November 15, 2010

Hillstrom's 2011 Almanac ... Sample #1

Here's a sample tidbit/observation from the new book, available soon:

May 3

If you want to measure the incremental value that Facebook delivers to your business, give regression analysis a try.
Pull purchase data for all customers who placed at least one order in the past twelve months ending March 31.  Here are the variables.
  • Recency, Months Since Last Purchase.
  • Number of Orders, Past 12 Months.
  • Number of Orders, 13+ Months Ago.
  • Number of Channels Purchased From.
  • Number of Merchandise Divisions Purchased From.
  • 1 if Customer is a Facebook Fan, 0 Otherwise.
Pull demand spent by customers in this audience during April.  Run a Regression Analysis.
  • April Demand is the Dependent Variable.
  • The Variables above are Independent Variables.
Look at the coefficient for being a Facebook Fan.
  • If is isn't significant, then being a Fan results in no incremental value.
  • If the coefficient is negative, then being a Facebook fan hurt the business.
  • If the coefficient is positive, then multiply the value of the coefficient by the number of customers who are a Facebook Fan.  This multiplication yields the incremental dollar value provided by Facebook. 
    • Example:  Coefficient = $0.50.  Customers = 1,000.  Incremental Value = $0.50 * 1,000 = $500.

6 comments:

  1. Scott2:11 PM

    How do people go about identifying facebook fans on their housefile? I'd love to be able to perform the analysis you are suggesting, but we're struggling to find a way to "export" fan info and associate to a name/address/email on the database. Suggestions?

    ReplyDelete
  2. It doesn't have to be hard. Each time a customer orders from your business, have a box on the checkout page asking if the customer is a Facebook Fan ... ask the customer to volunteer information.

    It's not perfect, but it is better than nothing, in fact, it's significantly better than nothing.

    ReplyDelete
  3. Chris7:18 AM

    I had the same question. Another way is to allow consumers to register for your site using Facebook Connect.

    ReplyDelete
  4. Don't over-think it ... do what Chris said, there's a thousand ways to capture enough information to make valid inferences.

    ReplyDelete
  5. Hi Kevin,

    First of all love the blog and have been a reader for quite sometime.

    An issue I have with this type of analysis is the old correlation versus causation argument which I know you're familiar with.

    Despite having a strong positive coefficient after running the regression, one shouldn't draw a straight line that there is any incremental value of these customers that has anything to do with Facebook.

    As an example, it stands to reason that if I'm a huge fan of Best Buy (and of course I am) and I'm also on Facebook - it's already pretty likely that I may also be a fan of that brand on Facebook.

    I'd like to think of myself as a high value customer for them with the amount I've spent over the years.

    It wouldn't make sense though for Best Buy to assume there's incremental value driven in our relationship as a result of me being a fan of their page on Facebook. If anything, it simply says, "I want people to know I like Best Buy."

    I can say with the Best Buy example, I'd be just as likely to purchase form Best Buy whether they were on Facebook or not which IMHO makes it very difficult to draw the "incremental value" conclusion that you've pointed out above.

    Best,
    Mike

    ReplyDelete
  6. The theory is that the recency/frequency/monetary information controls for the fact that you are a huge Best Buy customer, leaving any residual impact to the Facebook coefficient (and random error).

    Not perfect.

    But better than what people are out there are publishing.

    ReplyDelete

Note: Only a member of this blog may post a comment.

Dashboarding New Customers by Week

Each week, you acquire new customers. Count the number of new customers by segment. Elite = 0 Customers (0%) Loyal = 1 Customer (< 0.1%) ...