How long you take to reply: measure it channel by channel

Ask any business owner what they turned over last month and they'll tell you almost to the euro. Ask how long it takes them to answer someone who messages on Instagram at half seven on a Tuesday evening and you get "quickly". Quickly isn't a number. And because it isn't a number, you can't improve it, compare it across channels, or work out whether your problem is getting enquiries or simply answering them.

This is for owners of small businesses fielding messages across four or five different places with no idea which one is quietly leaking customers. By the end you'll have a boring, cheap method — one week, one spreadsheet, no new software — for finding out how long you actually take on each channel, plus a way to tell which part of the gap automation fixes and which part it doesn't.

The number you don't have

First response is where most of it gets decided. Someone asking your price is almost never asking only you: the same message went to two or three places, and they go with whoever shows a sign of life first.

There's a well-known data point here, and it's worth reading carefully. In 2011 Harvard Business Review published The Short Life of Online Sales Leads. The authors audited 2,241 US companies by submitting a test lead through each firm's own website. 37% replied within an hour, 23% never replied at all, and among those that did reply within 30 days the average was 42 hours.

Be careful what you take from that. It's over a decade old, it's US firms, and it's web forms rather than WhatsApp. It isn't a target ("we must reply in five minutes") and it isn't a benchmark for your sector. What it's good for is one thing: the gap between how fast companies think they answer and how fast they actually answer is enormous, and it almost always runs the same way. Your business isn't the exception until you've measured it.

Decide what counts as a reply before you measure anything

This is where homemade measurement usually falls apart. Set the rules first, or everyone logs whatever they feel like and the number means nothing.

• The clock starts when the message lands, not when you see it. Your phone being face-down in a drawer is precisely the thing you're trying to measure, not a reason to discount the entry.
• The clock stops at the first useful reply. "Hi, we've got your message, someone will be in touch shortly" doesn't stop it — that's an acknowledgement. Log it separately if you like, because it does have value, but don't mistake it for answering.
• Measure in wall-clock time, not working hours. A message that arrives Friday at 19:05 and gets answered Monday at 09:30 is 62 hours, not 30 minutes. Customers live in wall-clock time. Record both if you want — the gap between the two figures tells you exactly what your opening hours are costing you.
• A channel is a channel. WhatsApp, phone (including missed calls you never return), email, the website form, Instagram DMs, marketplace portals, Google Business messages. Each one separately. A single blended average across all of them is a useless number.

One week, one spreadsheet

Don't build anything. Five columns:

1. Channel.
2. Date and time the message arrived.
3. Date and time of the first useful reply (or "never answered").
4. Who answered.
5. Type: new customer, existing customer, supplier, spam.

Log for seven calendar days, weekends included. Thirty or forty entries is enough to see the pattern; you don't need three hundred. Split the logging across whoever picks up each channel, and log it as it happens rather than from memory at the end of the day, because memory always rounds in your favour.

One honest warning: the week you measure, you answer better. That happens to everyone. Assume your real number is somewhat worse than the one you get, and change nothing during those seven days — no new rota, no new alerts, no new hours — or you'll be measuring the experiment instead of the business.

At the end of the week, pull four figures per channel:

• The median (the middle value when you sort them), not the mean. One enquiry that sat dead for three days wrecks the mean and convinces you that everything is broken.
• The worst 10%, which is where the angry customers and the one-star reviews live.
• The share with no reply inside 24 hours.
• The share that arrives outside your opening hours. This is usually the one that shocks people.

What almost always turns up

Do this across enough different businesses and the same findings come back with a regularity that's almost funny.

The orphan channel. There's always one — usually the website form or Google Business messages — that belongs to nobody. It lands in a shared inbox everyone can open and nobody reads. Medians there aren't measured in minutes, they're measured in days. And it's frequently the channel used by people who have already decided to buy.

The dead slot. You don't answer at the same speed at 10am as you do at 3pm. In hospitality and retail the bad slot is service; for trades, it's the hours in the van. That gap is closed by three well-placed notifications, not by hiring someone.

The lying average. Your mean is 4 hours and your median is 11 minutes. Nearly everything gets answered fast and four things rotted. That's a hole problem, not a speed problem, and the fix is a rule — "nothing sits unanswered longer than X" — rather than asking the team to hurry up.

Fast but empty. You reply in two minutes with "I'll confirm this afternoon" and the real answer arrives the next day. Separating acknowledgements from useful replies makes this visible immediately. It matters, because from the inside it feels like you're flying.

If your worst channel has a two-day median, your problem isn't that you need more leads.

Not everything here is an automation problem

Time to be straight about this. There are three possible fixes and only one of them involves technology.

The first is to close the channel. If Instagram DMs are open and nobody staffs them, staffing them badly is worse than not having them. Turn them off, or bounce them to WhatsApp with a fixed message. Fewer doors, properly attended, beats six doors half-attended.

The second is to give it an owner. A huge share of orphan-channel cases is solved by one sentence: "Anna checks the website form at 9:00, 13:00 and 17:00." Zero euros, zero tools.

The third — once the volume no longer fits inside one person's day, or once 40% of what arrives comes in outside your hours — is to have an agent answer: give the genuinely useful reply (ballpark price, availability, what documents are needed), clear the repetitive stuff, and hand you only what deserves you, with the groundwork already done.

And one more: if you get eight messages a week, don't measure anything. This exercise starts making sense somewhere around twenty or thirty enquiries a week. Below that, the noise swamps the signal and your time is better spent almost anywhere else.

What to do with the number

Pick one channel — the worst one — and give it a target you can actually check: "website enquiries get a real answer within two working hours." Make one change. Just one. Measure again for a week a month later, with the same rules. If it hasn't moved, don't blame the team: you picked the wrong change.

Once that number is sitting in front of you, the argument about whether you need help stops being a matter of opinion. That's how we set up the receptionist agent: it picks up what comes in via WhatsApp, web or phone at any hour, gives the first useful reply, and leaves the things that genuinely need a person ready for you. You approve, it executes. But measure first — if only so you know what you're playing against.