Contact Centre Metrics That Drive Action

The dashboard nobody acts on

Ask a COO what their contact centre's core metrics are, and the answer comes fast and confident. Ask what changed last quarter because any of them moved, and the answer slows right down. That gap is the actual problem, and it isn't cheap.

Walk into most weekly operations reviews and you'll find the same scene: a screen full of numbers, a manager talking through each one in turn, and almost nothing that changes once the meeting ends. Average handle time is up a few seconds. Adherence dipped on Tuesday. Someone nods, a note gets taken, and the review moves to the next line.

Oliver has sat on both sides of that screen for 25 years, running operations directly and sitting in reviews with outsourced partners. The pattern is recognisable within the first ten minutes, and the issue is rarely a shortage of data. Most centres, whether run in-house or delivered through a partner, already collect far more than anyone has time to use properly.

The cost of that is concrete: the licence fee for reporting nobody consults before a decision, the hour a week a supervisor spends preparing a readout instead of coaching, and the risk sitting quietly underneath a dashboard that tracks speed and cost while attrition and repeat contact drift the wrong way unwatched, until the numbers that do get read start moving too.

A number on a dashboard doesn't change behaviour on its own. It changes behaviour when someone owns it, reviews it on a fixed schedule, and knows what decision follows if it moves the wrong way. Remove any one of those three things and the metric becomes wallpaper, reported without being read and forgotten by the next meeting. That's the honest starting diagnosis for most dashboards worth fixing: not "we need better metrics," but "we're paying to report things nobody is using to decide anything."

Activity metrics and outcome metrics are not the same thing

Most dashboards are dominated by what's easiest to measure in real time: average handle time, schedule adherence, occupancy, contacts handled per shift. These are activity metrics. They describe effort and pace, update continuously, and attach cleanly to one agent, which is exactly why they end up on every screen in the building.

Outcome metrics are harder to build and slower to move: first contact resolution, repeat contact within a set window, escalation rate, complaint themes. They describe whether the customer's actual problem got solved, which is a different question from how quickly someone got off the call.

The trouble starts when activity metrics get treated as a stand-in for outcomes, because they can be optimised in isolation from them. An agent under pressure to hit a handle-time target can hit it three ways: rush the close, transfer a query they could have worked through, or log a contact as resolved when it isn't. Handle time improves. Nothing about the customer's problem has.

The lag that hides the damage

Repeat contact and escalations tend to rise a few days later, but they land in a different report, reviewed by a different person, on a different day. Cost per contact looks like it's improving while cost per resolved issue is quietly getting worse, and the gap between those two numbers doesn't show up until someone totals repeat contacts a quarter later. By then the team has already been praised for the number that moved.

None of this makes activity metrics useless. Handle time and adherence matter for staffing and cost. The fix is pairing them: report one alongside the other in the same conversation, every time, so speed can't be bought at quality's expense without someone seeing it happen in real time.

Building a metric set you can govern

The metric count on a dashboard tends to grow every time something goes wrong: a complaint spikes and someone adds a report; a target gets missed and someone adds a warning indicator. Almost nobody ever removes anything, because retiring a metric can feel like admitting it never mattered. A few years of that produces a screen nobody can hold in their head, which is exactly why most of it gets skimmed rather than acted on.

The ownership test

One discipline cuts through this reliably: for every metric currently reported, ask who owns it and what decision would follow if it moved. A clear owner and a clear next action earns a metric its place. "Nobody specific" or "nothing would really change" means the metric is decoration, nothing more, and it should be retired or folded into a report where it can genuinely inform a decision.

A structure that scales

Beyond the ownership test, most metric sets that hold up under pressure end up covering three kinds of signal:

A cost or efficiency signal: what this costs to run, day to day.
A quality or outcome signal: whether the customer's actual problem got solved.
A people signal: attrition risk, engagement, or coaching activity, because people problems show up in the other two numbers weeks after they start.

This isn't a template to apply without judgment. The right mix depends on the operating model, the channel and what's driving performance in a given team, and a dashboard missing one of the three tends to have a blind spot exactly where that category would sit. It's usually the people signal that's missing, because it's the hardest of the three to measure in real time.

Most teams find that the metrics worth a supervisor's weekly attention number in the single digits, well short of the twenty or thirty a dashboard can accumulate over a few years. Getting to that shortlist is harder than it sounds from the inside, because whoever built each report can usually explain why their own number matters. It's the whole set that needs an outside look, which is often where a contact centre consulting engagement earns its keep: someone with no attachment to any single line, asking the same "who owns this, and so what" question all the way down the screen.

Governing metrics through an outsourced partner

When delivery sits with a partner, the dashboard problem doesn't disappear. It changes shape. A provider will report what's written into the contract, because that's what's measured and easy to produce: service level, handle time, adherence, sometimes a satisfaction score. Those are almost always activity metrics, not outcome metrics, and a partner can be fully compliant on every one of them while first contact resolution quietly erodes underneath.

Most of the time, that has nothing to do with the partner being bad. Outcome ownership was simply never written into the relationship in the first place. SLA compliance measures whether the contract is being met. It doesn't measure whether the customer's problem is getting solved, and treating the two as interchangeable is one of the more expensive assumptions an executive overseeing an outsourced relationship can make.

