The 21x figure everyone attributes to Harvard is not in the Harvard article. It comes from a 2007 study co-published by a company that sold software for calling leads faster, a fact the study states about itself with unusual candour.

The Harvard article is real, and it reports different numbers, from two different samples, which are also worth knowing. Getting the attribution right matters here, because the advice built on these figures is expensive to follow.

This page attributes each number to its actual source, states the problem both studies share, and then covers the part of this subject nobody optimises: the page the visitor lands on immediately after clicking submit.

Two studies, routinely merged into one

The 2011 business review article contains two distinct datasets, and the most famous statistic is in neither of them.

The audit. The authors audited 2,241 US companies, measuring how long each took to respond to a web-generated test lead. 37% responded within an hour. 16% within one to 24 hours. 24% took more than 24 hours. 23% never responded at all. Among companies responding within 30 days, the average response time was 42 hours.

The separate study, described as such in the article. 1.25 million sales leads received by 29 business-to-consumer and 13 business-to-business companies in the US. Firms that tried to contact potential customers within an hour of receiving a query were nearly seven times as likely to qualify the lead as those trying an hour later, and more than 60 times as likely as companies that waited 24 hours or longer.

What is not in the article. The 21x figure. Neither is the 100x figure that travels with it.

Where those actually come from. A 2007 study presented at an industry summit that October, whose executive summary states: the odds of contacting a lead if called in 5 minutes versus 30 minutes drop 100 times, and the odds of qualifying a lead if called in 5 minutes versus 30 minutes drop 21 times.

That study’s sample, verbatim. Three years of data across six companies that generate and respond to web leads, from over fifteen thousand leads and over one hundred thousand call attempts.

Why the confusion is understandable. The two publications share an author and a subject. They do not share a sample, a method or a finding, and citing one for the other’s numbers is the single most common error in this literature.

Attribution of the widely cited lead response time statistics to their actual source studiesTable attributing each of the widely cited lead response time statistics to the study that actually produced it, since three separate datasets are routinely merged into a single claimed source. The first dataset is an audit reported in a business review article published in March two thousand and eleven, in which the authors audited two thousand two hundred and forty-one United States companies by measuring how long each took to respond to a web-generated test lead; thirty-seven percent responded within an hour, sixteen percent responded within one to twenty-four hours, twenty-four percent took more than twenty-four hours, and twenty-three percent of the companies never responded at all, while the average response time among companies responding within thirty days was forty-two hours. The second dataset is described within that same article as a separate study, covering one point two five million sales leads received by twenty-nine business-to-consumer and thirteen business-to-business companies in the United States, finding that firms attempting to contact potential customers within an hour of receiving a query were nearly seven times as likely to qualify the lead as those attempting contact an hour later, and more than sixty times as likely as companies that waited twenty-four hours or longer. The third dataset, which produces the most frequently quoted figures, appears in neither of those and originates in a two thousand and seven study presented at an industry summit in October of that year, whose executive summary states that the odds of contacting a lead called at five minutes versus thirty minutes drop one hundred times and the odds of qualifying a lead drop twenty-one times, based on three years of data across six companies covering over fifteen thousand leads and over one hundred thousand call attempts. The two publications share an author and a subject but share no sample, method or finding, which makes citing one for the other’s numbers the most common error in this literature.Three datasets, one attributionThe 2011 article, dataset one: the audit2,241 US companies sent a test lead.37% replied within an hour · 24% took over 24 hours · 23% never repliedAverage response time: 42 hours, among those replying within 30 daysThe 2011 article, dataset two: “a separate study”1.25 million leads, 29 B2C and 13 B2B companies.Contact within an hour: nearly 7x as likely to qualify the leadAgainst waiting 24 hours or more: more than 60xThe 2007 study, which is where the famous numbers actually live”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.”Sample: three years, six companies, 15,000+ leads, 100,000+ call attempts.Co-published by a company selling lead response software. Not in the 2011 article at all.
The 21x is not in the 2011 article. It is from a 2007 study by a company selling faster dialling. Source : HBR March 2011; Lead Response Management, 2007 (2011)
Response times recorded when two thousand two hundred and forty-one companies were sent a test leadChart presenting the response times recorded in an audit in which two thousand two hundred and forty-one United States companies were each sent a web-generated test lead and the time taken to respond was measured. Thirty-seven percent of companies responded within one hour. Sixteen percent responded within one to twenty-four hours. Twenty-four percent took more than twenty-four hours to respond. Twenty-three percent of the companies never responded at all. Among the companies that did respond within thirty days, the average response time was forty-two hours. This audit is methodologically the strongest element of the lead response literature, because unlike the associational findings about qualification rates it was produced by direct measurement: the researchers sent the leads themselves and observed what happened, so there is no question of selection among the companies observed and no confound between response speed and any other organizational characteristic. The finding that almost a quarter of companies never responded at all is also the most immediately actionable, because any individual business can determine within a single afternoon whether it belongs to that group by sending itself a test lead from a personal address and timing the reply, and businesses that discover they are in that group almost invariably discover it this way rather than through analysis.2,241 companies were sent a test leadMeasured directly by the researchers, not reported by the companies.Replied within an hour37%Replied in 1 to 24 hours16%Took more than 24 hours24%Never replied at all23%Average, among those who replied at all42 hours, counting only replies within 30 days.Why this is the best number hereThey sent the leads. Nothing is confounded.You can find out which group you are in this afternoonSend yourself a test lead from a personal address and time the reply. That is the whole method.
Almost a quarter never replied. That was measured directly, and it is the most useful number in the whole literature. Source : The Short Life of Online Sales Leads, HBR (2011)

