Speed to lead SLA: what response time to promise, and how to measure it

A speed to lead SLA is a written promise about how fast a new inbound lead gets a first useful reply. Define when the clock starts and stops, set a target by lead source and hours, measure the median and the slowest 10 percent, and name who acts when a target is missed. The 2011 Harvard Business Review study of 2,241 companies found only 37% replied to a web lead within an hour, so a target you actually keep already puts you ahead.

Most teams have a feeling about speed ("we reply fast") and no number. An SLA turns the feeling into a number that someone owns. This post covers what to promise, how to define the clock, how to measure it without fooling yourself, and a worked example. The study figures below come from the Harvard Business Review article, read on 25 October 2026. Everything else is method, and the example numbers are made up and labelled.

What does the research say about response time?

The best known evidence is the 2011 Harvard Business Review piece by James Oldroyd, Kristina McElheran and David Elkington. Its authors audited 2,241 U.S. companies by sending each a web generated test lead and timing the answer:

The average response time among companies that did reply within 30 days was 42 hours. A separate study in the same article looked at 1.25 million leads received by 29 B2C and 13 B2B companies in the U.S. Firms that tried to contact a lead within an hour were nearly seven times as likely to qualify it, defined as a meaningful conversation with a key decision maker, as firms that tried an hour later. They were more than 60 times as likely as firms that waited 24 hours or longer.

Two cautions. The data is from before 2011, so treat the exact multiples as old evidence, not a law of nature. And you will see "five minutes" and "100 times" attributed to this article on many blogs. They are not in it. The article's own windows are one hour, one to 24 hours and beyond. We would not put a number in an SLA because a blog repeated it.

What does the SLA actually promise?

Write one sentence: "For every new inbound lead from [source], a [person or agent] sends a first useful reply within [time] during [hours]." Each bracket needs a decision.

Who counts as a lead. Someone who asked for something: a form, a chat, a direct message, a reply to an ad. A newsletter signup is not a lead for this purpose unless you decide it is.

What counts as a first response. A reply that addresses what the person asked or moves them to a next step, such as a booking link with a reason to use it. An automatic receipt is not one. If receipts count, you will hit your target by definition and learn nothing.

Which hours. A lead that arrives at 2 a.m. on a Sunday needs a rule. You either promise the same time around the clock, which usually means an agent or a rota, or you promise "within 15 minutes of opening" and measure from opening time. Both are honest. A promise of five minutes that quietly means "in office hours" is not.

How do you start and stop the clock?

Start the clock when the person acts, not when your system notices. That is the part teams skip, and it hides most of the delay.

• Start: the timestamp the lead submitted the form or sent the message. If you can only see the time the lead reached your CRM, use that and write down that the gap before it is not measured.
• Stop: the timestamp of the first reply that meets your definition, sent by a person or an agent.
• Pauses: none. Do not stop the clock because the lead is "not qualified yet". The point is to reply.

Polling is a hidden part of the clock. Sellinger polls Facebook lead form leads from a sheet every 60 seconds and Google Ads leads every 300 seconds, then creates them as leads and contacts them by the campaign, as described in our post on ad lead speed. A Google Ads lead can therefore wait up to five minutes before any workflow starts. If your SLA is five minutes, the poll alone can use all of it. Know every delay in the chain before you pick a number.

What target should you set?

Set tiers, not one number. The right tier depends on how fast the person expects an answer and how much the lead is worth.

The table is a made up example for a fictional B2B software company. The numbers are illustrations of structure, not benchmarks.

Sellinger's own goal library includes a goal to respond to every inbound inquiry within five minutes, classify intent and route it to the right sequence or team member. That is a default we ship, not a result we claim improves outcomes. We publish no figure showing that five minutes beats ten.

Choose each target by asking one question: could we keep this at 3 a.m. and on a Sunday? If not, either lower the promise or change who answers.

How do you measure it without fooling yourself?

Four rules.

1. Use the median and the tail. Report the median first reply time and the slowest 10 percent. An average can be pulled up by a single lead that waited three days, and a median hides that lead. The tail is where leads go cold.
2. Report attainment. The share of leads answered inside the target. "92% inside five minutes" is something a manager can act on.
3. Split by source and by hour. A good overall number can hide a bad Saturday or a bad channel.
4. Count replies, not touches. Do not let an auto receipt, an out of office or an internal note stop the clock.

A worked example, with made up numbers

A fictional team gets 200 inbound leads a month. For their demo request form the SLA says 15 minutes in hours. They pull timestamps for one month.

Reading it: the median looks fine, but 29% miss the target, and the tail is long because the 38% that arrive out of hours wait for morning. Two fixes follow. Put an agent on the form so out of hours leads get a real reply with a booking option, and give the on call rep an alert at the 10 minute mark so in hours misses are caught before the target passes.

What happens when the SLA is missed?

A target with no consequence is a wish. Define three things in the same document.

• An alert at a fraction of the target, for example at two thirds, to the person who can act.
• An escalation to a named backup if the alert is not handled.
• A review: a weekly look at every miss, with a one line cause. Patterns beat blame.

Speed is only the front of the problem. In the Hotel Pirin case study, the median first reply to a guest was 41 seconds over 54 inquiries, yet once a guest was handed to staff the wait ran from 3 hours to 14 days. Our post on the handoff bottleneck covers that second clock. Consider adding a second SLA: time from handoff to the human's first reply.

On our own LinkedIn outbound, reply time fell from about 19 minutes to about 2 over a year and meetings per conversation did not rise, according to the Sellinger case study. The list and the message were the bigger levers. Speed is the price of staying in the conversation, not a guarantee of winning it.

Where does an agent fit?

An agent can hold the first reply target around the clock, from a knowledge base with your prices, hours and rules, while a person owns the handoff. You decide the mode: a campaign can run in approval mode, where every message waits for a person to approve it, or send automatically. A person can take over any conversation, and the agent stops on its own when it cannot answer, for example a price it has no answer for. Our speed to lead guide for messaging channels covers the setup, and the glossary entry gives the short definition.