First contact resolution only counts if the customer agrees

First contact resolution is one of the few support numbers I actually like, and one of the easiest to fake. It asks the right question: did the customer get their problem solved the first time they asked? The trouble is who gets to answer it.

In most queues the answer comes from the agent. They close the case, tick “resolved”, and the report counts it as a first contact resolution. Whether the customer felt the same way turns up somewhere else, a week later, as a new ticket in a different channel that nobody connects to the first one.

The agent’s tick box isn’t the answer

Say a customer emails because their invoice is wrong. The agent explains the charge, the customer reads it, doesn’t quite follow, and calls two days later. On the call they get transferred once, and the second person issues a credit. On most dashboards that’s two cases, and the first one went down as resolved on first contact. For the customer it was one problem, three conversations and four days.

That’s the gap I care about. The figure is counted per contact, but customers live per problem. Count it per contact and every handover, and every vague answer that gets a case closed, quietly improves your score.

It gets worse once it becomes a target. Put a first contact resolution goal on a team and you’ll get closed cases, because closed cases are what’s being counted. People stop asking whether it worked, since a “no” means the case stays open. Some will close and hope. I don’t blame them for that, they’re doing what the number asks. A target only ever gets you to the target, and this one can be hit without anyone being helped. It’s the same problem as metrics that measure the team instead of the customer.

Count first contact resolution from the outside

What I’d measure instead is simple to explain and a bit more work to set up. A case only counts as first contact resolution if the same customer doesn’t come back about the same problem within a set window, say seven days, in any channel. Email, phone, chat, a message to their account manager, all of it. If they come back, the first contact didn’t resolve anything, whatever the case status says.

That needs two things a queue doesn’t usually give you by default. You have to recognise the customer across channels, which normally means matching on email address or customer number rather than ticket ID. And you need a rough idea of what “the same problem” is, which can be as basic as the contact reason the agent picks. It won’t be perfect. It doesn’t need to be. A rough repeat-contact rate tells you more than a perfectly clean tick box.

Then ask the customer. One question after the case closes, “Is this solved?”, yes or no, with no score attached. Most of the work is in the no answers, because every one of them is a case your report called resolved.

Count it this way and expect the number to drop, possibly a lot. That’s fine, it was never as high as it looked. What you get back is a list of the exact problems that bring people back, and that list is worth far more than the percentage. It’s the same idea as using support data to stop issues arriving at all. The repeat contacts point at where the cause sits: the product, a letter, a process, or what the first agent was allowed to do.

That last one gets less attention than it deserves. A good share of second contacts happen because the first person couldn’t issue the credit, change the booking or send the part. They knew the answer and had to pass it on. Give the frontline the judgement and the access to finish the job, with light guardrails, and first contact resolution goes up without anyone being told to push it.

Some problems shouldn’t be solved on the first contact

Not every case can end on the first contact, and some shouldn’t. A machine that needs an engineer on site, or a refund that has to go through finance. Forcing those into one contact gets you rushed answers and guesses, and a wrong answer today costs more than a right one tomorrow.

For those I’d rather see a clear promise: what happens next, and when the customer will hear back. Then keep it. The customer still has one problem with one owner, and as long as they never have to chase, I’d call that handled properly even if it took two steps. Leave those cases out of your first contact resolution figure and judge them on total case time instead, from the moment the problem happened to the moment it was fully solved.

So this week, pull last month’s closed cases and check one thing: how many of those customers came back within seven days about the same problem. That’s your real first contact resolution, and the cases behind it are your to-do list.

Which of your resolved cases came back, and why? If you want a second pair of eyes on that, it’s what in-house optimisation is for, or you can tell us how your queue runs today.