How CX metrics can hide a broken customer experience

Your metrics are improving. Your experience might be getting worse.

The metrics are moving in the right direction. The story you can tell upstairs is a good one. And yet inside the teams, everyone knows the system is kind of broken in ways the dashboard doesn't show. The question is why it happens. And more importantly, what to do about it.

In this episode, Jochem van der Veer makes the case that most CX organizations are trapped in a measurement system that rewards what's easy to model, not what actually reflects experience health. He draws on real examples to show what it looks like to measure the integrity of a system, not just its outputs.

Why consequences and integrity aren't the same thing

Transactional NPS captures how a customer felt at a specific moment. It doesn't capture the structural friction accumulating underneath. You can reduce calls by making support harder to reach. You can reduce churn by locking customers into longer contracts. You can improve NPS at one touchpoint while the end-to-end journey gets more fragmented. All of these are measurable. None of them make the experience stronger.

The result: organizations get very good at improving lagging indicators while invisible friction builds underneath. It doesn't surface until it does, suddenly, as a churn spike or a trust problem that seems to come from nowhere. It didn't. It accumulated through blind spots the dashboard never showed.

Fixing economics, not scores

A utilities company Jochem worked with asked a different question. Not "how do we raise NPS?" but "what does it cost us per customer when someone doesn't onboard digitally, and what's pushing them away from it?" By tracing friction across the whole system, from onboarding to invoicing, and prioritizing fixes by impact, they shifted the economics of their entire customer base. Gross margin improved. Not because prices went up or service was cut, but because the system got healthier.

AI amplifies the wrong definitions at scale

AI doesn't create definitions. It accelerates whatever definition you already have. If your organization optimizes for call deflection, AI deflects more calls, faster. If transactional NPS is the target, agents will get very efficient at improving a number that doesn't reflect reality. The risk isn't that AI replaces CX. It's that AI locks in measurability bias before anyone stops to question what's actually being measured.

Definition before optimization

Most organizations skip defining integrity metrics because they don't fit neatly into a board pack. How fragmented is the journey across teams? How often do customers repeat themselves? How long do they sit in uncertainty? These get labeled soft, and soft things don't get prioritized. The mature move is treating system integrity as a leading indicator, not an afterthought, and getting the definition right before the optimization starts.

Watch the episode

If your metrics are moving in the right direction but something still feels off inside your teams, this episode is the conversation you need to have.

Spotify | YouTube