A bloke I know runs a small flooring business. Six fitters, decent work, been going twelve years. He rang me a few months back absolutely fuming because a customer had left him a one star Google review about a job his lads finished three weeks earlier.
The complaint was small. The fitters had left offcuts in the back garden instead of taking them away. That's it. A five minute fix if anyone had known about it.
Nobody knew about it. The customer didn't ring. Didn't email. Didn't mention it when the invoice went out. She stewed on it for three weeks, then typed it into Google where nine hundred future customers could read it.
Here's the thing that made him go quiet when I said it. That's not one unhappy customer. That's one unhappy customer who happened to be annoyed enough to type. The rest of them just go away.
Most of them never tell you
Qualtrics XM Institute asked consumers what they actually did after a very bad experience with a company. Only 31.3% sent feedback directly to the company via a call, letter, email, or on their website.
Look at that number the other way round. Roughly two thirds of people who had a genuinely bad experience never bothered to tell the business about it.
They didn't complain. They didn't give you a chance to fix it. They just decided, without a word, that you were a bit rubbish, and next time they needed that job doing they rang someone else. No drama, no argument, no feedback. Just a customer who used to spend money with you and now doesn't.
And you'll never see that in your numbers. Revenue doesn't flash red when a happy customer turns into an ex customer. It just sits slightly lower than it should, month after month, and you blame the market.
What it's actually worth
Novuna Consumer Finance ran the maths on retention for UK retailers. Their findings: securing a new customer is roughly five times more expensive than retaining an existing one, and converting a new shopper costs about 29% more. And they also found that improving retention by just 5% can lift profits by 25% to 95%.
So the customer who silently walked isn't a small loss. Replacing her costs five times what keeping her would have. And you didn't even get told what went wrong, so the same thing is happening to the next one.
So my flooring bloke does about sixty jobs a month. If even a handful of those end with a mildly annoyed customer who says nothing, that's a few thousand quid a year walking out of the business through a door he can't see.
The fix is one question, sent automatically, at the right moment
This isn't a customer experience programme. It's not a quarterly survey. It's not a consultant with a framework. It's one message, sent by a machine, a few hours after the job is done.
Something like this:
> Hi Sarah, Dave here from Ashworth Flooring. Quick one: out of 10, how did we do today? Just reply with a number.
And that's it. No form. No login. No "we value your feedback". One number, sent by text, while the job is still fresh in their head.
Then the machine does the sorting:
- 9 or 10. Reply with thanks, then a second message asking for a Google review. These are your happiest people and nobody ever asks them.
- 7 or 8. Reply with thanks and a single open question: what would have made it a 10? That answer is worth more than the score.
- Anything 6 or below. Stop. Don't send a review link. Ping the owner's phone immediately with the customer's name, the job, and the score. Someone rings them that day.
That last branch is the whole point. You've just bought yourself the chance to fix the offcuts in the back garden on the same day, instead of reading about them three weeks later underneath a one star rating.
Why timing beats wording
But every business owner I've said this to has had the same objection. "We tried a survey. Nobody filled it in."
Yeah. Because it went out as an email with a link, a week late, from an address nobody recognised.
Retently studied 25 million survey invitations sent by 600 ecommerce brands across 2025. Surveys sent by email got an average response rate of 3.24%. Surveys asked in the moment, in the app, at the point the thing actually happened, got an average response rate of 32.34%, roughly 10 times higher than email.
Same question. Same customers. Ten times the answers, purely because of when and where you asked.
Do the maths on sixty jobs a month. A generic email survey at 3.24% gets you two responses. Two. You can't spot a pattern in two. Ask at the moment the work finishes and you're looking at a proper handful of real answers every single month, including the ones from the people who were about to go quiet on you.
Now, that study is ecommerce and in-app, and he fits carpets in Stockport. The exact percentages won't transfer. The principle absolutely does: ask right after the moment, in the channel they already use, and you get answers. Ask a week later, by email, with a form, and you get silence.
Building it, roughly
Tool-agnostic, because this sort of thing works in whatever you already pay for.
1. Find the trigger. Something in your business already marks a job as finished. An invoice going out. A job status flipping to complete. An appointment ending in the calendar. That event is your trigger, and you probably already have it.
2. Wait a bit. Two to four hours is about right for a service business. Long enough that they have made a cup of tea and had a proper look at the work as well. Short enough that they still remember it.
3. Send one question by text. Not email. Text. It gets read. Ask for a number, nothing else, and keep it human.
4. Branch on the answer. High score goes to a review request. Middling score goes to one follow-up question. Low score goes straight to a human being's phone with a name and a number attached. Every automation tool on the market can do an if/then on a number.
5. Log everything in one place. A spreadsheet is fine. Date, customer, job, score, comment. After three months you'll spot things you genuinely did not know. One of my clients found every low score came from the same two-day window each week, which turned out to be when their most experienced installer was off. That's not a customer problem, that's a rota problem, and no amount of marketing would have fixed it.
6. Get AI to read the comments, not the scores. The scores are easy. The free text is where the money is, and nobody reads it after week three. Have an AI summarise the month's comments into three themes and one recommended action, then send that to you on the first Monday of the month. Takes about twenty minutes to set up and it turns a dead spreadsheet column into something you actually act on.
What changes
So the obvious win is the reviews. You stop asking everybody and start asking the people who already told you they're delighted, so your rating climbs instead of rolling the dice every time.
But the bigger win is a lot less obvious. You find out about the offcuts in the garden the same afternoon. You ring, you apologise, you send someone round. She tells three friends how well you handled it instead of telling Google how badly you did the job.
And the biggest win doesn't show up as a review at all. It's the customers who would have drifted off without a word, who now get asked, and who tell you the small thing that was bothering them. That's the two thirds. That's where your missing revenue lives.
One trigger. One question. One branch. An afternoon to build.
The alternative is finding out three weeks late, in public, from someone who'd already decided not to come back.
Sources:
- How Consumers Give Feedback (XMI_DS_HowConsumersGiveFeedback-2020), Qualtrics XM Institute, November 2020
- Survey Response Rate Study, Retently, March 2026
- Why customer retention is crucial for retailers, Novuna Consumer Finance, July 2025
Want to try this yourself? Grab the free AI Setup Kit. 77 plug-and-play skills for content, sales, marketing and operations. Works with Viktor, Claude or ChatGPT. No course, no tool to configure, just markdown files and your AI.
Brewed by Steven, poured by Viktor
About Steven Tann: Steven helps business owners build systems that run themselves using AI. After 10+ years helping 7,000+ businesses and building his own autonomous operations, he's the bloke who actually does it, not just talks about it. Find out more at steventann.com.