential statistics: no model, no p-value, no confidence interval, no control variable. The word “odds” is used without one.

The authors were more careful than the people quoting them. On whether the pattern generalizes, page 30: “Dr. Oldroyd emphasizes that he finds these clear patterns in the data only when data from several companies is combined together. Patterns vary significantly from company to company.”

On why speed would matter at all, page 31: “And finally, WHY is response time so important? Though we don’t exactly know for sure, we have a few educated guesses.”

And then there is the third study in the same document, the FranklinCovey case on page 29, which found the opposite of the rule the document is famous for: “we found that the worst time to call back was within the first 24 hours. People often opted out because they didn’t have their credit card handy or they decided against the purchase for some reason. If reps called back too quickly they would often get a negative emotional response.”

One more limit worth knowing before anyone builds a sales process on this. Page 20: “This study did not address close ratios.” The outcomes measured were contact and qualification, and both were defined differently by each of the six companies. A contact was a call lasting long enough to count, and the footnote says the threshold “was different for each company’s data ranging from 2 minutes to 6 minutes.”

The other citation people reach for is a Harvard Business Review piece from March 2011. It is a short magazine item, not a research paper, and it has no digital object identifier. It reports two things, both quoted here.

The audit: “We audited 2,241 U.S. companies, measuring how long each took to respond to a web-generated test lead. Although 37% responded to their lead within an hour, and 16% responded within one to 24 hours, 24% took more than 24 hours” and “23% of the companies never responded at all. The average response time, among companies that responded within 30 days, was 42 hours.” The figure is 42. A separate 2012 InsideSales.com summary of its own audits gives “nearly 47 hours”, from different data.

The conversion study: 1.25 million leads across 29 business-to-consumer and 13 business-to-business companies. Firms that tried to make contact “within an hour” were “nearly seven 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.

Read those two paragraphs next to each other and the drift is obvious. Different sample, different companies, different definition of qualifying, and a comparison of one hour against two. None of it is about five minutes. The article closes: “We’re conducting further research to more fully understand the causes and identify possible solutions.” We found no published follow-up.

Almost nothing, and that is the finding. We searched Crossref, OpenAlex and Semantic Scholar for any peer-reviewed study estimating the effect of first-response speed on inbound lead conversion. We found none, and no journal version of this work by either academic author.

The closest thing is Canadian. Ohiomah and colleagues at the University of Ottawa, in the Journal of Business Research, note a lack of research on how IT use affects inside sales performance, a neighbouring question rather than response speed itself. Their own study surveys 108 sales professionals and measures follow-up intensity rather than speed.

What does exist, and is genuinely peer reviewed, is the literature on waiting. Taylor’s 1994 paper in the Journal of Marketing on delays and service evaluation. Zohar, Mandelbaum and Shimkin in Management Science in 2002, who found that customers’ patience adapts to the delay they have learned to expect, which is a direct problem for any claim that a buyer’s willingness to wait is a fixed number. And the standard call-centre review by Gans, Koole and Mandelbaum. None of it says five minutes.

Every version of this advice assumes the business can act on it. That assumption is testable, so we tested it.

We took the frame of Calgary business websites behind our earlier studies, 395 sites drawn from curated “best of Calgary” lists across restaurants, gyms, cafes, contractors, dental, medical aesthetics, law, auto, mortgage broking and retail. For each one we fetched the homepage and up to two contact or booking pages as served, with no JavaScript run, and recorded what a visitor is offered. 326 sites answered.

Eighteen sites out of 326 tell a customer when to expect a reply. Another 39 say something like “we’ll get back to you shortly”, which is a tone, not a commitment. The rest say nothing at all, and the customer is left to guess.

The number we keep coming back to is the 22.4%. Nearly a quarter of these businesses run an enquiry form that never asks for a phone number. They could not call a lead back in five minutes, or five hours, because they did not collect the means to do it. The advice is unreachable from where they are standing, and no amount of urgency fixes it. The median enquiry form asks for six fields, so this is not a case of ruthless minimalism; the phone field simply is not there.

The 8.3% is the other end of the same idea. A booking widget removes the question entirely. The customer does not wait for a reply because there is nothing to wait for. Eight in a hundred of the Calgary sites we measured already have one.

