When file counts are small, we're just kind of stuck having to use what we have.
Remember, in yesterday's example, most of the mobile customers were graded as "A".
|Grade = "A"||50,000||394||2.58|
|Grade = "B"||50,000||176||1.15|
|Grade = "C"||50,000||88||0.58|
|Grade = "D"||50,000||65||0.42|
|Grade = "F"||50,000||42||0.27|
So here's what we can do. Let's take everybody at the end of March who had a grade of "A". Split that segment by those who used a mobile app recently, and those who did not.
Now, for each group, measure the incremental demand generated during the month of April. Your table should look something like this:
|Grade = A, Mobile||394||3.4%||$165.00||$5.61||17.5%||$4.63|
|Grade = A, Other||50,000||3.1%||$170.00||$5.27||0.4%||$5.25|
There are a couple of things worth noticing here. First of all, the mobile app buyers, after equalizing a bit for customer quality, are worth 6.5% more in the month of April than are other customers. That's what we want to see, we want to see that the new channel creates value among customers.
Also notice that the number is 6.5% ... it's a small number. In cases like this, don't expect mobile apps to be the life saver you've been looking for to generate huge sales increases (unless the percentage of mobile app users are disproportionately new customers).
Finally, notice that non-mobile volume is nearly 12% less among mobile app shoppers. If you see an outcome like this, then you know that your efforts are causing these customers to shift business from your core website to the mobile app ... in other words, your online sales are being cannibalized by the mobile app.
Cannibalization matters, folks. Most businesses miss cannibalization until it is too late. Just ask catalogers, just ask the newspaper industry, just ask independent record stores. When sales are cannibalized, you have strategic issues that need to be addresses, so that "channel shift" is not devastating to your business.
Ok, you've read several posts about measuring fundamental relationships in mobile marketing, what are your thoughts, what questions do you have? We've got more posts coming ... that being said, this would be a good time for input.
Is your presentation regarding conditional probability available (emetrics)?ReplyDelete
I'm not making that presentation available a this time, though I will cover much of the content on the blog during the next month.ReplyDelete