First Response Time and the Window That Decides the Job

Why this matters

This is the one number in the quote-to-cash chain where the target is genuinely known, genuinely short, and genuinely within a small shop's control. The window that decides an inbound web or text lead is measured in minutes, not hours, and the drop-off past it is steep rather than gradual. A shop reading a monthly average of a few hours has no idea whether it is winning that window or losing it, because the average is the one statistic that cannot tell it apart.

What it counts, and the leads it cannot count

Mean hours from a lead arriving to the first logged activity against it, over the leads received in the window.

The catch sits in that sentence. A lead nobody ever responded to has no first activity, so it contributes no hours, so it cannot be in the mean. The worst cases in the month are the ones the figure physically cannot see, which is the survivor-bias shape a sibling card owns.

That makes the never-answered count a mandatory companion, not an optional extra. Report the share inside your target, the mean, and the never-answered count on one line, over the same denominator of leads received. Any one of those on its own is misleading.

The target, and it is minutes

Commit to it rather than hedging: first human contact attempt within 5 minutes during staffed hours for an inbound web form or text.

The evidence behind that number is the lead-response timing research summarised in Harvard Business Review's work on the short life of online sales leads, which found the odds of reaching and qualifying an inbound web lead falling by roughly an order of magnitude between a five-minute response and a thirty-minute one, and continuing to fall through the first hour. The exact multiple varies by study and by market. The shape does not, and the shape is what you set a standard against.

A sibling card in this library sets an office standard of 15 to 30 minutes for a web request and 5 to 10 minutes for a missed call, which is what one person covering the phones alongside other work can realistically hold. Both numbers are useful and they are not in conflict as long as you know which is which: 5 minutes is the target that wins the job; the sibling's band is the line that, once you are breaching it regularly, is telling you to change the staffing rather than try harder.

What the customer is doing in those minutes

The mechanism is worth understanding, because it explains why the curve is steep and why the fix is structural rather than motivational.

Somebody with a failed water heater at 10:02 is not evaluating shops. They are working down a search page or filling the same form on three sites, and their goal is to stop doing that. The first competent human who calls back ends the search, because the search itself is the thing they want over with.

So the advantage is first-mover, not best-offer. A second-best price delivered at minute 4 beats the best price delivered at hour 3, routinely, and a shop that believes it lost on price has usually lost on order of arrival and never found out.

The window is also physical. A form submitted at 10:02 is a person sitting at a screen with the problem in front of them at 10:03, and somewhere else entirely at 14:30. You are not competing for their decision. You are competing for the few minutes in which they are still holding the problem.

That is why outcomes behave like a step rather than a slope. Interest does not decay smoothly; at some point somebody else answered. And an average taken across a step function lands in a region with no operational meaning at all.

A fast average with a long tail

A month of 180 inbound web and text leads:

One hundred and eighty leads, of which 168 got some response.

Taking each band at its midpoint, the mean across those 168 is about 3.5 hours. The median sits between the 84th and 85th of the 168, both inside the 5-to-15 minute band, so the median response is around 9 minutes.

Both figures describe the same month. The mean says the shop is slow enough to be losing most of its inbound work. The median says it is fast enough to be beating almost anyone. Neither is a lie and neither is usable.

The number that runs the shop is the share inside the target. 105 of 180 leads received, 58.3 percent, got a human inside 15 minutes. The other 75, 41.7 percent, did not, and 12 of those, 6.7 percent of the month's leads, never got one at all.

The mean is being driven by 23 leads that waited more than four hours, which is 13.7 percent of the 168 that got a response. Twenty-three records out of 180 are setting a figure the owner reads as a verdict on the whole month.

Where the tail actually lives

Cross the response time against the hour the lead arrived, and the diagnosis stops being about effort.

Of the 180 leads, 52 arrived outside the staffed band, which for this shop ends at 17:00 on weekdays. That is 28.9 percent of the month's leads, arriving when nobody is scheduled to be there.

