CSAT — customer satisfaction score — is the metric most support teams lead with, and it's one of the least trustworthy signals in the function. The problem isn't that customers lie. It's that a positive CSAT can attach to an interaction that completely failed, because satisfaction and resolution are different things, and CSAT only measures the first.
How a failed interaction scores "satisfied." A customer asks a chatbot for help. It can't resolve the issue. They give up, Google it, find a workaround, or solve it themselves. The survey arrives; they rate it positively — out of politeness, because they eventually solved it (without you), or because their expectations were already low. Your dashboard records a win for an interaction where the customer left without the help they came for.
Why FCR is the honest metric. First-contact resolution asks the question CSAT dodges: did the problem actually get solved, in this conversation, without the customer having to go elsewhere? You can't score well on FCR by being pleasant about failing. A support function optimizing for CSAT can look healthy while quietly failing its customers; a function optimizing for genuine resolution can't hide behind politeness.
What to do. Keep CSAT if you like, but stop treating it as the quality metric. Measure genuine resolution, and measure it honestly — including the customers who gave up rather than escalated, who are invisible to deflection counts and flattering to CSAT. The gap between your CSAT and your true resolution rate is the size of the problem you can't currently see.
Frequently asked questions
Is CSAT a reliable support quality metric?
Not on its own — customers often rate failed interactions positively out of politeness or because they self-resolved, so a high CSAT can mask interactions that didn't actually help.
What's a better metric than CSAT?
First-contact resolution, which measures whether the problem was genuinely solved in the conversation rather than whether the customer felt okay about giving up.