I once watched a restaurant change its background music because one customer said it was too loud. Not a complaint that came up repeatedly, and not from a regular. One person, once, on a Tuesday. The volume came down and stayed down, and about three weeks later the place felt like a waiting room, and nobody could quite say why. The owner had been responsive. He'd also been wrong.
The opposite failure is more common but harder to notice, because nothing visibly happens. Someone writes a careful message explaining that your online booking form rejects addresses with apostrophes in them, and it gets filed under "one person's opinion" alongside the music complaint, and it sits there for a year while everyone named O'Brien fails to book.
Both mistakes come from the same place: treating a complaint as a unit of measurement. It isn't. One complaint tells you what one person experienced, and whether that's worth acting on depends entirely on what kind of claim they've made.
Facts, preferences, and the difference between them
Some complaints report a checkable fact about your business. The form rejects apostrophes. The confirmation email doesn't say which entrance to use. The toilet lock is broken. The price on the website is different from the price on the menu. These need no corroboration, because the number of people who reported it is irrelevant to whether it's true. Go and look. If it's true, fix it. One person is a perfectly adequate sample for the existence of a broken lock.
Other complaints express a preference. The music is too loud. The lighting is too dim. The portions are too big. The tone of your emails is too casual. These are real experiences and they're not facts about your business, they're facts about the fit between your business and that person. Acting on one of them means changing your business to suit one customer's taste, which is only a good idea if you happen to have picked the right customer.
The music was a preference dressed up in the grammar of a fault. That's what fooled the owner. "The music is too loud" sounds like "the lock is broken," and it isn't, and there's no way to tell from the sentence alone. You have to ask what kind of claim is underneath it.
A third category sits between them: the preference that repeats. Once four separate people have mentioned the music, you're no longer looking at one person's taste, you're looking at a mismatch between your room and your customers. That's a signal. The difference between the complaint and the signal is entirely the repetition, which means the only way to see it is to have written the first three down.
Why the loud ones win
In practice, what determines whether a complaint gets acted on is rarely how well evidenced it is. It's how vividly it was delivered and who heard it.
A furious review at eleven at night produces action the next morning. A polite note in a form produces a nod. A complaint made directly to the owner's face outranks fifty responses in a folder, because the owner remembers it, and memory is what feels like evidence. If your mother in law mentions something about your shop, that complaint has roughly the weight of a small survey, and it shouldn't.
That's how businesses end up with a set of changes that traces the history of who complained loudest rather than what customers generally need. Each individual decision felt responsive. The cumulative effect is a business shaped by its most confident critics.
The habit that fixes it
Write complaints down and act on the second one. That's most of it.
For anything that reports a checkable fact, act immediately, because the cost of checking is ten minutes and the cost of being wrong for a year is a steady trickle of people who couldn't book. For anything that expresses a preference, log it and wait. Not forever, and not as a way of ignoring people, but because the second and third instance are what turn it into information. When the count reaches three or four, it's earned a decision.
The logging is the part that fails. Most businesses have no place where the music complaint from March and the music complaint from June end up next to each other, so the June one arrives as a fresh single data point and gets the same shrug. Any list works, as long as it's the same list. When the same complaint shows up twice is about that moment of recognition and why it's easy to miss.
At low volumes this matters more, because a single response is a bigger fraction of everything you've got and the temptation to read a trend into it is stronger. Reading feedback when you only have a dozen responses covers the sorting in more detail, and it lands in the same place: check the facts, park the preferences, don't calculate percentages.
Once you've got a list of parked preferences, the next question is which ones to spend on when several have crossed the threshold at once. That's a prioritisation problem rather than a listening problem, and how to prioritise customer feedback deals with it properly.
Seeing the count without doing arithmetic
The practical difficulty is that complaints arrive in different places. One in a Google review, one in a form response, one said out loud at the till and never recorded anywhere. Those three are the same signal and they look like three unrelated events.
This is the specific job Qria does: pulling the reviews you get publicly and the structured feedback you collect yourself into one view, with the AI reading across both and telling you what keeps recurring. It won't tell you whether a complaint is a fact or a preference, that's still your call, but it will tell you it's the fourth time. Which is usually the only piece of information you were missing.
The restaurant, for what it's worth, put the music back up about two months later, after enough people said it was flat in there. Two preferences, one reversal, and roughly ten weeks of a place that didn't feel like itself. Cheaper to have written the first one down.
