September 20, 2015

Channels Seldom Offer Multiplicative Gains

To date, there are four components to my system:
  1. Annual Repurchase Rates dictate the strategy a marketer needs to employ - and in most cases, that strategy is to find a ton of new customers.
  2. If we have to find a ton of new customers, and most of us have to find a ton of new customers, we need healthy levels of merchandise productivity to get the job done.
  3. If we have poor merchandise productivity, we optimize via profit to get the most out of a bad situation. If we have good merchandise productivity, we optimize via profit to obtain healthy sales and profit growth.
  4. It is very difficult to sustain long-term annual repurchase rate increases, new customer increases, or profit optimization without healthy new item development. Therefore, new item development is a critical component of business health.
The fifth component is this one ... "channels seldom offer multiplicative gains".

Oh, I know, this one goes against everything you've been taught (#multichannel #omnichannel).

Before there were channels, there were "battlin' business units", as I used to call them at Lands' End, some twenty-five years ago. There was a good customer - and this customer would be pummeled by each business unit. Instead of being mailed 12 catalogs a year, this customer received a monthly Lands' End catalog, and 9 Home catalogs, and 9 Kids catalogs, and 6 Mens Tailored catalogs, and 5 Womens Tailored catalogs ... that's 41 catalogs. That's a lot of catalogs.

And yet, something was wrong. If I measured annual repurchase rates over time, the proliferation of catalogs from 12 to 41 did not correspond with a marked increase in annual repurchase rates - and corporate-wide merchandise productivity barely increased. The small increase in sales aligned with essentially no increase in profit, yielding a diluted EBITDA. When your bonus is based on EBITDA percentage, you decide to dig deeper into the mess.

Turns out that these catalogs were competing against each other. If we mailed a Mens Tailored catalog, we learned (via testing - more on testing coming soon) that half of the sales were cannibalized from other channels. In other words, if a Mens Tailored catalog generated $5,000,000 in sales, $2,500,000 were incremental sales, and $2,500,000 would have happened anyway had the catalog never been mailed.

#OhOh.

Then I get to Eddie Bauer ... I was asked to measure what happened when we opened new stores in new markets, and what happened when we opened new stores in existing markets. The results were not multiplicative.
  1. A new store in a new market resulted in the decrease of catalog/online sales for a period of time, until the new store reached its potential.
  2. A new store in an existing market resulted in the decrease of sales in the stores that already existed, and resulted in a decrease in sales in catalog/online channels, until the new store reached its potential.
  3. The new store in the new market maybe generated $1.6 million in sales ... of which, $1.1 million was new business, with $0.5 million cannibalized from existing channels/stores.
In other words, what I observed in retail at Eddie Bauer aligned with what I observed in old-school cataloging at Lands' End.

Then, I get to Nordstrom. We noticed something similar to what was noticed at Eddie Bauer. - open stores, and sales shift around a bit (though not as much as at Eddie Bauer). Even better, we shut down $36,000,000 in catalog advertising. When we shut the catalog down, the trade journalists called me ... "are you an idiot?" They printed their nasty-grams for all to see. Those were fun calls and fun articles to read. Nobody called a year later when sales INCREASED and profit INCREASED and we maxed-out bonus payments, the trade journalists conveniently ignored the positives. Turns out sales shift. Not everything is multiplicative. In fact, it is extremely rare to find anything that is multiplicative - you need network effects to make that happen - and you need immense word-of-mouth to make that happen - both are nearly impossible.

What you learn when you actually write your own code to measure your own results is that channels are interactive, they are not multiplicative. Gains in one channel come at the expense of sales in another channel.

What you learn when you read trade journals and listen to consultants / researchers is that channels are multiplicative, and if you don't expand into every channel and you do not thoroughly digitize your business from top to bottom, you will go out of business.

Ask yourself why retailers cannot grow at a time when they are following the vendor playbook of adding hundreds of digital channels? Look at Macy's, for instance. 

These facts became a key part of my system - maybe the most controversial part of my system. Most do not believe in this part of my system. They'll tell you that I am "measuring engagement incorrectly". Fine. Engagement has nothing to do with profit. I'll measure profit instead. When you measure profit, you learn channels are not multiplicative. And that's ok.

There are now five components of my system:
  1. Annual Repurchase Rates dictate the strategy a marketer needs to employ - and in most cases, that strategy is to find a ton of new customers.
  2. If we have to find a ton of new customers, and most of us have to find a ton of new customers, we need healthy levels of merchandise productivity to get the job done.
  3. If we have poor merchandise productivity, we optimize via profit to get the most out of a bad situation. If we have good merchandise productivity, we optimize via profit to obtain healthy sales and profit growth.
  4. It is very difficult to sustain long-term annual repurchase rate increases, new customer increases, or profit optimization without healthy new item development. Therefore, new item development is a critical component of business health.
  5. Channels seldom offer multiplicative gains. Channels interact with each other, causing customers to make tradeoffs instead of spending more money.
Two more components follow. You can hardly wait, can you?