Resources · Inspection
Audits that actually find something.
An audit that has never found a problem is not evidence that there are no problems. It is evidence that the audit does not work, and that is how it will be read.
A clean audit is a warning sign
Twelve months of green ticks describes no service that has ever existed. Everybody in the room knows it, which is the problem.
Audits go wrong in a predictable way. They start as a genuine check, become a monthly task, then become a task somebody completes at the end of the month from memory, and finally become a form that is filled in because it has to be filled in. At no point does anybody decide to stop looking properly. It just erodes.
The tell is the findings column. If it is empty, or says "no issues identified" every month, the audit has stopped being a check. And an inspector reading twelve identical clean audits will not conclude that the service is flawless. They will conclude that governance is not working, which is a far more serious finding than whatever the audit should have caught.
The counter-intuitive part: an audit that finds three things every month and shows all three fixed is stronger evidence of a well-run service than one that finds nothing.
Test your own governance
The inspection check looks at whether your audits find anything and whether findings get closed. Those two questions carry most of the weight.
Why audits stop finding things
Three causes, in order of how often we see them.
The auditor audits their own work
The person who wrote the care plans is checking the care plans. They are not being dishonest; they simply cannot see it fresh. Nobody can mark their own homework and find much.
The questions only have one answer
"Are care plans up to date?" invites yes. "When was this plan last changed, and what changed in the person's life that prompted it?" cannot be answered without opening the file.
Findings go nowhere
The audit finds something, it is written down, and nothing happens. After that happens twice, the person doing the audit stops finding things, because finding things has been shown to be pointless.
How to build one that works
Four changes. None of them make the audit longer.
Sample properly, and small. Three files chosen at random, read carefully, beats twenty files skimmed. Choose them before you know what you are looking for, and never let the auditor pick.
Ask questions with a real answer. Replace every yes-or-no question with one that requires reading something. Dates, names, what changed. A question you can answer from memory is not an audit question.
Give every finding an owner and a date. Not "to be addressed". A person and a day. Then start the next audit by checking the last one's findings, before anything else.
Rotate who does it. A deputy from another service, a manager from a sister site, somebody who was not involved. Fresh eyes find things that familiar eyes cannot.
If your audits have been clean for a year, do not back-fill them. Start a real one this month, let it find what it finds, and record honestly that the approach has changed and why. A service that has just improved its own governance reads well. A service with a suspiciously perfect history does not.
Required audits are specified elsewhere
Which audits a service must carry out, how often, and what must be reported are set by the regulator and vary by service type. Take the list from the source and treat it as the floor.
This guide is about making the audits you already run capable of finding something. A required audit that finds nothing satisfies the requirement and protects nobody.
Ring us before it is urgent. It is a much shorter conversation.
Thirty minutes, no charge, no obligation. If you want to know what an honest audit would turn up in your service, we can run one with you.
Text or WhatsApp
07426 919823If it is easier to write it down than say it out loud, send a message.