Computerworld Archive

Misunderstand the symptoms, fail with the cure

Article comments

If you go to see your doctor they are unlikely to diagnose an illness or even order tests based on a vague description of “I feel unwell”. They will ask you about your symptoms.

They will want to know if you feel unwell all the time or just at certain times. If they sent you off to have a battery of tests just based on the description of “unwell” you’d think they were an idiot.

A few years ago we were asked to look at a problem with an insurance brokerage system. The users said that the system was too slow. Each of the technology teams had spent a considerable amount of time checking the infrastructure for an underlying problem.

The first thing we did was visit some users and what they told us was a revelation. A change had been made to the application that meant the user had to pass through two additional screens to complete a business process, and this was the complaint. It was a design issue and nothing at all to do with infrastructure.

It’s worrying that an IT department is happy to throw man days of effort at problems based on a description like “the CRM system is slow”. What’s slow – all transactions, or just some? And how slow is slow?

We’ve just completed the investigation into a problem impacting tens of thousands of users for four months. A team of twelve very bright people have spent more than 300 man days trying to determine the root cause, all based on a problem description of “it goes slow at month end”.

Maybe you’re thinking that these are unusual examples, but in my experience they’re common. I’ve sat in countless problem investigation meetings where techies (and I’m one) spend hours describing in the finest detail how a piece of technology works and the configuration of a system, only to discover that we are all working off a vague symptom description.

So here’s a tip straight out of the RPR Problem Diagnosis Method – you need to understand a symptom to a level such that, in theory, you could walk up to a PC and replicate the user actions that resulted in the problem. If you are investigating a problem and you don’t understand it to that level, you’re probably wasting your time.

Email this to a friend

* indicates mandatory field

ComputerWorldUK Webcast