Lead response time and what the research really shows

Speed to lead is real. The number everyone quotes for it is not, and the gap between those two sentences is where a lot of bad software gets sold.

I am spending the first half of this article on sourcing rather than tactics, because you are going to be shown the fake number in a sales deck, and knowing where it came from changes how you buy.

Where does the 21x figure actually come from?

A conference presentation, not a journal, and not Harvard.

The source is the Lead Response Management Study, presented by Dave Elkington of InsideSales.com and James Oldroyd, then a faculty fellow at MIT Sloan, at MarketingSherpa's B2B Demand Generation Summit on 16 October 2007. Its own description of the sample: three years of data across six companies, from over fifteen thousand leads and over one hundred thousand call attempts.

The lines everyone repeats come straight off those slides. The odds of contacting a lead if called in five minutes versus thirty drop 100x. The odds of qualifying one drop 21x.

Harvard Business Review contains no such number. Not in the 2011 article it gets attributed to, not anywhere.

Four things are wrong with the citation as it circulates. Six companies is a small sample for a claim this precise. It is proprietary vendor data presented at a marketing conference. It was never peer reviewed and the data was never released. And the "MIT study" branding rests on one faculty fellow's participation, not on institutional sponsorship.

There is a fifth problem that is almost funny. The domain that hosted the study for years, leadresponsemanagement.org, has rotted. It now carries paid outbound links to casino review sites. If a vendor cites that domain at you, they have not opened it.

None of that makes speed unimportant. It makes the multiplier unusable. Ask where it came from and watch what happens.

There is a general rule underneath this. A statistic that is very precise, very dramatic, and always attributed to a prestigious publication without a link is usually none of those things.

What does the real research say?

There is a genuine Harvard Business Review piece, and it is better than the myth.

Oldroyd, McElheran and Elkington published "The Short Life of Online Sales Leads" in HBR in March 2011. They audited 2,241 US companies to see how fast each responded to a web generated lead.

The results, in their words: 37% responded within an hour, 16% responded between one and 24 hours, 24% took more than 24 hours, and 23% never responded at all. The average response time among companies that answered within thirty days was 42 hours.

Separately, across a dataset of 1.25 million sales leads received by 29 business to consumer and 13 business to business companies, firms that tried to contact potential customers within an hour were nearly 7x more likely to qualify the lead as those that tried an hour later, and more than 60x more likely than companies that waited 24 hours or longer.

Two disclosures have to travel with that citation, and most people who quote it give neither.

First, the conflict of interest. Co-author David Elkington was founder and chief executive of InsideSales.com, and InsideSales.com's platform generated the data. That does not make the finding wrong. It makes it vendor research published under an editorial masthead, and you would want to know that about any other study.

Second, HBR is a magazine, not a peer reviewed journal, and this piece is a short "Vision Statement" feature. There was no external review of the method.

So here is the honest ladder.

Defensible: responding within an hour makes you roughly 7x more likely to qualify a lead than responding an hour later, and roughly 60x more likely than waiting 24 hours. Source: HBR 2011, 1.25 million leads, with the disclosure above.

Defensible with caveats stated out loud: five minute versus thirty minute response changes qualification odds by roughly 21x. Source: InsideSales and MIT, 2007, six companies, vendor data, never peer reviewed.

Not defensible: attributing the five minute, 21x, or 100x figures to Harvard Business Review. That sentence is simply false, and it is in a great many proposals.

Which number should actually change your behavior?

The 23%, and it is the one nobody quotes.

Nearly a quarter of audited companies never responded to a lead at all. Not slowly. Never. That is not a speed problem, it is an existence problem.

Think about where a lead can arrive: your main line, a form on the site, a message on your Google Business Profile, a text, a chat widget, a direct message on social. Each has a different owner, and at least one has no owner.

Before you optimize response time, find the channel where response time is infinite. That fix is free and it is bigger than anything you will buy. Routing leads from four channels without dropping any is the practical version of that audit.

How do you measure your own response time in an afternoon?

How fast do you actually need to be?

Inside the hour, reliably, on every channel. That is the defensible standard the research supports.

I would go faster on the phone specifically, and not because of a study. A caller who does not reach you is holding a phone with your competitors' numbers on it. There is no comparable delay tolerance because there is no gap between their intent and their next option.

There is also a hard reason the callback strategy is weaker than it looks. A TNS survey in July 2022 found that 75% of Americans never answer calls to their wireless phone from a number they do not recognize. Your callback is an unknown number. Resolving the request while the person is still on the line is worth far more than capturing a number and ringing back.

The dental data shows the same thing from the other end. Patient Prism's scoring of 11,552,668 calls found that of every 100 callers, 13 walked away without booking, 4.8 received a follow up call, and 0.7 came back and booked.

For web forms, an hour is the standard and same day is the floor. Consistency matters more than speed records. A business that responds within forty minutes every time beats one that responds in three minutes on Tuesday and never on Saturday. Nobody measures the Saturdays.

This is the whole case for automation on the phone. An AI receptionist answers within a ring at any hour, which makes the response time question moot for calls and lets you spend your human attention on the channels that still need it.

What are the rules if the fast response is an automated outbound call?

This is where speed to lead runs into telecom law, and it deserves a straight answer.

An AI agent answering a call the customer placed is not a robocall. Nobody dialed anybody. The moment you point the same technology outward, the analysis changes. The FCC's declaratory ruling of 8 February 2024 holds that calls made with AI generated voices are "artificial" under the Telephone Consumer Protection Act, which means they require the prior express consent of the called party. That ruling took effect on adoption.

So an automated outbound callback in an AI voice, made seconds after a form submission, is a different legal object from an AI agent answering your main line. It may be fine. It may need written consent depending on the message and on who you are. That determination is your attorney's, not mine and not your vendor's. What I can tell you is that vendors sell "instant AI callback" without mentioning any of this.

Two related items belong in the same conversation with counsel. Consent can be revoked by any reasonable method, including spoken words on a call, and revocations have to be honored across voice and text together rather than in separate systems.

Does faster always mean better?

No, and this is where speed to lead gets oversold.

Speed matters when the caller is choosing between vendors and the first competent response wins. Home services, emergency work, and most local trades sit squarely there.

Speed matters much less when the buying cycle is long or the decision involves several people. Responding in ninety seconds to a proposal request for a project that starts in six months signals eagerness, not competence.

And speed is actively harmful when it costs accuracy in a regulated setting. A dental practice answering a coverage question in ten seconds has broken a rule that a slower, correct process would have held. The limits on what a dental AI may say do not bend for response time, and a fast wrong answer is the most expensive kind.

Single variable thinking fails the same way elsewhere. Proximity does not automatically decide which contractor ranks first, and long pages sometimes convert better than short ones. The variable everyone fixates on is rarely the one doing the work.

What to do this week

List every channel a lead can arrive on, including the ones nobody checks, and put a person's name against each.

Run the thirty lead audit above and produce your own four bucket split.

Then test each channel yourself. Submit a form, send a Google Business Profile message, text your main line, fill in the chat widget. Time the response. Do it on a Saturday.

Whatever comes back slowest, or not at all, is your project. It will not be the one you expected.

Questions about where your response time is going?

Email me at eric@seod.com with how long it currently takes you to call back a web form lead, and the channels you accept leads on. I will tell you which part of that chain is eating the time and which channel is most likely to have no owner at all.

I do this myself, and I will also tell you if a statistic in a proposal you were sent is one of the fake ones. That check takes two minutes and it has saved people real money.

More on the surrounding pieces sits in the AI phone and lead response library.