The CX Metric That’s Lying to You (And the Data That Proves It)

I have a confession to make about CSAT.

I've spent more than 30 years in customer experience research, building VoC programs, leading operations teams, and sitting across from executives who wanted to know, in plain terms, whether their customers were happy. For most of that career, CSAT was part of the answer I gave them

It still is, but not the way most organizations use it.

The problem is: CSAT tells you how customers feel about a specific interaction. It was never designed to tell you whether they're staying. And somewhere along the way, organizations started confusing the two. The result is a reporting environment where the scores look fine, leadership feels reassured, and customers quietly leave.

We have the numbers to prove it.

What the Q1 2026 Data Actually Shows

Every quarter, QuestionPro's Experience Benchmarks study measures NPS, CSAT, and Customer Effort Score across eight U.S. industries and dozens of brands. In 2026 we added Technology, which immediately topped every metric. Interesting, but not the story here. The story is in banking.

Banking and Credit Unions posted the lowest NPS of any industry we tracked in Q1 2026, at 19. That same quarter, CSAT for the sector was 72%, which sounds like nearly three out of four customers are satisfied. It is also the lowest CSAT in our study. But the gap between how those two numbers feel is significant. A 72% satisfaction rate registers in most executive dashboards as "room for improvement." A 19 NPS registers as a problem that is actively destroying your customer base.

Both numbers describe the same customers, same companies, same quarter. That's not a coincidence; it's a structural feature of how CSAT is designed.

CSAT Measures the Moment. NPS Measures the Relationship

To be clear: CSAT is not a bad metric. It's a good one that gets misused.

CSAT asks how satisfied you were with this interaction, a specific, answerable question. NPS asks whether you'd recommend the brand, a question that carries the full weight of every interaction you've ever had with it, the good ones and the ones you remember at 2 in the morning.

Look at the full 2025-2026 banking data and you see something instructive. In Q1 2025, Banking and Credit Unions had an NPS of 41. By Q4 2025, it had dropped to 26. By Q1 2026, it was down to 19. That is a 22-point decline over four quarters. Meanwhile, CSAT went from 82% in Q1 2025 to 72% in Q1 2026. A 10-point drop that, by itself, would not set off most internal alarms.

The NPS trajectory was the warning signal. CSAT was the lagging indicator that made everything look more manageable than it was.

The Brands Where the Gap Is Most Visible

Let me give you a specific example that makes this concrete.

Capital One posted a CSAT of 76% in Q1 2026, respectable, the kind of number that gets a "holding steady" in a weekly review. But their NPS was 21, and that's also a trailing-indicator story: Capital One's CSAT dropped from 90% in Q1 2025 to 76% in Q1 2026, while their NPS dropped from 53 to 21 over the same period.

CSAT looked like a manageable decline, but NPS told you the relationship was fracturing.

This is not unique to Capital One. Hyatt's CSAT was 81% in Q1 2026 while its NPS fell to 41 from 58 a year earlier. Southwest's CSAT held at 76% while its NPS dropped to 16 from 48, following policy changes customers experienced as working against them

In each case, the CSAT created a floor that felt acceptable but the NPS was telling a different story underneath it.

Why Organizations Keep Trusting CSAT Anyway

CSAT is easier to explain. '76% satisfied' processes in three seconds; NPS needs context, and its smaller numbers feel less impressive, even when they're more meaningful.

There's also a timing issue. CSAT deploys right after an interaction, while it's still fresh. The friction from prior visits, last month's hold time, the renewal they dreaded, none of it shows up. The score reflects the best version of the relationship, not the full one.

This is not a flaw in any research team's work. It's a structural feature of what CSAT measures. The question is whether you're using it to understand loyalty, or just to feel better about it.

Customer Effort Score: The Metric That Closes the Gap

In 2026 we replaced Likelihood of Repurchase with Customer Effort Score in our methodology. CES asks how easy it was to interact with a brand, and effort is one of the strongest predictors of churn. It captures what NPS and CSAT both miss: friction that accumulates quietly, below the threshold of a complaint.

In our Q1 2026 data, Banking and Credit Unions and Insurance both landed at 76% ease, the lowest of any industry we track. Digital-first sectors like streaming and phones came in at 86-88%. That gap isn't just a technology story. It's an operational one, about where friction lives in the journey and whether anyone owns removing it.

When CES drops, it almost always precedes an NPS decline. Customers find interactions difficult, disengage, stop recommending, and eventually leave. CSAT stays stable through all of this because individual interactions, when they happen, are handled reasonably well. The problem isn't any single touchpoint. It's the cumulative weight of dealing with the organization.

The Local Credit Union Exception

I want to end with something that gives me genuine optimism, because this piece should not read as a condemnation of measurement itself.

In Q1 2026, Local Credit Unions, a compiled aggregate of smaller regional credit unions tracked anonymously in our study, led the banking and credit union category across all three metrics: NPS of 47, CSAT of 82%, and CES of 89%. That's not a coincidence. Local credit unions run on a different relationship model: simpler products, more consistent human interactions, and a member-ownership structure that aligns their incentives with the customer's, not against it.

The result shows up in the data: their NPS is 28 points higher than the industry average, their CSAT is 10 points higher, and their CES is 13 points higher.

This matters for CX practitioners because it illustrates something I have believed for a long time: the organizations that score well on NPS are almost always the ones that have reduced friction and built genuine accountability into the customer journey. They are not gaming a metric. They have built something that earns the score.

That is the point. CSAT is not lying to you because it is a bad metric. It is lying to you because you are asking it a question it was not designed to answer. If you want to know whether your customers are satisfied with their last interaction, CSAT is a reasonable tool. If you want to know whether they are staying, you need to look at NPS. If you want to know why they are leaving before they do, you need to look at Customer Effort Score.

Use all three. Understand what each one measures. And stop letting your highest CSAT score of the quarter be the headline in your next executive review.

The customers who left last quarter were probably satisfied the last time you asked.

Data sources: QuestionPro Experience Benchmarks, Q1 2026 (n=1,200, survey conducted March 2026); quarterly comparisons from Q1-Q4 2025. All data from proprietary QuestionPro research.