The Faster He Mows The Faster It Grows

A few years ago I met a retail analyst who told me "I feel like Ali Sard from that Dr. Seuss story." I had to head home and dig through my son's collection of Dr. Seuss books before I understood the analogy.

Ali Sard is a character from the book "Did I ever tell you how lucky you are" . The book is about horrible jobs, and Ali's is to mow his uncle's yard. The problem is that the yard is so huge that by the time Ali is done, the grass has grown so much that he needs start over, right away. As the story goes, "the faster he mows, the faster it grows." This is exactly what this retail analyst is going through every day of his work life.

He suffers because his bosses (and many other retail executives) look at analytics the same way that Ali's uncle looks at his lawn in this children's story. They are trying to mow acres of grass with faster and bigger mowers. But as the speed and power of our computers increases, so does the desire to look at more data, in greater detail. Retailers used to create monthly sales reports, then weekly, and now they are daily! Yet sales growth still comes in at two to three per cent. All of this noise is hurting, not helping.

Many retailers have large teams of analysts creating detailed sales reports that keep the CEO and executive team up to speed on everything that goes on in every store across the country. An analyst's job boils down to answering one key question with precision: how did we do against plan last week?

Sure enough, after every Monday sales review meeting, there is a question. For example, why were we down in the West? And the analysts have to drill down into the available data. Unfortunately they don't have easy access to industry data which would explain almost all of the week-to-week noise. At Fusion, we have recently completed an extensive study of eight major retailers in North America and found that on average 96% of their weekly sales variances were due to either industry changes or shifts in the retailer's promotional calendar from one week to another. Only 4% were due to anything that could be considered insight. These retailers are being completely blinded by data.

Worse yet, it is a dirty little secret among analysts that changing the pre-period in a formula by just one week - let's say from a six week pre-period to seven weeks - allows them to swing the results by one to two per cent. Most programs are lucky to lift sales by two to three per cent, but the noise in most analytical models, even under test and control, is more than three per cent! This makes the whole measurement exercise unusable at best, and leads to the wrong decisions at worst. Most of the time it is just multi-million dollar retail voodoo, or what derivatives-trader-turned-essayist Nassim Taleb would call being "fooled by randomness."

Even if the analyst does manage to filter out the 96% noise, they are only doing analysis with one dataset: sales, transactions and unit analysis. Great surgeons don't use just one kind of surgical tool, and great retail analysis is the same; it requires access to a complete set of analytical tools. However, for most retailers each department only gets one tool, and for the weekly Monday meeting the only tool allowed is sales analysis. So the analysts try their best to uncover the story by looking at the available point-of-sale data, diving deeper into sub-categories and geographical pockets. Just when they think they have found the answer, the topic of the week is over and a new week begins with a new set of reports and questions.

If the analysts can't keep up with demand, retail executives hire more analysts or approve better software. As a result, the teams of analysts get bigger which leads to more reports which leads to more questions. Around and around they go, mowing more and more grass. Too often retail decision makers pursue non-critical information - information that appears relevant but isn't. That's why today's retail CEOs receive more and more data and invest more and more in analytics. It also explains why their decisions don't get any smarter. Retailers need to stop trying to do things better and faster, and start to do things differently.

The solution is to move away from cult of tracking and daily/weekly performance through a fog of noise, and instead embrace "Opportunity Modelling" which seeks out clarity on where the largest investment opportunities are for the company and aggressively align resources to them.

There is little planning action to be taken from tracking. Retail executives already intuitively know the data they are seeing is not working, but only a few have started to rethink the whole approach, that it isn't about more data, but about moving from tracking to opportunity. Then, and only then, can you track to measure how well you executed on the plan. Tracking data in of itself, be it sales data or consumer insight data, can never tell you what to do.

Opportunity modeling means we don't look at historical performance, we look at future opportunity. Let's use measure a TV brand campaign as an example. How can a retailer know what the results of a TV branding campaign should look like if they have never done one? How can they know if a brand campaign focusing on Category A is better than the campaign they never ran on Category B? There is no historical tracking data for most of the questions that matter in retail. This is why tracking leads to incremental improvement on what you are already doing best; it keeps you at the industry growth rate of 3%, but doesn't create transformational growth at 10%. Tracking simply does not identify insight that will have a material impact on your business.

Nate Silver, the famous American political statistician, likes to use the analogy of poker. You can actually play the best strategy given your cards but lose the round, and you can play the wrong strategy given your hand but win the round. The short-term can lie to you, as it is full of noise. A TV campaign might have worked or not worked for dozens of reasons and noise, but if you really want to drive your business forward you shouldn't care about the short-term noise, you want to be focused on the right opportunity. This means aligning to the safest bet, which in TV is focused on categories with large market size, high margins, high brand upside and good flow-through from visits to conversion. Consistently doing this will win you the game.

By moving analysis from mowing more backward looking data ("how did we do?") to a forward-looking formula ("where is the opportunity?"), you will enable the right conversation, and insight.