What are you doing with your data?
Sept. 29, 2014 – I read with interest a column by Jeff Hajek called “11 Common Continuous Improvement Mistakes You Are Probably Already Making.” Two of the items on the list — “Using averages” and “Collecting data without using it” — struck me as pertinent to any discussion of automation software, ERP software and other automated data collection systems.
In the “Using averages” section, the author explained that the average size, time or other measurement might make it look like you’re within tolerance, when in reality you have a high defect rate or a number of outliers that aren’t reflected in that average. Any average without a measure of dispersion or variability will be useless because it’s uncommon for averages alone to help you improve processes. He goes on to say, “Let’s say that your takt time on an assembly line is 12:07, and your average cycle time is 11:52 on all of your stations. Using the average would tell you that you are in good shape. However, you know nothing about how often you have to stop the line. The same holds true in virtually any meaningful metric. The average ship time compared to your target time tells you very little about on-time delivery…Make sure you have some way of looking at the spread of your results rather than just the average.”
The same advice applies when using automation. You can collect vast amounts of information without answering the question of what you’ll be doing with it, whether everyone understands it and whether it will help you improve your processes.
In the “Collecting data without using it” section, Hajek makes the point that everything you do should be adding value to the customer — everything you do should have a purpose. I agree.
We often forget to ask how this step will add value for our customers during customer-facing processes because we get caught up in the way that we’ve always done things. This only gets worse when we think about our management and administrative processes. When I deal with manufacturers, I often tell them that delays are caused by inefficiencies in the office.
Very often we believe that the more information we collect, the better off we’ll be. We spend huge amounts of resources collecting data. Then we spend more resources sorting, processing and compiling it. But how many times do we actually use that data? Typically, we never get around to actually using it to make an improvement. There is no return on that investment.
Here are some things to keep in mind as you automate your data collection to fit in with your lean endeavours:
• Keep the numbers simple. Collect them in real time. Make sure that everyone knows what those numbers mean. Don’t have numbers that only one area will recognize. For example, one factory I was in had fancy charts set up — different ones for different areas. They were fed from their ERP system, but none were in real time. I set up a simple Andon board that used real-time data. The simple numbers meant more to them than the data that was a day or two old. We took the time to complete a task down from over 11 hours to just two and a half with the Andon board!
• If you’re going to automate your data collection, make sure you’re collecting the right data, displaying it at the right time and using it for the right purpose. Ideally, this data should be collected automatically and displayed immediately. If you’re a small shop, this still holds true (although you might have to make do with paper charts and tables instead of computer display screens). I often tell my clients that it doesn’t matter how the information gets down to the floor; it’s only important that it does.
• Co-ordinate with your ERP provider to develop an Andon board to show the status of all the jobs you’ve got going on. Knowing is better than guessing! Let people know how they’re doing currently and let them know how well they’re doing compared to standard. This information will be motivating for them. After all, engagement surveys tell us time and time again that people want feedback about how well they’re doing and they want their managers to listen to them.
Question from the floor
Question: I’m trying to bring lean to a multi-plant operation, but it’s looking as if I’m getting as many different versions of lean as I’ve got locations or departments! How can I rectify this?
Answer: I’ve seen this happen when Factory “A” tries something, and then Factory “B” tries something else. They wind up following divergent paths and everyone gets their own version of lean. It’s best to try to co-ordinate the sites and start with something simple (i.e., a 5S or a kaizen), and then have them compare their notes by establishing a competition. First hold this “Kaizen Off” within each facility, and then the leaders go on to compete across facilities. By leading the way within the plants, you’ll be lining them up and showing them how to do the 5Ss or kaizens. By having them compete across plants, you’ll be reinforcing the standard approach to continuous improvement. I find that having the kaizen teams present in-plant to a management meeting works best. It brings the management team in on the kaizen efforts and lets everyone in the plant know what’s what. Try this first before you try co-ordinated training or hiring a development person. Remember, it takes about two years to build up a lean culture, so take your time.
This column originally appeared in the September 2014 issue of Manufacturing AUTOMATION.