Here's a thought experiment I keep coming back to. A customer has agreed to answer exactly one question before they walk out the door. Not a survey, not a page of sliders. One question, and then they're gone. What do you ask?
Most people's first instinct is "how did we do?" or some cousin of it. A rating out of five, a smiley face, a "would you recommend us." Those feel safe because they give you a number, and a number feels like an answer. But sit with the constraint for a second. If this is your only shot, a five-star rating buys you almost nothing. You learn that someone was roughly happy or roughly not, and you're left to guess at everything that made them feel that way. You've spent your one question collecting a feeling you already half knew.
The exercise is useful precisely because it's artificial. In real life nobody makes you cut down to one question. But forcing the choice reveals what you actually want to learn, and most feedback forms are built without anyone ever having asked that. They accreted. Someone added a question, then another, then a section, and now there's a fourteen-field form that nobody can explain the purpose of, least of all the customer staring at it.
What the single question should do
If I only get one, I want it to surface the gap between what the customer expected and what they got. Something open, and pointed at the friction rather than the glow. "What almost stopped you from booking?" "What's the one thing you'd change about today?" "Was there anything that nearly went wrong?" The exact wording depends on your business, but the shape is consistent: an open question that gives people permission to say the slightly negative thing they'd otherwise keep to themselves.
That phrasing matters more than it looks. Ask "how was everything?" and you'll get "fine, thanks," because that's the socially frictionless answer and people default to it. Ask what nearly went wrong, or what they'd change, and you've quietly told them that criticism is welcome and expected. You get the near-miss that didn't quite become a complaint, which is usually the most actionable thing a customer can hand you. It's the problem you can still fix before it costs you the next person.
The reason an open question wins the single slot is that it carries its own context. A rating tells you where the needle sits. A sentence tells you why, and the why is the part you can act on. Read a hundred of those sentences and the patterns start to stand out on their own, the same phrase surfacing again and again until it stops being one person's gripe and becomes a thing you clearly need to deal with.
Why "just make it shorter" misses the point
It would be easy to turn this into a sermon about brevity. Kill your long surveys, one question is all you need, and so on. I don't actually believe that, and I think the advice does damage. Plenty of businesses genuinely need to know several things, and there are moments where a customer is happy to give you five minutes because they care about the outcome. A post-purchase survey and an in-depth research interview are different tools for different jobs. Cutting everything to a single field would throw away real signal for the sake of a tidy principle.
The problem with the fourteen-field form was never the number fourteen. It was that nobody could say what each field was for, and that it turned up at a moment when the customer had no patience for it. Length is a symptom. The disease is asking without a purpose, and asking at a time that suits your reporting calendar rather than the customer's memory. A form earns every question it contains when someone can say, for each one, what decision the answer will change. Most can't, and that's the actual fault.
So the honest version of the one-question rule isn't "only ask one thing." It's "know which single thing you'd keep if you had to, and make sure it's actually in there." Once you know that, adding a second or a third question is a deliberate choice rather than an accident of accumulation.
The pairing that earns its place
In practice, one question is slightly too austere. What tends to work is a pair: your one good open question, plus one structured signal you can count.
The structured signal is a single rating, or a yes/no, or a pick-one. On its own it's shallow, exactly as I complained above. But sitting next to the open answer, it earns its keep. It gives you something you can track over weeks without reading every response, and it gives you a way to sort. Pull up every low rating and read only those open answers, and you've gone straight to the customers worth understanding. The number is the index; the sentence is the content. Neither does the job alone, and two fields is a length nobody resents.
Where you place that pair matters as much as what it says. The best-written question still flops if it arrives three days after the visit, when the details have gone soft. When you ask shapes the answer as much as what you ask, and a mediocre question at the right moment beats a perfect one that shows up late.
This is the setup Qria is built around, gathering that structured signal and the open answer together and reading across both with your public reviews in one view, so the rating gives you the trend and the AI summary tells you what the sentences underneath are actually saying. If you're starting from scratch and want the wider picture of channels and methods, the practical guide to collecting feedback covers where the pair fits.
Try the thought experiment on your own form this week. Cross out every question except the one you'd defend to the end. Then look at what's left and ask whether the customer would recognise it as the thing you most wanted to know. If they wouldn't, you've found the one worth keeping.
