In essence, the word "parsimony" means that "less is better".
And in multichannel marketing, we look to incorporate every possible piece of information we can find, don't we?
I recall back in 1991, the "good 'ole days" at Lands' End. You submitted a request to have customer information pulled from the mainframe computer system and written to tapes. Statistical models were built off of programming code written against the mainframe computing system.
My boss, a statistically minded individual who came to Lands' End via the oil industry, built simplistic statistical models that were used to mail catalogs to Lands' End customers. He used probit models (yes, this was before SPSS and SAS had procedures for Logistic Regression ... old school!) to predict response, using recency and frequency. He used ordinary least squares to predict average order value, using historical average order value.
He multiplied his predicted response rate (0.03) by his predicted average order value ($100), yielding a predicted dollar per book ($3.00). He ranked ten million households from most productive to least productive. Our circulation managers decided who to mail based on the predicted values he generated.
He built glorified RFM models, for crying out loud!
Oh, was I excited to get my hands on this model!
At all of 26 years old, I strongly believed that complex models with many variables and many interactions would do so much better.
So I built the models, using every possible level of complexity that a mainframe computing system would tolerate.
After one year of testing (old model vs. new model, actual mail/control tests in every catalog), my models generated exactly 1% more demand than the old models generated.
Now there's nothing to sneeze at when you're talking about 1% of $500,000,000 ... the $5,000,000 of demand and $1,500,000 of profit I contributed to the company on an annual basis were appreciated by shareholders, no doubt.
Parsimony should be a big part of our multichannel world, but it isn't.
We make multichannel marketing so much harder than it needs to be.
In social media, we have a blog, a Facebook presence, a Twitter presence, Plurk, instant messaging, a cell phone, a business phone, multiple e-mail accounts, Friendfeed, video sites, on and on and on. We believe we have to be in twenty different places. And our most loyal followers join us on these channels ... THESE ARE OUR BEST FOLLOWERS!!
But that doesn't mean we need to be in all of these places. What would happen if we only participated in two or three social media channels? I'm not saying we shouldn't experiment ... I am saying we should be strategic marketers. What is the strategic marketing purpose of each social media channel?
Same with old-school multichannel marketing. Who says you have to keep sending out direct mail at a 1% response rate? Who says that every product has to be in every channel at the same price at all times?
And it is the same thing with our BI or Web Analytics or Statistical Modeling groups ... we simply don't need to make things so complicated --- and if we're going to make it complicated, don't tell anybody. Our executive partners have no idea what the complexity we're communicating means ... and that is bad for our career development.
RFM is great for targeting one catalog to one customer. However, RFM is tough to manage in a multichannel environment. This becomes clear ...
If you don't like geeky math, please skip this post, because I am about to show you how the sausage is made! I have eight variables in...
It's common for folks to measure cost per new customer. Total Marketing Cost = $10,000. Total New Customers = 130. Cost per New C...