[Dividing Line Image]

Formulating Using Design of Experiments II




Dom Ruggeri

October 2008:

As I have said many times, in the early Nineties I was working for a major formulating house.  Since this company was considered a large account to any raw material supplier we had salesman constantly seeking an audience with any formulator who would give them the time of day.  As expected every chemist had their favorite salesman - one that they always made time for.  Occasionally a company would bring in the top marketing people and invite the entire lab staff to a seminar.

A major corrosion preventative supplier scheduled a presentation to discuss their latest and greatest product.  These were always good days - not only did you get off the bench for about an hour to listen and learn, with any luck you would even get a midmorning snack.  Of course everyone signed up to go, everyone but me.  I knew these guys.  Further I was sure this meeting would be nothing more then a chest thumping, hard hitting bag of hot air neatly packaged to sound impressive.  The Technical Manager noticed I did not sign up and wanted to know why.  I shared my opinion about this group of presenters with him and his answer shocked me.  He felt the reasons I did not want to go were the exact reasons I should be there - as he put it someone had to "sift through the goop".

I attended the meeting and, as I expected, the material they were touting was the greatest thing since sliced bread and all their data proved it - or did it?  Each formulation, no matter how basic, improved as soon as you added their product.  Why, the corrosion threshold dropped from five percent to less then one percent.  I began to notice everyone in the room nodding in agreement with every claim these marketers spoke.  Once again I looked at their sample formulations and it became clear to me - all they were showing were the ingredients without showing how much of each component was in the formulation.  I was stunned no one else saw this!

When the presentation was over and the applause ended, they took questions.  There were none.  All right, my colleagues were on to this scam... or not.  I looked around and many were still nodding their agreement with the presented data.  I raised my hand and asked how much of their corrosion inhibiter was in each of the formulations shown.  The lead presenter replied, “That number is not up there”?  His answer was "the standard amount", less then helpful at best.  I pressed on.  I wanted to know what standard amount meant.  Finally, after much useless dialogue, he told me the standard amount they used was twenty-five percent.  At eight dollars a pound, no one could afford to use this material.

In August 2006, I had the privilege of attending a course on data presentation and analysis given by Edward Trufte.  What does this have to do with a DOE?  In a DOE you will generate reams of data.  That data will help you formulate a product or solve a problem.  Either way, your data is the evidence you need to back up and sell your solution.  Here are a few tips I learned at the seminar:

1. Document everything and report it:
Whenever you are reporting data your credibility is on the line.  Therefore, you must show your audience that your data is accurate and complete thus enhancing your credibility.  This leads us to the second tip.

2. Evidence or Evidence Selection (Cherry Picking):
Many marketing types will engage in this practice selecting only the data points that agree with their solution or make their product look good.  If you want a hit in the credibility, try cherry picking data points.  Someone, or everyone, in the room will catch you.

3. Your data must answer these questions:
a. What is the problem?
b. Why is it important?
c. What is your solution?

If your data presentation cannot answer the three questions above you have a very big problem.

Packaging is everything.  This is true for data presentation and report writing especially.  Many companies want charts and graphs attached to the back of the report.  Avoid this practice whenever possible.  When a reader has to flip to the back of a report to see a graph or data chart you have already lost their interest.  Place charts and graphs in the body of reports, with detailed explanations under the data so that the reader can easily understand how the data is presented and what it means.

What I learned at that seminar could fill perhaps twenty articles, but that would be quite boring.  However, if Edward Trufte is offering a seminar in your area, I highly recommend you consider attending.  Until next time, should you have any questions please feel free to e-mail me in care of the magazine.  I hope to see all of my readers in Philadelphia.

Good Luck