How do you measure a WhatsApp response SLA?

Setting a reply target is the easy half. How fast should you reply covers choosing one from your own numbers. This is the other half: once you have said “4 working hours”, how do you know whether your team is keeping it? Most of the ways managers measure it get a flattering answer, and each one flatters for a different reason.

Everything here assumes your reps sell from the WhatsApp Business app on company phones. The numbers come from linking those company numbers, not from an API or an inbox the reps work in.

Attainment, not an average

An average reply time answers the wrong question. A customer does not experience your average; they experience whether they heard back before they gave up. The SLA question is a yes or no per wait: did this customer get a reply inside the hours you promised? SLA attainment is the share of waits where the answer was yes.

Keep the median and the 90th percentile beside it. The median tells you what a normal wait looks like. The 90th percentile tells you what the slowest tenth of your customers went through, which is where the lost deals are. Attainment can fall while the median holds perfectly still, because the slow replies are getting slower and the fast ones are not changing.

Count the waits that never ended

The most expensive failure has no reply time at all. A customer who wrote on Tuesday and never heard back does not appear in a reply-time average, because there is no reply to time. A team that ignores its hardest conversations can post a better average than one that answers all of them late.

So count open waits past the SLA as a separate figure, by name, oldest first. That list is the one to act on today; attainment is the one to discuss on Monday.

Working hours, stated once

A message that arrives at 11 pm on a Friday and gets an answer at 9:10 on Monday was answered in 10 minutes, if your team does not work weekends. Counting it as 58 hours makes the SLA meaningless, and reps stop looking at it. Count the SLA on the working hours you actually keep, say so wherever you report it, and keep the wall-clock wait as well. The customer did wait the weekend, and some of them will say so.

One customer is one wait

Sales teams on WhatsApp share customers. A group chat can hold 3 of your company numbers, and a customer can write to one rep and hear back from another. Measured per number, that group is 3 waits, 2 of them never answered. Measured per customer, it is 1 wait, answered.

The fair split is to ask 2 questions. For the customer, the wait ends when anyone on your side answers, on any number. For the rep, it ends at their own reply, so the per-rep figure still shows who let a colleague pick it up. A connected WhatsApp call to the customer is an answer too. A team that moves hard conversations to the phone looks slow on every messaging-only measure; call tracking is what fixes that.

Promises need their own clock

The first reply is only the first thing a customer waits for. After it come the quote, the documents, the options the rep said they would send by tomorrow. None of those has a message to be late on, so a reply SLA never sees them.

Give each of those its own SLA: what starts the clock, what should be said and by whom, and how long it may take. “Customer asks about price, rep sends a figure within 2 working hours” is an SLA exactly like the first reply, with an on-time rate of its own. Track the 2 or 3 that cost you deals, not 20.

Per rep, last

Read attainment for the team first, then per rep, and put workload beside it. A rep with the lowest share inside the SLA may simply be carrying the most conversations, and team performance shows the two together for that reason. Tell your team what is measured and why; what your employer can and cannot see is written for them.

Where this lives in WAOversight

SLA tracking does each of these: a reply SLA in working hours, SLA attainment per rep and for the team, every customer waiting past it in the queue, and deadline rules with their own on-time rate and median and p90 time to meet. The definitions are in the metrics guide. It never replies or chases anyone; the reply still comes from the rep, on their phone.