# Speed to lead: why the first five minutes matter

- URL: https://plandora.com/blog/speed-to-lead-with-a-digital-mind/
- Authors: Kit (Co-founder)
- Published: 2026-10-22
- Updated: 2026-10-22
- Facts checked: 2026-10-22
- Category: Growth

> Speed to lead is the time between an enquiry and your first response, ideally a useful reply. The 21-times figure compares the odds of qualifying web-form leads when the first phone call came within 5 minutes rather than 30, at six companies, in a study presented in 2007. It did not measure sales or AI chat. Speed helps only when the reply is accurate and hands over cleanly.

## Key takeaways

- The 21-times figure compares the odds of qualifying web-form leads when the first phone call came within 5 minutes rather than 30, at six companies (presented 2007). It does not measure closed sales or AI chat.
- The studies differ in metric and population (21-fold, nearly seven-fold, eight-fold), so their numbers should never be combined.
- An automatic acknowledgement is not a useful reply. Track time to first attempt and to first useful reply separately.
- Measure your own median for both, split by in-hours and out-of-hours, before setting a target.
- A Digital Mind can answer, ask a qualifying question, offer booking and hand over with context where your channels are connected. It is not the human expert.

Speed to lead is the time between an enquiry and your first response. The best-known research, presented in 2007 and covering six companies, found the odds of qualifying web-form leads falling as the first phone call slowed (its [21-times figure](https://www.mortech.com/hs-fs/hub/25649/file-13535879-pdf/docs/mit_study.pdf) compares 5 and 30 minutes). It measured phone calls, not closed sales, chat or AI representatives.

## What does speed to lead mean?

Speed to lead is the elapsed time between a person asking and your business first responding. The studies below mostly time the first attempt, such as a call, connected or not. This article also recommends tracking the first useful reply: one that answers the question, or asks one specific question that moves the enquiry on. A receipt or a generic greeting does not count. Three choices matter.

- **When the clock starts.** When an enquiry is submitted or a message arrives, not when someone opens it. A scan or a chat open is not an enquiry until the visitor writes.
- **When it stops.** Report the first attempt and the first useful reply as two numbers.
- **Which hours count.** All of them. An enquiry sent at 9 pm waits until morning on a team that works nine to five. The 2014 audit below sent test leads only between 8 am and 5 pm, so its benchmark is an in-hours figure.

A form submission starts the clock but rarely a conversation, so the choice between [lead forms and conversations](/blog/lead-forms-vs-conversations) affects speed.

## What did the five-minute research actually measure?

The 21-times figure comes from the InsideSales.com/MIT Lead Response Management Study. InsideSales.com's chief executive, David Elkington, and James Oldroyd of MIT's Sloan School of Management [presented it on 16 October 2007](https://www.mortech.com/hs-fs/hub/25649/file-13535879-pdf/docs/mit_study.pdf) at a business-to-business demand-generation summit. The summary describes three years of data from six companies: more than 15,000 web-form leads and more than 100,000 call attempts. It does not give the calendar years, the companies' industries or whether their customers were businesses or consumers.

Two outcomes were counted. A "contact" was a call that connected with a live person for a defined length of time. A lead counted as "qualified" when it was willing to enter the sales process, in some cases by setting an appointment, and each company defined that in its own way. Headline results: the odds of making contact fell 100-fold, and the odds of qualifying fell 21-fold, when the first call came after 30 minutes instead of within 5. The summary states its own limit: "This study did not address close ratios."

The table compares later work.

