Quality & Research
Customers are not your canary
Complaints are a late signal from a tiny minority; most harmed customers leave silently. Find defects in testing, not in production.
Teams that treat customer complaints as their defect-detection system are outsourcing quality control to the people paying them. Complaints arrive late and from almost nobody: most harmed customers never say a word, they just leave. Real quality assurance happens before release, under conditions that mirror actual use.
Some teams move fast and break things, treating customer complaints as their defect-detection system. Ship it, wait for tickets, fix whatever gets reported. The approach fails twice. Complaints arrive late, after the damage is done. And they barely arrive at all: most people who hit a defect never say a word. They abandon the task, lose a little trust, and quietly take their business elsewhere. Using customers as your canary means the canary usually dies silently, and you keep mining.
Why it matters to the business
The math is brutal. Esteban Kolsky's ThinkJar research found only 1 in 26 unhappy customers complains, meaning the other 25 churn silently, and 67% of customers cite bad experience as their reason for leaving. A complaint-driven quality process therefore sees a small fraction of the harm, and sees it last. The damage is also fast: PwC's study of 15,000 consumers found 32% will walk away from a brand they love after one bad experience. Even satisfaction surveys will not warn you. Bain's Reichheld found 60% to 80% of lost customers said they were satisfied or very satisfied shortly before defecting. A quiet inbox is not evidence of quality.
How to use it
- Move detection upstream: test every release under realistic conditions, on real devices, before customers touch it.
- Treat each customer-reported defect as a process escape; root-cause how it got past testing, not just the bug itself.
- Instrument silent failure directly, through task abandonment, error rates, and journey drop-off, so harm shows up without a ticket.
- Track an escape rate, the share of defects found by customers versus found internally, and drive it down each quarter.
- Protect speed by shrinking release scope, never by cutting testing rigor.
Where teams get it wrong
They mistake a quiet support queue for a healthy product. Silence is the sound of customers leaving. Move fast and break things becomes outsourcing quality control to the people paying you, and the invoice arrives as churn nobody can trace back to its cause.
Ask your team
- What share of last quarter's defects did a customer find before we did?
- How do we detect harm from the customers who never contact us?
- When a customer reports a bug, do we root-cause the process gap or just patch the bug?
It's not the customer's job to tell you that you got it wrong.
Apply this
Reading about customers are not your canary is one thing. Seeing where it applies in your journey is the useful part.