At Amazon, one of the most impactful weekly moments was when, in addition to diving deep on our metrics, we also reviewed anecdotes. In each WBR, our customer support team shared an example of a poor customer experience from the prior week.
These were technically edge or corner cases, but they were emotionally resonant cases that reflected particularly frustrating customer experiences.
In one particular case, a customer bought a $30,000 diamond ring on Amazon. A few hours later, they canceled the order. The hold we placed on their credit card wasn’t lifted immediately, so they couldn’t use it for other purchases. We were unknowingly holding their card hostage.
Now, this customer was probably the only person to buy a diamond ring on Amazon that week, but that wasn’t the point. The value of these examples was in what they revealed — faulty assumptions, process gaps, breaks at the margins. They surfaced defects that could destroy customer trust.
More often than not, when we dug into the story, we found that the issue wasn’t completely isolated. It was the first signal of what could be a broader problem if left unaddressed.
We treated these stories from the customer service team like data points. They gave us direction and highlighted which systems weren’t working as intended.
Many companies err when measuring customer experience by relying on reported data, such as NPS scores, surveys, or star ratings. These tools are blunt instruments— without diving deep into the root causes and CX driving the ratings, they reveal little to nothing; they are output metrics or lagging indicators.
If you want to understand the quality of your customer experience, you have to measure it through instrumentation (input metrics—leading indicators), and you have to dig deeper into real customer stories—every day and every week.
