The Worst Experience is Your Best Friend
There is a folder somewhere in your company. Maybe it's a Slack channel, maybe it's a support queue, maybe it's a tab in a spreadsheet nobody opens on purpose. It is full of customers telling you, in their own words, exactly what is wrong with your business. It has been there for years.
Almost none of it has been fixed.
This is the part of customer experience that nobody says out loud. It isn't that companies don't collect feedback. Most collect far too much of it. It's that the feedback arrives in a form that cannot be acted on, gets compressed into a number that means nothing, and dies in a meeting about the number. We wrote a paper about why that happens and what to do instead. This post is the short version of the part that stings.
The thing your customers actually remember
Start with a finding that should have reorganised the whole industry and somehow didn't.
In the mid-1990s, two researchers recorded the pain of patients undergoing colonoscopy and lithotripsy, minute by minute, as it happened. Then they asked the patients afterwards how bad it had been, and compared the memory against the tape.
What people remembered came down to two things. The worst moment, and the last few minutes. That was basically it. How long the procedure lasted barely mattered. The colonoscopies ran anywhere from four minutes to sixty-nine, and the correlation between how long one took and how bad the patient said it was came out at 0.03. Near enough to nothing.
Then they ran the experiment that proves the point. They took the procedure and added a few extra minutes at the end, with the instrument left in place, still uncomfortable but milder than everything before it. That group went through more total discomfort, for longer. They remembered it as less unpleasant, and they were more likely to come back.
Sit with that for a second, because it breaks something. You can make an experience objectively worse and have it remembered as better, by changing one moment near the end. What lives in your customer's head is a peak and an ending. Everything in between washes out.
Which means the number you are probably tracking, the average, describes an experience no customer ever actually had.
Why the worst moment never reaches you
If the worst moment is the thing that governs whether a customer stays, tells their friends, or quietly leaves, you'd think every company would have it written down. Almost none do. There are four reasons, and three of them are baked into the tools people already pay for.
The customer averages it away before you ever see it. Ask someone "how was your experience?" and they collapse a peak, an ending, a mood and a general sense of your brand into a single digit. By the time that digit lands in your database, the worst moment is gone. It went into the number and it cannot be pulled back out.
Then the scoring averages it again. Bucket people into promoters and detractors, subtract, and a customer who gave you a 6 and a customer who cannot stand you land in the same pile, while a 6 and a 7 end up on opposite sides of the line. Two companies with completely different customers can post the identical score.
Your customers are also just being polite. Most people will not volunteer a criticism to a company that has given no sign of wanting one. Complaining unprompted feels like an imposition, and there's no reason to believe anything will come of it. So they tick a 7 and leave, and the one useful thing they knew goes with them.
And the score gets gamed at the point of collection. You already know this one. "If you can't give me a ten, let me know first." A number you can lift just by lifting the mood in the room is a number that is easy to corrupt, and it gets corrupted constantly.
Four mechanisms, all pushing the same way. Your most valuable data gets destroyed by the customer, destroyed again by the scale, held back out of politeness, and massaged at the point of collection. Then whatever survives gets averaged and put on a slide.
The five-word fix, and the wall you hit right after
There is a question that recovers most of this, and it is five words long.
What was the worst part?
It works because of how it's built. "How was it?" can be answered with "fine", and most people take that exit, which is why your comment box is empty. "What was the worst part?" cannot be answered with "fine". The grammar demands an actual thing. And because the question already assumes a worst part exists, the customer isn't sticking their neck out to complain any more. You've paid that social cost for them, in advance, just by asking.
You can add this to one touchpoint tomorrow. You don't need to buy anything to do it, and you should. Read the answers yourself for a fortnight, the actual sentences, not a sentiment score. Watch what happens the first time you read one aloud in a meeting and notice that nobody argues with it. A number invites a debate about the number. A sentence like "the confirmation email arrived after the courier did" invites somebody to go and check, and then fix it.
And then you hit the wall.
You'll have a folder of true, specific, unarguable statements about what is wrong with your business, and you still won't be able to get any of it funded. Because when you take a customer complaint to a finance director, you get asked how many customers, how much it costs, and what happens if you do nothing. "Several people mentioned it" doesn't get a budget, and honestly it shouldn't.
This is the graveyard. It is full of teams who collected excellent feedback, presented it with real conviction, and were asked for a number they didn't have. The question on its own gets you better anecdotes. Better anecdotes get you a better folder. The folder still changes nothing.
Putting a price on it
Getting the fix funded means knowing what the bad moment actually costs, and that's where the second half comes in. The full paper lays out the mechanism: take a quick reading before the experience and another after, and the difference strips out the customer's mood, their loyalty, their expectations, everything they walked in carrying, and leaves you with what the experience itself did. Attach that to the themes coming out of the worst-part question and the anecdote turns into something a finance director will act on. A named problem, holding a measured share of a measured loss, across a known volume.
That's the whole argument. The worst moment is where your churn is decided. The question recovers it. The measurement prices it. Together they produce the one thing a CX programme was always supposed to produce and almost never does: a ranked list of what to fix, with a number next to each line.
Bill Gates wrote that your most unhappy customers are your greatest source of learning. That line has been on the wall of every CX team for twenty-five years. The uncomfortable part is that the industry then went and built the instrument that guarantees you never hear them.
The paper is called The Worst Part. It covers the memory research, why the standard score fails, the question and how to keep from breaking it, the measurement that prices the answer, and a full section on where our own method could be wrong. If any part of the folder problem sounded familiar, it's worth the read.