Who published it, and what they said about why

This is not a hidden conflict. It is disclosed, in the source document, more openly than in the journal article.

The 2011 article’s disclosure. None as such. It carries a biographical line identifying one of the three authors as the chairman and chief executive of a lead response software company, presented at the same level as the two academic affiliations, with no note that the company sold precisely the capability the article recommends.

The 2007 document’s disclosure. Considerably franker, and inside the research document itself: that the company is a leader in lead management and the pioneer of lead response management solutions, and that this study caused a significant shift in our corporate positioning. It goes on to state that companies typically see a 2 to 4 times increase in contact ratios and lead qualification rates using its technology.

The later report from the same publisher. Describes its principal researcher’s role as conducting research in order to promote the company’s position of innovation and leadership. That is an unusually direct statement of what the research function was for.

How to read all this. Not as fabrication. The data may well be accurate. But it was collected and published by a party with a direct commercial interest in one conclusion, and that belongs in any citation of it.

What is missing from the field. I could not find an independent, peer-reviewed, recent study of B2B lead response time with a declared sample. Everything current comes from software vendors, which is the same structural problem twenty years later.

The problem both studies share

Beyond the sourcing, there is a design issue neither document addresses.

Neither is experimental. No leads were randomly assigned to fast or slow follow-up. Both observe what companies already did.

The confound. A company that responds in five minutes is a company with a staffed sales desk, a rota, a working routing system and someone whose job it is to watch the queue. That company will also qualify leads better for a dozen reasons unrelated to speed.

What that means. Response time is partly a measurement of sales operations maturity. Some of the association between speed and qualification is the maturity, not the speed.

What neither document does. Discuss it. The 2011 article is descriptive and offers informal explanations. The 2007 document notes only that the patterns appear clearly when data from several companies is combined.

How much this should change your behaviour. Very little, and here is why: the practical recommendation survives the critique. A buyer who submitted a form is comparing suppliers now. Arriving while the question is open is obviously better than arriving after it closed. You do not need a 21x multiple to justify that.

What it should change. How you cite it. Stop attributing a software vendor’s 2007 finding to a business school, and stop presenting an observational association as a causal law.

Elements of the lead response literature that survive methodological scrutiny and those that do notTable separating the elements of the lead response time literature that survive methodological scrutiny from those that do not. What survives is the practical recommendation itself: a buyer who has submitted a form is comparing suppliers at that moment, so arriving with a credible reply while the question remains open is self-evidently better than arriving after it has closed, a proposition requiring no statistical multiple to justify. Also surviving is the audit finding that a substantial proportion of companies respond slowly or not at all, since that measurement was taken directly by sending test leads to two thousand two hundred and forty-one companies and observing what happened. What does not survive is the attribution, since the most quoted figures of twenty-one times and one hundred times originate in a two thousand and seven study co-published by a lead response software vendor rather than in the two thousand and eleven business review article to which they are routinely credited. Also failing is the causal reading, because neither study randomly assigned leads to fast or slow follow-up and both merely observe what companies already did, leaving unaddressed the confound that a company capable of responding within five minutes possesses a staffed sales desk, a rota, a functioning routing system and a person whose responsibility is watching the queue, all of which would independently improve qualification rates. Response time therefore functions partly as a measurement of sales operations maturity, and some portion of the observed association between speed and qualification is attributable to that maturity rather than to speed itself. Neither document discusses this confound: the two thousand and eleven article is descriptive and offers informal explanations, while the two thousand and seven document notes only that the patterns appear clearly when data from several companies is combined.What survives the critiqueSurvivesThe advice itself.A buyer who filled in a form is comparingsuppliers now. Arrive while it is open.The audit findings.2,241 companies were sent a test lead and23% never replied. That was measured.Does not surviveThe attribution.21x and 100x are from a vendor study,not from the business review article.The causal reading.Nothing was randomly assigned. Bothstudies observe what firms already did.The confound neither document discussesA company that answers in five minutes has a staffed desk, a rota and a routing system. It wouldqualify leads better for a dozen reasons that have nothing to do with speed.Keep the behaviour. Change how you cite it.
The advice survives. The multiple, the attribution and the causal reading do not. Source : HBR March 2011 and Lead Response Management, 2007 (2011)

