> because customer experience isn’t a KPI and so won’t get them promoted?

In a way, I think you may be on to something. At least indirectly.

Amazon are constantly pushing the narrative that they are obsessively customer-focused and constantly thinking about long term. I have my reasons to believe this might have been true at one time, but now is mostly about optics.

Having interviewed at the company a few years back, there sure was no long-term thinking about anything they internally advocate for potential new hires.[ß] To be honest, I was only exposed to one segment of their business, but I still find it hard to believe that kind of culture would be limited to a single part.

ß: an extreme example: one of their more senior managers explained to me that a 1-year old team is already too old and ready to be disbanded. Knowing, from personal experience, that it can take two years to build a well-functioning team - that was a massive red flag.

During my time there I never saw any evidence that they cared about employee retention. The same pain points came up year after year on the tech survey, and were invariably ignored.

Amazon is customer obsessed in the same way that Google does no evil.

Amazon may have created a more object way of measuring the same thing, but without needing to actually ask customers. They probably know at least a some of garbage products sold on their platform. If that's the case, merely tracking visits and revenues from customers who received poorly made products and comparing with customers who've received well made products over time probably produces a more accurate picture of the effects that low-quality products have on their bottom line over time.

The metric you’re talking about is “churn.” It’s distinct from NPS, as NPS predicts growth—the additional value your brand has in people’s minds above the threshold for leaving, where it becomes a virality coefficient—whereas churn is just a self-evident measure of the percentage of people who see your brand as having value below the threshold.

The fact that NPS (and measures like it) are predictors is useful, in that you end up also collecting real growth data, and the difference between NPS-predicted growth and actual growth can tell you how much of your growth was organic vs. marketing-driven, and thus give you a picture of your brand’s “inertia”—how valuable it is just sitting there doing nothing, separate from any marketing projects. Tracking that allows you to judge the brand impact of everything you do. But you can’t do it without some measure that actually gives you brand value, rather than simple growth.

FWIW, I worked on a team that built a platform which used customer activity to predict their NPS score. The algorithm was straight-forward and the results were pretty accurate; enough so to be used by A/B testing platforms for testing new features on users. A side-effect of our research was insight into the key features that dictated a positive response. Even with a complicated product, success was highly dependent upon about four of the 200 or so features we collected.

I suspect that the minds at Amazon could come up with a much better solution than what we did. By virtue of having smarter, more experienced people on average, and a more simplistic product.