The Data Heads

I mentioned in yesterday’s post announcing the release of GainSeeker Suite Version 8 that I have been working on a case study about the new report writer.

You can read the case study here, and download a PDF version to share with others.

A little bit of the back story:

This project came about during a training class we held for Valeo (the subject of the case study) early in the new year. During the class, Stacey (who is quoted extensively in the case study) made the comment “You’ll never find a kid who wants to grow up to be a data jockey.”  What a great comment.

Mel (the guy from our staff doing the training) was intrigued and scratched away at it. What did he mean by data jockey? Why had he become a data jockey? Who cared about the results of his data jockey work? What difference would it make if we could eliminate the data jockey work? What would it take to eliminate the work?

At the time, Version 8 hadn’t been released, and Mel had had only minimal exposure to the power of the new Dynamic Reports module. He came back and started asking his colleagues “So this is what they really want. Could Dynamic Reports handle it?” Dale, one of our senior developers whom I sometimes call Obi-Wan Kenobi, put together a prototype and we were off to the races.

In manufacturing circles it not at all uncommon to talk about “The hidden factory.” The hidden factory is rework. Another case study about our customer Titleist includes this analogy from the rubber and plastics industry:

Another benefit of reduced scrap is that equipment is freed to do productive work. A shop with a 5 percent scrap rate and 20 molding machines has one machine dedicated to making scrap. Using real-time production data to eliminate scrap is the equivalent of buying a new machine.

Working as a Data Jockey is a hidden factory in our offices. Like a machine producing scrap, it is not value added. Eliminating the data shuffle - freeing the data jockey - pays huge benefits to your organization. The case study outlines some of them.

What about you? What is your experience as a Data Jockey? What have you done to eliminate this hidden factory in your office? Use the ShareThis button below to mark this page, leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

I came up with the Data Cost / Value Matrix to help me explain what organizations can do to become more Data Driven.

On the horizontal axis we have cost of data, going from High to Low. On the vertical axis we have the value of data, going from Low to High. The Data Driven Organization lives in the upper-right quadrant, where data is inexpensive and of high value.

Unfortunately, the Data Driven Organization is in sharp contrast to most businesses. Many business leaders don’t realize that they can have it both ways: high value data at low cost. While there may be an infinite number of ways companies fall short of being data driven, our experience is that they fall into three broad camps. I’ve seen these companies, and maybe you have too.

Some gain high value from their data, but pay way too much for the knowledge. They’re in Quadrant B on the Cost / Value Matrix. This is typical of many Six Sigma companies. Six Sigma companies, for the most part, understand listening to the Voice of the Process better than many other organizations. They apply proven, disciplined techniques of project management and statistical problem solving to get to the bottom of chronic, entrenched problems. Payback from these programs is huge.

However, many Six Sigma Black Belts spend an inordinate amount of time scrubbing and massaging data in order to get something useful. We call this effort the Six Sigma Data Shuffle.

Oddly this pattern is seen as normal in the Six Sigma world. If somebody in your organization has to copy files from one folder to another, reorganize the data in some new format (convert from .csv to .xls) and then scrub the data so that all descriptive fields match (make McConnell, F. and Frank McConnell into F. McConnell), all before you copy it into MINITAB where you have to group it in into appropriate samples and manually enter specifications before you can begin to analyze the data, then you are paying too high a price for clean data.  See our white paper “Freeing Six Sigma from the Data Shuffle” for more on this topic.

Others (Quadrant C) pursue data for data’s sake. They build  elaborate data collection systems that effectively protect the user from their customer, but provide little or no additional value to the business. An example of this is an automotive supplier who told me “If the customer calls with a complaint, I’ll print out a blast of several thousand data points and email or fax it to him. He gets real quiet when he discovers I have the data.” The fact that this data is rarely used to make improvements to the process (and eliminating the customer’s complaint) doesn’t seem to concern this manager.

A third camp (Quadrant D) has the worst of both worlds: they pay a high price for data, but have almost nothing to show for their efforts. This is typical of mature organizations with a long tradition of inspecting quality into a product. These businesses may have enormous file cabinets full of hand written data sheets. Data are written on an inspection sheet and then filed away.  Getting to the data is a laborious process requiring the patience of Job and the dogged determination of Wiley Coyote.

In a future post I’ll share a quick test that you can take to determine just how Data Driven your company is. In the meantime, what’s your best quess? Which quadrant does your business live in?  Use the ShareThis button below to mark this page, leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

If you haven’t been following it, the discussion about the data shuffle has been continuing over at LInkedIn.

Laura Wright posted a comment yesterday:

Is there SOME value to the ‘data shuffle’? E.g., deep knowledge that can help the green or black belt discern nuances to their process analysis that otherwise wouldn’t be had…and so a better solution comes to light? Don’t get me wrong - I do believe data shuffle is wasteful…but some fruit can be gleaned from the exercise.

I think this is a great question, and I agree with her comment. But I also wanted to push the topic out a little further.

