Data Out of Thin Air

The caption is from "The Rime (Rhyme) of the Ancient Mariner." It refers to the plight of sailors stranded in the middle of the ocean, surround by water that couldn't slake their thirst. In a similar manner, do you sometimes find yourself awash in data, but none of it useful? It's surprisingly common.

In many Lean implementations we find that we need different data than that captured by computer systems. For that reason, one of the first things we do is to teach workers how to manually capture data. We use simple tools like check sheets (right).

Collecting data is surprisingly easy. You teach the workers what conditions are important and what "good" looks like. If you are already using Standard Work, they'll know that. Then you ask them to record when and how a condition is violated.

To make this simple and keep inputs uniform, we often introduce defect codes: e.g.

L = too Long; S = too Short; T = too Thick; etc.

With just a few pieces of data, you can begin to understand your problems.

In a factory where defects kept slowing down the production process, we asked workers to use a blank piece of paper to write down the DATE, the PROBLEM, the DEFECT CODE and the DATE RESOLVED. Later, we'd introduce the name of the person accountable for resolving the problem. We reviewed these data daily.

With that little information, we began to better understand our problems. We tallied all the information from the sheets then began charting that data. The problems became visual.

In very little time, a picture emerged. It indicated that we frequently began work without a complete parts kit. We worked on and improved that.

As we got more sophisticated, we realized that parts weren't getting kitted because the Bills of Material (BOM) were wrong or incomplete. We worked on improving the BOMs and defects dropped again.

In another industry, the only defect data came from a final inspection process. Catching defects there meant delays as we tore the product apart to make the repair. Moreover that method of discovery rarely pinpointed the root cause of the problem.

We implemented an "in process" inspection. Before "adding value" themselves, each worker inspected the Work In Process (WIP) coming to them. If there was a flaw, it was recorded and the WIP returned to the previous worker.

Looking for What, not Who

When we had enough data to see a picture emerging, we gathered workers and managers to discuss what was driving our defects. You can expect to get workers complaining they were rushed or didn't have the proper tools. We took those claims seriously. After all, we were looking for what, not who.

We conducted time studies of workers with good quality records to determine a fair time. Then that became the standard.

We also looked at tool problems. We repaired or replaced them when needed. We also set about retraining workers in the proper way to perform their job. Defects dropped.

As most of you know, when workers realize their concerns are being taken seriously, they become more engaged, better workers. Morale improved. So did both our First Pass Yield and the speed at which the product moved. Everyone won, and it all began with data collection.