Tom Webster, writing and speaking

Marketing Metrics And Head Injuries

Added on by Tom Webster.

NewImageThe two aren't as unrelated as they might seem. Consider the following scenario: You run an e-commerce site that sells real or virtual products with some kind of online shopping cart system. You make a tweak to your sales process--maybe add a step, or make your offer page longer--and observe that your percentage of "abandoned carts" (when people add your product to their cart and then, after seemingly reconsidering, exit your site without making a purchase) increases considerably. Did the changes you made to your sales process have a negative effect?

Well, there are three answers to this. The obvious answer is "yes." The obvious answer for contrarians who always suspect trick questions is "no."

The answer that would get you hired at Edison is…you don't know (yet.)

So why wouldn't an increase in abandoned shopping carts automatically be bad? The short answer is that abandoned shopping carts, in a vacuum, is the wrong metric to track.

To explain, here's a little nugget of statistical lore from World War I. In the early days of the war, soldiers were sent out with cloth hats--about as bullet-proof as my Mets hat. By 1915, the mounting, horrific number of deaths caused by head wounds prompted the armies of the Great War to begin issuing metal helmets to their soldiers. The Brodie Helmet (pictured), developed in 1915, represented a real breakthrough in personnel armor, as each helmet could be stamped from a single piece of steel, making these helmets efficient to mass produce.

As more and more soldiers became equipped with helmets, a curious statistic was observed: the number of head injuries that army hospitals had to treat skyrocketed. That's right--head injuries increased dramatically as a result of the introduction of helmets.

The answer to this conundrum, of course, is that head injuries increased because more soldiers were surviving head shots. Tracking head injuries, in this case, is the wrong metric--or, at least, it doesn't tell the story in a vacuum. The right metric to track comes down to outcomes--are fewer soldiers dying of head shots?

So, back to shopping carts. Abandoned shopping carts (and other peripheral stats) may or may not tell the story of the data you have. For example, if cart abandonment goes up, but so do sales, then the answer might be that your changes have attracted more "non-core" consumers--more casual shoppers--but the bottom line is that you have attracted more shoppers. You cannot gauge a metric like abandoned carts negatively or positively unless you consider your outcomes.

Now let's be careful out there.