| Source | Published | What was counted | Headline result |
|---|---|---|---|
| [InsideSales.com/MIT Lead Response Management Study](https://www.mortech.com/hs-fs/hub/25649/file-13535879-pdf/docs/mit_study.pdf) | Presented October 2007 | Three years of data (years not stated), 6 companies, 15,000+ web-form leads, 100,000+ call attempts. Qualified was defined by each company | Odds of qualifying 21 times lower when the first call came at 30 minutes than at 5. Odds of contact 100 times lower |
| [Harvard Business Review, lead decay study](https://web.archive.org/web/20111113015409/http://hbr.org/2011/03/the-short-life-of-online-sales-leads/ar/pr) (archived print view) | March 2011 | 1.25 million leads, 29 B2C and 13 B2B US companies. Qualified meant a meaningful conversation with a key decision maker | Contact within an hour was nearly 7 times as likely to qualify as an attempt an hour later, and more than 60 times as likely as one after 24 hours or more |
| Harvard Business Review, response audit (same article) | March 2011 | 2,241 US companies audited with a web-generated test lead | 37% replied within an hour, 23% never replied. Average 42 hours among those that replied within 30 days |
| [InsideSales.com 2014 Lead Response Report](https://resources.insidesales.com/wp-content/uploads/2019/11/2014-Lead-Response-Report.pdf) | Tests run in 2013 | 9,538 companies that received a test lead | 47% never responded. Median first phone response 3 hours 8 minutes (vendor-reported) |
| [InsideSales.com Lead Response Study 2021](https://www.insidesales.com/response-time-matters/) | 2021 | Reviewed "over 55 million sales activities on 5.7 million inbound leads at 400+ companies" | Conversion "8x greater" within 5 minutes than between 5 minutes and 24 hours. Conversion is not defined (vendor-reported) |

A 21-fold, a nearly seven-fold and an eight-fold figure describe different populations, time windows and outcomes, so none can be averaged, converted or stacked into one rule. InsideSales.com published or co-wrote every source above, and [HBR's page](https://hbr.org/2011/03/the-short-life-of-online-sales-leads) lists its co-author Elkington as the company's chairman and CEO. That page showed only the abstract when checked, so the HBR figures here come from the archived print view. None of the sources checked independently replicates the five-versus-thirty-minute multiplier.

## How quickly do businesses actually reply?

HBR's audit also found that 16% of companies replied within one to 24 hours and 24% took longer than a day; it does not say when the audit ran. InsideSales.com's 2014 report audited companies in 2013, with test leads sent only between 8 am and 5 pm. Among companies that replied by phone, the median first call came after 3 hours 8 minutes and the average after 61 hours 1 minute. The 23% and 47% no-reply rates come from different samples and methods, so they show no trend, and the 2014 figures are vendor-reported.

The 2021 page says only 0.1% of inbound leads were engaged in under five minutes and 57.1% of first call attempts came after more than a week, without saying how leads or engagement were counted. Treat these as evidence that slow replies were common in those samples (US companies for HBR and the 2014 audit, unstated for 2021), not a description of your market today.

## Why can't you treat 21 times as a promise?

- **Qualified is not sold.** HBR defines it as "having a meaningful conversation with a key decision maker", and the 2007 summary excludes close ratios.
- **It is a record, not an experiment.** The data came from a call-tracking system, and nothing in the summary says leads were delayed on purpose. Differences between fast and slow responders, such as which leads they chased first, may explain part of the gap.
- **The channel is the phone, or unstated.** The 2007 study counted calls by sales representatives to web-form leads, and HBR does not say how contact was attempted. None of the studies in the table tests chat, messaging or AI replies.
- **The source is a conference summary.** The page publishes charts, not the underlying data or the statistical model, and the study is nearly twenty years old.
- **The seller wrote it.** InsideSales.com sells tools for speeding up lead follow-up.

What survives is the direction. HBR's separate 1.25-million-lead study points the same way under a different definition of qualified. In these datasets, longer delays are associated with lower qualification rates, with the steepest drops in the first hours, but they do not show how much the delay itself caused. Only your own data can show that for your business.

## How much demand arrives outside office hours?

Salesloft's [Conversation Trends Report](https://assets.contentstack.io/v3/assets/blt48e9deb40787970b/blt10355646adde4b3e/66abcfd1eb20b48bb62d2355/Conversation_Trends_Report.pdf) (copyright 2024) analysed more than 30 million conversations across all instances of Drift, its chat product, during 2023. For the East Coast part of its customer base it reports that "39% of all conversations happened outside of normal business hours", with most still falling between 9 am and 5 pm Eastern. Within that group, 41% of meetings booked through Drift happened outside 9 to 5.

That is vendor data about B2B website chat from one group of customers. A [Zoom blog post](https://www.zoom.com/en/blog/chatbot-statistics/) repeats the figure under Salesloft's name as "39% of all conversations occurred when offices were closed" and adds that "around-the-clock availability matters just as much in B2B and B2C settings". Zoom drops the East Coast scope and the year, and the B2C claim is Zoom's inference, not Salesloft's.

The report also has the one chat-speed finding among this article's sources: live agents who replied within two minutes of a visitor engaging with a chatbot had the highest chance of booking a meeting, with no sample size or method given. It concerns human agents, not AI replies.