Speed has a legal shape in Canada, and it cuts in a useful direction.

Phoning a web enquiry fast is lightly regulated. Canada’s Anti-Spam Legislation applies to commercial electronic messages, and section 6(8)(a) puts a live voice call outside it. On the telemarketing side, section 41.7(1)(b) of the Telecommunications Act, read with the definition of an existing business relationship in section 41.7(2), exempts a call to someone who made an enquiry in the previous six months, and has not asked you not to call, from the National Do Not Call List. Business numbers sit outside that list altogether. You still have to identify yourself and the purpose of the call at the start under section 41.7(3), and you still have to keep your own internal do-not-call list under section 41.7(4), where, under the National Do Not Call List rules, a number must stay for three years.

Emailing or texting fast is the regulated act. An enquiry gives you implied consent for six months under section 10(10)(e), not indefinitely. Every message needs the prescribed identification, contact details that stay valid for 60 days, and a working unsubscribe, and you have ten business days to honour an opt-out.

The practical read: if you want to move fast, pick up the phone. It is the least regulated channel and the one your form is least likely to have equipped you for.

Collect a phone number. This is the whole post in three words for nearly a quarter of the businesses we measured. Add the field, make it optional if you must, and you convert a channel you cannot use fast into one you can.

Publish what you actually do. Eighteen sites out of 326 say when they will reply. “One business day” is a promise you can keep, and it beats silence and it beats “shortly”. If you cannot keep one, that is worth knowing before a customer finds out.

Prefer removing the wait to shortening it. Where a booking widget suits the work, the fastest response is no response required. That is a structural fix rather than a speed target.

Treat speed as one variable, not the variable. The evidence for calling fast is a vendor report with no denominators and a magazine page about a different comparison. That is not nothing, and answering promptly is obviously better than not answering. It is a long way from a law of nature, and the same report contains a case where calling fast made things worse.

If you test it, test it properly. The way to know whether speed moves your numbers is to vary it deliberately and compare, not to read a slide. Route half your enquiries to an immediate call and half to a scheduled callback and look at closed business, not contact rates.

The 2007 report was read in full from the copy hosted on MarketingSherpa’s content server, and every quotation carries its page number. The Harvard Business Review quotations were checked against the text on the publisher’s own page in October 2026; the citation is volume 89, issue 3, March 2011. Searches for peer-reviewed work were run against Crossref, OpenAlex and Semantic Scholar, most recently on OpenAlex in October 2026. Legal provisions are quoted from the Justice Laws website. We are marketers, not lawyers; check the sections before a process relies on them.

Our own measurement was collected in September 2026. The frame is 395 Calgary business websites assembled for our earlier studies from curated listicles (those studies count 403 entries because eight domains appear twice), which is a convenience sample and not a random draw from Calgary businesses: sites that appear on “best of” lists are plausibly better resourced than average, so treat these figures as a favourable reading. 326 sites returned a usable page; 69 did not, mostly bot blocks and dead domains. We fetched the homepage and up to two contact or booking pages and read the HTML as served, so anything rendered only by JavaScript is invisible to us and a site could offer chat or a form we did not see. Detection is pattern matching: forms that are only a search box or an email-only newsletter signup were excluded, and chat and booking widgets were detected from the embed of 16 chat providers and 14 booking providers, so a self-built widget would be missed. Every response promise the detector flagged was read by hand; 20 were detected and two were rejected, one because the sentence was a vendor’s own marketing copy and one because it sat inside a customer testimonial rather than being a promise by the business. The frame also still contains a small number of non-Calgary and non-business sites that survived the original exclusion list. The dataset, the method and the exclusions are published with this post: the full result table is metrics.csv and the method and limits are method.json.

• Your contact form is quietly losing you leads: the build-side problems that stop an enquiry ever arriving.
• 41% of Calgary businesses cannot be phoned in one tap: the earlier study on the same frame.
• The decoy effect, and four more pricing claims retested: the same treatment applied to pricing folklore.
• Five marketing myths that peer-reviewed research says are wrong: more claims that do not survive their sources.