The fix is the same ownership test applied across a contract boundary rather than an org chart: decide, before signing or at the next renewal, which outcome metrics you need visibility into, who on your side reviews them, and what happens if they move the wrong way while every SLA stays green. Not every partner will already track the outcome metric you need in the exact format you want. That capability gets matched during selection, which is precisely why it belongs in the selection criteria rather than a request made after go-live. That scoping conversation is worth having before delivery starts.

A short set of questions tends to surface where the gap is:

Which of the numbers reported monthly are outcome metrics rather than activity metrics?
Who on our side reviews those specific numbers, and how often?
What's the escalation path if an outcome metric moves the wrong way while every SLA stays green?

Where metrics change behaviour

Metrics don't change behaviour in a monthly readout. They change it in a specific conversation, close to the event, between a supervisor and the person whose number it is.

"Your handle time is above target this month" produces one of two responses: defensiveness, or a promise to try harder that fades within a week, because it gives the agent nothing concrete to do differently. "Let's listen to this call from Tuesday together and figure out where the time went" produces something else, because it turns a number into a specific, examinable piece of work. The metric flagged that something was worth a look. It didn't do the coaching.

That distinction separates centres where metrics genuinely move performance from centres where they're simply published. The dashboard's job is to point at where attention should go. The supervisor's job is the conversation that follows: grounded in a real interaction, specific enough that the agent can see what "better" looks like next time, and close enough to the event to still be useful.

That conversation is a skill in its own right. A capable agent promoted into a supervisor role inherits the scorecard. What almost nobody teaches them is the method for turning a line on it into a coaching conversation. It's a large part of what our frontline leadership training is built around, because the gap between having the numbers and using one to change what happens next is where most coaching intentions quietly fail, and where the return on every metric above it gets realised or lost.

Failure modes worth naming

A handful of habits are worth retiring from most operations reviews, because they consume attention without changing anything:

Reporting a number with no owner. If moving it isn't part of anyone's job, it shouldn't be on the weekly screen.
Leaderboards that reward volume over resolution. Ranking agents on contacts handled, without a resolution or quality figure next to it, teaches people to chase the ranking rather than the outcome.
Reviewing everything monthly instead of a few things weekly. A monthly readout of thirty metrics is a report. A weekly conversation about three or four is management.
Redefining "resolved" until the number looks right. If first contact resolution is falling, the tempting fix is to loosen what counts as resolved rather than address what's driving repeat contact. The number recovers. The underlying problem doesn't.
Comparing an outsourced team and an in-house team on the same metric name with different definitions behind it. If "resolved" or "handled" means something different in each report, the comparison ends up measuring whose definition was looser. That corrodes trust in the whole dashboard once anyone notices.
Adding a metric every time something goes wrong, without checking whether the existing set should already have caught it. Growth without pruning is how dashboards reach thirty metrics in the first place.

None of this is a training exercise, and it's worth saying plainly: no workshop fixes a metric set that was never designed with ownership and consequence in mind. That's a leadership and design decision, made once and then defended every time someone proposes another chart. Training earns its place afterward, once the right few metrics are in place and what's left standing is a genuine skills gap in how supervisors coach against them. That's the moment for structured customer service training.

What to change this quarter

Redesigning a metric set isn't a project to hand to a BI team alone. Someone will resist losing "their" number, and the person who can overrule that objection needs to sponsor the audit directly, rather than simply receive a summary of it afterward.

The audit

Run this over one quarter rather than one meeting:

List every metric currently on the weekly and monthly screen, wherever it's produced.
Against each one, write down who owns it and what decision follows if it moves the wrong way.
Cut or fold in anything that fails that test, whatever resistance that produces.
Check that every remaining activity metric has an outcome metric sitting next to it.
Confirm someone reviews the shortlist weekly. A weekly cadence is what turns reporting into management.

The output is a screen where every remaining number has a name attached to it and a known consequence if it moves: the only version of a dashboard that was ever going to change behaviour in the first place.

If your team can't get through that list without disagreement about who owns half the numbers on it, that disagreement is the real finding, and it's worth resolving before the next review begins. That's a conversation worth having with an outside perspective in the room; starting one with a specific dashboard on the table tends to move faster than starting from a blank page.

Frequently asked questions

How many metrics should a contact centre track?

There's no fixed number that suits every team, but the principle holds everywhere: fewer than most dashboards carry, and each one with an owner and a known consequence attached. Teams that get real value from their numbers tend to hold a handful in weekly focus, far fewer than the two or three dozen a full report can accumulate over time.

Should we hold an outsourced partner to the same metrics as an in-house team?

Hold them to the same discipline, even if the exact metric set looks different. Definitions need to match before any comparison is made, so "resolved" and "handled" mean the same thing in both reports, and the outcome metrics that matter to your customer need to be written into the relationship rather than left to whatever the SLA happens to cover. No provider guarantees matching capability upfront: it gets confirmed during selection and reviewed afterward.

Is a metrics problem something training can fix?

Rarely on its own. Choosing the right metrics, assigning ownership and attaching real consequences is a leadership and process decision. It sits above any workshop agenda. Training earns its place once that structure exists and a specific coaching or skills gap is what's left standing in the way of better numbers.

About the author

Oliver Catt is the founder of The Cattalyst. He has spent 25 years inside contact centres, customer service, sales and training operations, and now helps organisations develop their leaders and build outsourced teams through a global partner network.

Start a conversation with Oliver