I was still trying to formulate a response when Terri Jostes weighed back in with a comment that said what I wanted to say far better than I could have:

I agree with Laura in that there is no substitute for an intimate knowledge of your data. Understanding where it came from, what it means and the process used to acquire the data is absolutely critical. But after that’s been figured out, a mechanism for streaming process data to managers and process improvement experts has to be put in place to free your belts from the ongoing task of “cleaning up” the data or linking files from multiple databases so it can be used.

In the interest of full disclosure I need to point out that Terri is a former user of the GainSeeker Suite. She comes to this after having lived with the data shuffle and found a different way of life. Actually some years ago I wrote up a case study about the experience of an unnamed Master Black Belt (who I just ‘outed’) at a financial services firm. Here is a link to read the case study, Building a Six Sigma Measurement System in Financial Services. At the end of the case study is a section on Lessons Learned, and the first lesson addressed this very point. Here is an excerpt:

Upstream manual data collection – According to the MBB who led the cycle-time-reduction initiative, the initial effort of capturing data manually first paid huge dividends as the deployment progressed. By engaging in manual data collection, the MBB was able to gain valuable insight into the nuances of the various operational definitions used by the process owners, and in the way the information system supported or did not support those definitions.

While an automated system has proved invaluable for collecting and analyzing massive amounts of transactional data, it is essential to develop an intimate, hands-on relationship with data in order to understand the system that produced it. This principle applies to any initiative or project that is focused on deriving long-term, leveraged benefit from an automated measurement system.

This lesson was reinforced later when the MBB implemented a similar measurement system in another part of the business. In this second application, she believed she knew enough about the system to go straight to automated data collection, but she discovered that there was no shortcut to forming a thorough understanding of the data by collecting it manually first. The second application took far longer to deploy, with many more false starts before realizing success.

Does this sound familiar to you? Use the ShareThis button below to mark this page, or leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

Over at the LinkedIn Continuous Improvement, Six Sigma & Lean Group, I posted a discussion question that generated some great responses. Here was my original question:

Lean Six Sigma practitioner’s ad hoc data survey
Recently I was talking with a black belt who told me “I know exactly what you mean by the data shuffle. Three hours of my 13 hour day yesterday were spent doing the shuffle.”

The data shuffle is all the compiling, massaging, and manipulating data that you do to get something useful so that you can make meaningful business decisions.

My experience is that all that busy work undermines the cultural transformation that is needed in Lean Six Sigma deployments. In other words, if data were easier to get to, the corporate culture would be more apt to make data driven decisions.

I have a couple of blog posts on the topic.

For an example of the data shuffle, see http://www.hertzler.com/blog/dataheads/index.php/2008/12/doing-the-data-shuffle/

For more background on the relationship between data and cultural transformation, see http://www.hertzler.com/blog/dataheads/index.php/2009/03/hitting-my-forehead-with-the-palm-of-my-hand/

Is the data shuffle alive and well at your organization?
How much time do you spend doing the data shuffle for your projects?
Does the data shuffle undermine cultural transformation?

Here are excerpts from some of the responses:

From Shaun Wurzner:

Regrettably, [Six Sigma] arrived at a point where managers and business leaders had to demonstrate Six Sigma improvements and for fear of at best, never being promoted or at worst, being let go…. Incorporating the fear of employment with continuous improvement was in my opinion, a disastrous train wreck. As a result, managers (and I AM GENERALIZING) felt compelled to generate data to show improvement. Regrettably, the more paperwork and reports, the better chance of being viewed as a valuable results oriented contributor…

To tie back into Evan’s original question, another fundamental requirement of Six Sigma was the ability to access data to characterize process, identify sources of variation, and drive process improvement through various means… The ability to capture timely data continues to be one of the fundamental issues that challenges companies and practitioners alike…. I can tell you that so many companies drive their business without fundamental data collection and analyze tools. As an example, how many times have you seen someone apply an algorithm to a data set without checking it for normality and these are statisticians or quality “experts”. Not to fault them, but everyone is so preoccupied with the end results that they by-pass some of the fundamentals.

This is a powerful indictment: “Managers feel compelled to generate data to show improvement.” Wow.

But why should we be surprised? It is entirely consistent with Deming’s 14 Points, especially #8 (Drive out Fear.), #10 (Eliminate slogans, exhortations, and targets for the work force.), and #11 (Eliminate numerical quotas for the work force and numerical goals for management.)

Maybe it is time to dust off your 1982 edition of Deming’s “Out of the Crisis”. We did not learn those lessons well.

And from Terri Jostes:

The 7 types of waste are now known as the “7 +1″ or 8 types of waste with the addition of “Creativity” or “Human Potential” as the 8th type of waste. All the wasted effort of overprocessing, rework and motion (in the form of moving data from one database to another) associated with the “data shuffle” causes our intelligent, highly trained, well-paid green belts and black belts to spend countless hours in unproductive, frustrating activity.