Where the delay actually comes from

Before buying software to respond faster, find out why the reply is slow. It is rarely what people assume.

The notification goes nowhere. The form emails an address that belonged to someone who left, or a shared inbox nobody opened since the reorganisation. This is the most common cause and the cheapest to fix.

The routing rule broke silently. A rule assigning leads by territory or product stopped matching when a value changed. Nothing errors. Leads simply queue.

Nobody owns the queue outside business hours. Which is fine, and should then be stated on the confirmation page rather than left as an implied promise you break every evening and weekend.

The lead arrives without context. The salesperson receives an email address and nothing else, so the first action is research rather than contact. A single qualifying field on the form removes that step.

The reply requires an approval. Pricing, availability or scope needs someone else, who is in a meeting. This is an organisational constraint dressed as a speed problem.

And the honest one. There is no agreed owner. Marketing thinks sales has it, sales thinks it is a marketing nurture lead, and the buyer waits. No dialler fixes this.

Common causes of slow lead response and whether software addresses each oneTable listing six common causes of slow response to an inbound lead and indicating whether purchasing lead response software addresses each one. The first cause is that the form notification is delivered to an address belonging to someone who has left the organization or to a shared inbox that nobody has opened since a reorganisation, which is the most frequent cause and the least expensive to correct, and which software does not address. The second cause is that a routing rule assigning leads by territory or product silently stopped matching when an underlying value changed, so no error is raised and leads simply accumulate in a queue, which software does not address. The third cause is that nobody owns the queue outside business hours, which is a legitimate arrangement but should then be stated explicitly on the confirmation page rather than left as an implied promise broken every evening and weekend, and which software does not address. The fourth cause is that the lead arrives carrying no context, so the salesperson receives an email address and nothing else and the first action becomes research rather than contact, a problem removed by adding a single qualifying field to the form rather than by software. The fifth cause is that the reply requires an approval concerning pricing, availability or scope from a person who is unavailable, which is an organisational constraint presented as a speed problem and which software does not address. The sixth cause is that no owner has been agreed, with marketing believing sales has responsibility and sales believing the submission is a marketing nurture lead while the buyer waits, which software also does not address. Only the mechanical acceleration of dialling, which presupposes that all six preceding conditions are already satisfied, is what such software actually provides.Why the reply is slowCauseDoes software fix it?The notification goes to a dead inboxNo. Most common cause.A routing rule broke silentlyNoNobody owns the queue outside hoursNo. Say so on the page instead.The lead arrives with no contextNo. One form field does.The reply needs an approvalNo. Organisational.No agreed owner between teamsNoFaster dialling presupposes all six are already solved. Check them before you buy anything.
Most of them are routing, ownership or approval problems. Software addresses the last of the six. Source : Method (2026)

The page nobody optimises

Everything above is about what happens after the visitor leaves. This is about the two seconds before they do, and it is the cheapest unclaimed asset on most B2B sites.

What it usually says. “Thanks. We will be in touch shortly.” Sometimes just “Thanks”.

What that moment actually is. The only point in the entire funnel where someone has just chosen to act, is still on your site, and is giving you undivided attention. You paid for that attention twice: once for the click, once for the conversion.

What the page should confirm. What was received. If they requested a specific thing, name it back to them, so they know the right form went to the right place.

What it should promise, in specific terms. Not “shortly”. A named window: today if before a stated hour, otherwise the next working day. Vague promises produce chasing emails, which cost your team more than the specificity would have.