Take the 35 worst cases, meaning the 23 that waited over four hours plus the 12 that were never answered. 28 of those 35, 80 percent of them, arrived outside the staffed band. Put the other way: of the 52 leads arriving outside the band, 28 became a worst case and 24 did not, which is 53.8 percent falling into the tail depending on whether somebody happened to check that evening.

That last figure is the finding. The after-hours result is not consistently bad, it is a coin flip, which is the signature of a habit rather than an arrangement.

Nothing here says the daytime team is slow. A 9-minute median is genuinely strong. The problem is a coverage gap with no owner, and no amount of pressure on the daytime team touches it.

The arrangements that actually move it

Three, in the order they are worth doing.

Name the owner of the first touch for each staffed band, by name, on a board. Not "whoever is free". A lead that belongs to everybody is answered in the gap between two people each assuming the other saw it, and that gap is where four-hour responses come from.

Decide the after-hours standard and write it down as a different standard. Two honest options. Run a rota where a named person covers 17:00 to 20:00 on a stated response promise, which for an evening band is realistically 60 minutes rather than 5 and should be written as 60 rather than pretended to be 5. Or concede the evening, send an auto-reply that states when a human will respond, and clear the overnight queue in a 07:45 sweep before the day's calls start. Both are legitimate. What is not legitimate is the current arrangement, where the evening is covered on some days and not others and nobody has decided which.

Size the concession before you make it. For this shop, conceding the evening means writing off the best part of the 52 after-hours leads a month, 28.9 percent of everything received, to whichever competitor is running the rota. Whether that is acceptable is a real business decision. Making it by accident is not.

What changes the answer: not every lead is in a race

The 5-minute target is about competitive inbound, and two conditions change it.

The lead was not shopped. A referral from a past customer, or a repeat customer calling back, is not sitting in a form on three other websites. The curve for that lead is much flatter, and a 90-minute response loses far less than it would on a search-driven enquiry. It still loses something, because a customer who has to wait learns what waiting on you is like.

The channel is different. A ringing phone is not measured in minutes at all: the standard there is answering live, and a callback is already the fallback. An after-hours emergency call is a different product with a different promise. Do not blend these into one figure, because the blended number has no target it can be read against.

Both of those point at the same practical move: compute the share inside target per lead source, not just for the month. A paid directory lead is typically sold to several shops at once and is the purest form of the race. A referral is not. If your worst response times are concentrated in the source you pay the most for, you are buying leads and handing them to whoever answers faster, which is the most expensive way to run this number badly.

What counts as the clock stopping

The measurement is only worth what its definitions are worth, and this is where a good-looking figure is usually manufactured. Five boundaries, and the exclusions matter more than the inclusions.

The clock starts when the lead arrives, not when somebody opens the record. A form submitted at 19:40 and opened at 09:00 is a 13-hour 20-minute response. If your clock starts at the open, the whole after-hours tail above disappears from the figure while continuing to happen.

An automated acknowledgement does not stop the clock. Send one, because it stops the customer wondering whether the form worked. It does not answer their question and it does not stop them calling the next shop, so a shop whose records stop the clock on the auto-reply reports a median of a few seconds and learns nothing.

An internal note does not stop the clock. Reading a lead and tagging it is not a response. The customer experienced nothing.

A voicemail or an unanswered call stops the attempt clock only. Carry two figures: time to first attempt and time to first two-way contact. The gap between them is a reachability problem, and reachability is fixed by when you call, not by how you sell.

A reply that does not answer the question is an attempt, not a contact. "We got your request and someone will be in touch" is the auto-reply with a person typing it. The clock that matters stops when somebody who can answer the customer's question is speaking to them.

Publish the definition alongside the number. Two shops quoting response times are usually measuring two different things, and so are the same shop's figures either side of a change in who logs what.

References

• See related: The Callback Response Standard Worth Setting - the office standard by channel, and how the 15 to 30 minute band relates to the 5-minute target here
• See related: Lead Conversion Rate Excludes the Leads You Never Touched - where the never-answered leads go in the funnel arithmetic
• See related: Average Sales Cycle Describes Winners Only - why this window shows up in the win count rather than in the length of a cycle
• Harvard Business Review, research on the short life of online sales leads, for the response-time-to-qualification relationship cited above