## Why is speed necessary but not sufficient?

Take an invented example, not a real conversation. On a Saturday evening a visitor asks whether there is a beginner class this weekend. The [solutions pages](/solutions) cover typical enquiries.

- **Fast and empty.** A greeting that only asks what the visitor needs arrives in seconds. InsideSales.com's 2014 report discounts automatic email the same way: a first email attempt is "less telling of response immediacy". The visitor has learned nothing.
- **Fast and useful.** The Mind gives the Saturday class from your published timetable, then asks one question, such as whether the class is for the visitor or someone else.
- **Needs a person.** The visitor asks whether the class suits a recent injury. The Mind is not the human expert and AI replies can be wrong, so keep a person involved in medical, legal, financial and other consequential decisions. The Playbook should have the Mind decline to judge, request the named person and say when your team next replies. The Mind keeps replying until someone takes over, so test that stretch too.

HBR lists why companies were slow to pick leads up: daily rather than continuous CRM pulls, sales staff busy with their own prospecting, and lead-sharing rules based on geography and "fairness". Faster software does not remove those causes; it moves them to the handoff. A Mind that answers in seconds and a person who takes two days still leave the visitor waiting.

## What can a Digital Mind do in the first few minutes?

A Digital Mind is an AI that answers visitors using your approved knowledge and instructions, and hands over to a person when needed. It is not the human expert and should not be presented as one. Depending on your set-up, plan or rollout, it can cover six jobs in the first minutes of an enquiry.

| Job | What a Digital Mind can do | Where you set it up |
|---|---|---|
| Answer | Reply from the knowledge and instructions you approved, in the channels you connect | Knowledge (Drive, Social media) and Mind › Playbook |
| Ask | Put one qualifying question and save the details a visitor clearly states on their contact | Mind › Playbook Flow and Mind › Data Collections |
| Offer | Offer your booking link when your conditions are met. A click shows interest and does not confirm a booking | Growth › Booking |
| Get help | Consult a Specialist for one focused job, such as reviewing a document | Mind › Specialists |
| Place | Move a contact into an automatic stage when a reply gives enough evidence. Your team confirms who became a customer | Audience › Stages |
| Hand over | Request a person. Your team is notified and the Mind keeps replying until someone takes over. Inbox holds the contact's stage, tags, collected answers and private notes | Mind › Playbook and Audience › Inbox |

Where it answers depends on the channels you connect, and on AI credits: when they run out, AI replies pause and skipped replies are not replayed. A website chat widget puts conversations in the same Inbox as your other channels. On WhatsApp, Auto-reply starts off, and an ordinary reply is possible only within 24 hours of the customer's last message. The Mind's instant reply falls inside that window. A handoff that stalls past it needs a WhatsApp-approved template, and approval does not replace the recipient's permission, so message only people who agreed to hear from you.

None of this is a claim about your conversion rate.

## How do you measure your own response time?

You need three timestamps per enquiry and a written definition of "useful".

1. Record when each enquiry arrived, when the first reply of any kind went out, and when the first useful reply went out. The first-attempt time is the closer match to the studies above. Set your target on the useful-reply time.
2. Pull the last 30 to 50 enquiries from each channel (form, chat, WhatsApp, email) by hand if you have no reporting. Count enquiries that never got a reply as their own row: HBR's audit found 23% of companies never replied to its test lead. Report scans and chat opens separately; Insights counts Quick Chat opens apart from engaged entries, which need a visitor reply.
3. Report the median and the slowest tenth, not only the average. The 2014 report's median was 3 hours 8 minutes against an average of 61 hours 1 minute.
4. Split the results by hour and day, because the 2014 benchmark says nothing about evenings or weekends. Your split shows how much of your demand waits for morning.