What it should offer immediately. A way to skip the queue. A booking link on the confirmation page lets the buyers who are ready right now put a meeting in the calendar without waiting for anyone. This is the single highest-value element on the page and most companies do not have it.

What it should deliver without a second wait. If you promised a document, put it on the page. Requiring an email round trip to receive something they already gave you their address for adds a failure point and a delay for no gain.

And what it should not do. Ask for anything else. The visitor has just given you something. Asking for a second thing immediately is the worst-timed request on the site.

Contents of a typical form confirmation page compared with what that page could usefully containComparison of what a typical form confirmation page contains with what such a page could usefully contain, on the basis that this moment is the only point in the funnel at which a person has just chosen to act, remains on the site, and is giving undivided attention that the business has paid for twice, once for the click and once for the conversion. A typical confirmation page displays only a brief acknowledgement such as thanks, sometimes accompanied by a vague promise to be in touch shortly. A useful confirmation page performs five functions. It confirms what was received, naming the specific item requested back to the visitor so they know the correct form reached the correct destination. It states a specific response window rather than an indefinite one, for example the same day if submitted before a stated hour and otherwise the next working day, since vague promises generate chasing emails that cost the team more than the specificity would have. It offers an immediate means of skipping the queue, typically a meeting booking link, which allows buyers who are ready at that moment to place a meeting in the calendar without waiting for anyone, and which constitutes the single highest-value element available on the page although most companies do not include it. It delivers without a second wait anything that was promised, placing a requested document directly on the page rather than requiring an email round trip to receive something the visitor has already supplied their address to obtain, since that round trip adds both a failure point and a delay for no benefit. Finally it asks for nothing further, since the visitor has just given something and an immediate second request is the worst-timed request on the site.The cheapest unclaimed asset on your siteWhat most pages say”Thanks.""We’ll be in touch shortly.”And that is the whole page.What it could doConfirm what was received, by nameGive a specific window, not “shortly”Offer a booking link to skip the queueDeliver the promised document here, nowAsk for nothing elseThe booking link is the one that paysBuyers who are ready right now can put a meeting in the calendar without waiting for anybody.It costs nothing to add and most B2B sites do not have it.Why “shortly” is expensiveAn unspecified promise produces chasing emails from the buyer, and those cost your team morethan writing “today, if you sent this before 4pm” would have.
You paid for that attention twice. Most sites spend it on the word 'Thanks'. Source : Method (2026)

What to fix, in order

Five things, cheapest first, none of which requires believing any particular multiple.

Send yourself a test lead. Today, from a personal address, as a stranger would. Time the reply. The 2011 audit found 23% of companies never replied at all, and every company that discovers it is in that group discovers it this way.

Put a booking link on the confirmation page. One line of work, and it converts the subset of buyers who were ready now and would otherwise have waited for you.

Replace “shortly” with a specific window. Then make sure the window is one you actually meet, because a missed specific promise is worse than a vague one.

Check where the notification goes. A surprising share of slow responses are not a staffing problem. They are a form emailing an address nobody monitors, or a routing rule that broke when someone left.

Then, and only then, argue about five minutes versus an hour. That debate is only worth having once the reply reliably happens at all.

Where to go next

You are deciding what the form should ask. How many form fields.

You want the page anatomy in detail. Anatomy of a high-converting B2B landing page.

You are deciding how many pages to run. How many landing pages.

Your conversion rate is the number in dispute. Conversion rate and its denominator.

Your cost per lead is under scrutiny. Cost per lead.

Your bounce rate looks wrong. GA4 bounce rate.

In short

  • The 21x figure is not from the 2011 business review article. It is from a 2007 study co-published by a lead response software vendor.
  • That study’s sample: three years, six companies, over 15,000 leads and over 100,000 call attempts.
  • The 2011 article reports different numbers: an audit of 2,241 US companies, and a separate study of 1.25 million leads across 42 companies.
  • From the audit: 37% replied within an hour, 24% took over 24 hours, 23% never replied, and the average was 42 hours.
  • From the separate study: nearly 7x more likely to qualify when contacting within an hour, more than 60x against waiting a day.
  • The commercial interest is disclosed in the 2007 document itself, which states the study caused a significant shift in its own corporate positioning.
  • Neither study is experimental, and neither addresses the confound that fast responders have better-resourced sales operations.
  • The advice survives anyway, and the confirmation page is where you can act on it today.

Send yourself a test lead this afternoon. Book a diagnostic, or see how we approach B2B websites.