Stop, Look & Think - Checking that a problem actually exists

By requiring keen observation before action, by demanding that one look beyond the obvious surface symptoms to better see the deeper causes, by never giving answers and only asking questions, Ohno taught people to stop and think.

~ Matthew E May In Pursuit of Elegance

Perhaps the problem, sometimes, is the notion that there's a problem
~ Oliver Burkman Help!


I had an interesting experience at work a few weeks ago which illustrates why continuous improvement has to start with a clear understanding of the facts, followed by a proper analysis. A knee jerk reaction might be OK if someone hits you on the knee with a little hammer, but otherwise it pays to stop, look and think.

My immediate manager had some concerns about a significant reduction in productivity within our unit. It appeared that we had dropped from processing around 4000 workitems one week to 2000 workitems the next.

Now, that does sound like a big drop doesn't it?

So we had a meeting about it with our team leaders and spent an hour discussing ways to improve productivity. I suggested that before doing this we should compare our productivity with other units to see if there had been a comparable drop, since this would indicate a common cause across units (such as a seasonal increase in sick leave) rather than something specific to ours, but this was dismissed by my manager out of hand. In the end we arrived at a "solution" which was pretty much a "solution" that we had used repeatedly over several years.

So problem solved, right?

WRONG!

There was no problem to begin with. As is my usual practice, I ignored what my manager said and went back to analyse the figures anyway.

The first thing I found was that the reporting cycle had been disrupted. Whereas previously we had received our productivity reports every seven days like clockwork, the last few reports had covered 9 days, then 5 days, then 9 days then 5 days. It doesn't take a rocket scientist to see that almost as twice as much work will get done in 9 days as in 5. The supposed drop in productivity was simply an artefact of the disrupted reporting cycle.

However, it was still possible that there had been some reduction in productivity so I compared the ratios of the number of workitems processed by my unit versus those of two other units. What I found was that the ratios had been fairly constant over the previous 3 months and that if anything our relative productivity had increased over the previous few weeks rather than decreased relative to other Units.

So there was no problem at all to be solved! We spent an hour trying to solve something that didn't exist at a cost that I estimate of over $300. The analysis that I did cost around $20, so if we had instead gone back to the actual data and done some analysis we would have saved the time equivalent of $280.

It has been said that one of the biggest mistakes that businesses make is to try to solve a problem at the same meeting at which it was raised.

This is a typical example of such a mistake.