## How do you set up for fast, useful first replies?

1. **Choose where enquiries start.** Connect the channels your visitors already use: the website chat widget, [WhatsApp](/blog/whatsapp-ai-agent-guide), and Quick Chat links or [QR codes](/blog/qr-codes-and-quick-chat-for-in-store-conversations) on packaging, displays and emails.
2. **Write the Playbook and keep facts in Knowledge.** Set the Goal, Role & style, Boundaries and Flow, and put prices, policies and answers in Knowledge. The Playbook then only says how to behave: answer first, then ask one question at a time.
3. **Decide what counts as qualified.** Create Data Collections fields only for details you will use, and write clear "Move here when..." criteria for each stage. [From conversation to qualified audience](/blog/from-conversation-to-qualified-audience) explains how placements are read.
4. **Name who takes over, and by when.** Tag the person or team who handles a handoff in the Flow, and add your Booking page in Growth. Test one normal request and one that should be handed off in the chat preview, check Inbox and select Take over. Decide who checks the Unreplied filter, which lists contacts still waiting on a reply, and how often.
5. **Cover failed replies and test out of hours.** Because AI replies pause when credits run out and skipped replies are not replayed, check Inbox for conversations where your Mind couldn't reply and assign someone to answer them. Message each channel at night and at the weekend, at different hours each month, read the first reply, and recheck your medians after a few weeks. Use Insights to find questions with no or partial answers and add them to Knowledge (answer analysis depends on your plan). [Measure AI conversations](/blog/measure-ai-conversations) covers what to count.

## Frequently asked questions

### What is speed to lead?

Speed to lead is the time between an enquiry arriving and your first response. The research timed the first attempt. This article also recommends tracking the first useful reply: an answer, or one specific question that moves things on, not an automatic acknowledgement. Measure each per channel and across all hours, including evenings and weekends.

### Does replying within five minutes make a lead 21 times more likely to buy?

No. A summary of an InsideSales.com/MIT study presented in 2007 reports 21-fold higher odds of qualifying web-form leads when the first phone call came within 5 minutes rather than 30, across six companies and over 15,000 leads. It says it did not address close ratios, so it measures qualification, not sales, and it covered phone calls, not chat.

### How fast should you reply to an enquiry?

No source checked for this article sets a standard for chat or messaging. The lead-response studies measured phone follow-up in minutes and hours and found delay costly early. A practical aim is to measure your own median time to first useful reply by channel, including out-of-hours enquiries, then set a target your team or Digital Mind can meet every day.

### Does an automatic reply count as responding?

Not by itself. InsideSales.com's 2014 report says an automatic first email is less telling of response immediacy, because many companies send one within seconds. Treat the useful-reply clock as stopped when a person or a Digital Mind answers or asks one specific question that moves the enquiry on. A generic greeting does not count.

### Can a Digital Mind reply outside office hours?

It can answer, at any hour, on each channel you have connected and switched on, using your approved knowledge, while sufficient AI credits remain. When credits run out AI replies pause, and skipped replies are not replayed. On WhatsApp, Auto-reply starts off and ordinary replies are limited to 24 hours after the customer's last message. It is not the human expert, and consequential decisions should go to a person.

### How do you measure response time?

Record when each enquiry arrived, when the first reply of any kind went out and when the first useful reply went out, for 30 to 50 recent enquiries per channel. Count unanswered enquiries separately, report the median and the slowest tenth rather than only the average, and split results by hour and day to see how long out-of-hours enquiries wait.

## Sources

- [The Short Life of Online Sales Leads](https://hbr.org/2011/03/the-short-life-of-online-sales-leads), Harvard Business Review (Oldroyd, McElheran and Elkington), March 2011, accessed 2026-10-22
- [The Short Life of Online Sales Leads (print view captured 13 November 2011)](https://web.archive.org/web/20111113015409/http://hbr.org/2011/03/the-short-life-of-online-sales-leads/ar/pr), Harvard Business Review, archived by the Internet Archive, accessed 2026-10-22
- [The InsideSales.com/MIT Lead Response Management Study (summary of the October 2007 presentation)](https://www.mortech.com/hs-fs/hub/25649/file-13535879-pdf/docs/mit_study.pdf), InsideSales.com and Dr James Oldroyd, archived copy, accessed 2026-10-22
- [Annual 2014 Lead Response Report](https://resources.insidesales.com/wp-content/uploads/2019/11/2014-Lead-Response-Report.pdf), XANT (now InsideSales.com), accessed 2026-10-22
- [Lead Response Study 2021](https://www.insidesales.com/response-time-matters/), InsideSales.com, accessed 2026-10-22
- [Conversation Trends Report](https://assets.contentstack.io/v3/assets/blt48e9deb40787970b/blt10355646adde4b3e/66abcfd1eb20b48bb62d2355/Conversation_Trends_Report.pdf), Salesloft, accessed 2026-10-22
- [Chatbot statistics](https://www.zoom.com/en/blog/chatbot-statistics/), Zoom, accessed 2026-10-22