Why do people really leave their internet provider?

The question

Every internet provider wants to know why customers leave. The easy answer is "the other guy is faster." It's easy because you can measure it. Run a speed test, compare the numbers, done.

I wanted to know if that was actually true, or just the answer that's easiest to count. Is it performance? Is it price, customer service, or how consistent the connection feels on a Tuesday night when everyone in the house is streaming? So I went looking for the parts of the experience that don't show up in a headline speed number.

The plan

Crowdsourced network data lets you look at the experience from the user's side instead of the network's side. Millions of people run speed tests on their own devices, in their own homes, on their own schedule. That gives you three kinds of evidence:

• Quality of service: the classic numbers, like download, upload, and latency.
• Quality of experience: what using the internet feels like. How long a web page takes to load, how long before a video starts, how laggy a video call is.
• Sentiment: how people rate their provider and whether they'd recommend it (Net Promoter Score, or NPS).

Then I lined those up against which providers were gaining and losing customers.

First clue: the fastest isn't the whole story

Comparing the US provider with the most low-latency connections against a major competitor, the speed gap was real but modest. The gap in how people felt about them was enormous.

A 32% speed advantage doesn't explain a swing from −16.5 to +20 on its own. Nobody feels megabits. What you feel is the page that hangs for half a second, or the lag on a video call with your boss.

Second clue: fiber shows up everywhere

Next I compared fiber providers with non-fiber providers city by city. Fiber was faster. Shocking, I know! But the gap that lined up most cleanly with how people felt wasn't download speed. It was latency.

Then the movement data closed the loop. Tracking which providers were gaining and losing users quarter by quarter, the operators investing fastest in fiber were the ones gaining and keeping customers in Houston, Los Angeles, Philadelphia, and Seattle.

The part where I overthink it

Here's the trap: "fiber providers have happier customers, so fiber makes customers happy." Maybe. But fiber providers also tend to be newer in a market, price differently, and run different marketing. The data shows these things move together. It doesn't prove one causes the other.

And there's a wrinkle that keeps me honest. Look at Philadelphia in the full table: non-fiber customers there gave their provider a higher star rating than fiber customers did, 4.0 vs 3.5. Yet the non-fiber NPS was still far lower. Ratings and recommendations are measuring different things. If you only looked at one, you'd tell yourself the wrong story.

"Who's fastest?" is the easy question. The useful one is where the experience breaks, and whether anyone notices.

That's why you need several kinds of evidence pointing the same way before you act: performance, experience, sentiment, and actual customer movement. One metric on its own is a guess. When four of them agree, you've got a story worth telling.

What happened next

Stories like this one change decisions. Analysis built on this approach helped drive a dramatic shift in how one provider marketed and packaged its services. The useful part was seeing what customers were actually reacting to, and it turned out to be a lot more than speed.