We find problems before your customers do

The problem

Something breaks at seven on Thursday evening. Nobody at the company notices, because nobody's looking. The first you hear of it is a forwarded complaint on Monday morning. By then it's been broken for the whole weekend and affected everyone who tried to use that part of the system.

Then comes the second problem. You ask what happened. The developers look into it, try to reproduce it, and can't. The answer comes back as "we think it was probably a timeout" or "it might have been a bad bit of data". A fix goes in based on the best guess. A month later it happens again.

Both problems come from the same gap: the system isn't telling anyone what it's doing.

What we do

Alerting

However many checks are in place before a release, occasionally something slips through. When it does, we want to know immediately.

The live system reports every error to a monitoring service the moment it happens. That service groups them, works out whether it's something new or a repeat, and alerts us. A problem at seven on Thursday evening is a notification at seven on Thursday evening, not a complaint on Monday.

The alerts also carry the detail: which page, which user, what they were doing, what went wrong. So the person picking it up starts with the facts rather than a description passed on second hand.

Logging

Alerting tells us something went wrong. Logging tells us what happened around it.

The system keeps a structured record of what it's doing: requests coming in, decisions made, things sent out, and where any of it went sideways. When there's a problem, we can read the sequence of events that led to it, rather than trying to recreate it from a customer's description.

This is the difference between "we think it was probably X" and "it was X, here's the fix". That difference is usually hours, sometimes days, and it's also the difference between a problem that's fixed and one that comes back.

What it means for you

You find out about problems from us, not from your customers, and usually before your customers have noticed. When you ask what happened, you get an answer rather than a theory. And problems get fixed once, because we fixed the actual cause.