Why do companies allow this to happen? In some companies, it is seen as a rite of passage for a belt or a necessary, if unsavory, part of the job. Some belts even like doing this type of work - they’re good at it! Of course they are…they spend a lot of time doing it!

Evan’s right…It’s time to be strategic about our data gathering, compilation and analysis. In order to create a “data driven” culture, clean, reliable data has to be readily available to all levels of the organization. Letting the “data shuffle” continue in our organizations guarantees lost opportunity in our improvement activities and operational performance.

I love that idea: “Rite of passage.” Did you have to go through this rite? What is your story?

And from Forrest Breyfogle:

Evan, you indicated that the data shuffle is all the compiling, massaging, and manipulating data that you do to get something useful so that you can make “meaningful business decisions.” If the result was to make meaning business decisions that would be one thing and would be waste in that the analysis was not performed efficiently. However, it appears to me that the problem is worse in that often there is playing games with the numbers to make a situation appear better than it is… We need to blend analytics usage within an overall business governance system.

Excellent point: This isn’t just an efficiency issue; we need better governance. Can better data help that?

And from Terry Burton:

Today’s turbulent economy requires much more targeted Lean and Six Sigma efforts that produce an accelerated level of tangible results. That means a rapid and perfected “value-added” execution. We need to stop the data shuffle, and the training of the masses and put the money on the table - Or it’s all just another bandwagon! Too many organizations are stuck on this”mad belt” and “data shuffle” mode with their Lean and Six Sigma deployments.

One of the largest inefficiences we observe that compounds the data shuffle is that people are using multiple versions of the truth (facts). This is prevalent in organizations where the formal enterprise system has broken down, or where people struggle with their own personalized kluge spreadsheets to get (data shuffle) the facts. Some people tend to grab and shuffle the data that is available, rather than think through the specific data elements needed to solve the problem.

Think about the obvious. Not only are these folks inefficient due to data shuffling, but they all have different versions of the facts when they’re done! Now this leads to the wrong actions and firefighting, and the data shuffle becomes a viscious cycle. Their organizations would be better off if they made their “data shufflers” sit there and do nothing! Unfortunately, these organizations are not improving at the rate of the economy so despite all the belts, they are falling behind.

The largest challenge with Lean and Six Sigma is quickly thinking through and acquiring the right data to make the right decisions and get the right results. Some of our clients have reached this “utopia:” Real-time, visual event-driven metrics, a single version of the facts, decide-act-measure in real time. Based on true potential, we see that well over 80% of Lean and Six Sigma deployments are failing. People are hanging up the window dressing and getting their belts, there’s a lot of data shuffling and charts, but cultural transformation is not happening.

This economic meltdown is a great opportunity to rethink the “what’s” and “how’s” of your Lean and Six Sigma deployments and shift into a higher gear of new results.

Important concept: “A single version of the truth.” This reminds me of a conversation I had with a test engineer at an audio amplifier company who told me “In the past, if I wanted that data, I had to go mine it myself… People would be reasonably questioning my political motives for saying ‘We had 94% first pass yield last week. Now we have this standard called Hertzler, and you can go get the same data I just got…”

And from David Back:

I can only echo the points made and pass on some of my experience with management teams. Management by facts is a major change for many leaders and it is far more than setting up data cemeteries and then having specialist analysts or Belts providing insight in parallel to routine business processes. Managers need to buy in to the strategic nature of the change and be equipped with the clear concepts to act in this way. The size of this step is generally underestimated, old habits die hard ! It falls to all of us that understand this issue to promote the change at every possible opportunity.

“Data cemeteries…” What a great term.

And from Ken Place:

Although I agree that “the data shuffle” is a problem, I see more instances where data is just not available, what is available cannot be used, and new or additional data collection activity is required before an effective Lean or Six Sigma effort can/should even begin. As a Master Black Belt, many of my Black Belt students come to the first week of class ready to hit the ground running with a project. When we discuss the details of “Good” data they realize that they must return to square one and generate some history of the current state before progressing. Rather than massage the data to make it useful I find they must disregard current data and re-collect. Clearly non-value added but seldom fixed at the root of the problem, The Business Management System (BMS). In my opinion the BMS, formerly known as the QMS, should be designed with the correct, robust metrics in place for each process. Those metrics should reflect the health of each process as it contributes to the system, so that a company is able to asses where their next opportunity for continuous improvement exists. Without ongoing correct measures of at least efficiency and effectiveness, projects are not properly prioritized and work cannot begin right away. Yes metrics should be strategically established and available at any time for accurate monitoring, measuring and continuous improvement efforts.

At the risk of over-simplyfying, build the measurement system and the the rest will be easier.

What do you think? Is the data shuffle alive and well at your organization? How much time do you spend doing the data shuffle for your projects? Does the data shuffle undermine cultural transformation? You can leave a comment, tweet me, schedule a conversation, or call 800-958-2709.