Skip to content

Web Analytics – Why The Math Always Wins

I didn’t know it then. Working for M&M/Mars as a newly minted Sales Information Analyst struggling to create a regression analysis by hand the future was visiting Mars. I was asked to determine why Royals, a mint M&M’s, wasn’t getting traction in its test market in Buffalo, New York. I’d identified dependent and independent variables, I’d thrown everything in to Lotus (and I know how badly that dates me J) and was working the numbers HARD.

It was my first big test. Having created a crude territory management system on a very expensive and not very powerful PC purchased with a loan from Marine Midland bank I’d been promoted to do something similar across the world of M&M’s. Holding one after another variable constant I watched the regression of the now dependant variables such as sales, profit per shelf facing or distribution point. Nothing made sense. I was lost in the data.

I’d been selected to be an M&M’s taste tester and trust me this sounds much more fun than it actually is since sitting in a room for hours eating and then spitting out the new Twix turned out to be painful. Coming back to my noisy desk in M&M/Mars large open national office, no one has an office at Mars, Incorporated, I decided to introduce taste panel data into my analysis.

I was young, stupid and oblivious to the dangers of mixing church and state. Introducing and then regressing the taste panel the conclusion was clear. Royals wasn’t very good, or more accurately a little Royals went a long way. I was asked to determine why 16 ounces of Royales were sitting on shelves. Our sales force was the biggest buyer of Royales. When the product hit its freshness date we would issue a credit and destroy the product so consumers didn’t eat stale Royales.

Consumers weren’t eating Royales because a few after dinner was great, a 16-ounce bag was death by mint M&M’s. Creating a five-page report I called a meeting (first mistake). Expecting applause, accolades and a big biscuit I finished my report, looked up (this is well before computer aided presentations) to see angry, cold silence. No one spoke for what seemed an eternity. Sometimes you know instantly you’ve been a fool despite your ability to predict it before hand and this was one of those moments.

The two most senior managers left the room without a word. Alone with my boss and mentors Charlie Purdy and Wolfgang Pfeiffer I learned of my faux pas. You didn’t tell senior Mars Brand Managers their products were so significantly flawed. It simply wasn’t done. Charlie and Wolfgang would help me understand the ropes well enough to climb them instead of hang myself over the next few years. The future couldn’t be turned back even by uber-powerful, freshly minted Wharton MBA Brand Managers.

Royales died an unmerciful death and some tough conversations took place well above my pay grade. I’d just experienced my first example of an important internet marketing rule (even though there wasn’t a Internet then) – the math always wins. Will share more contemporary examples of why the math always wins soon. In the meantime, if you have any Royales please throw them out as they are well past their freshness date :).


Director of Marketing

The Atlantic BT Manifesto

The Ultimate Guide To Planning A Complex Web Project