Predictive Modeling of Customer Behavior - Drilling Down in
The likelihood of a customer to repeat an action (purchase, visit, game play)
relative to other customers can be predicted using a customer LifeCycle metric
known as Recency. Recency is defined by
the number of days or weeks since the customer has performed the action
(purchase, visit, etc.) you are profiling. The more Recently a customer
has engaged in an action, the more likely they are to repeat the action,
especially when encouraged to repeat by
some kind of promotional effort.
Above is a graph of customer Recency (time elapsed
since last activity) based on
last visit (or page view) date. The number of unique visitors is on
the left (y-axis) scale; the number of days since last visit (or page view)
on the bottom (x-axis) scale. This site has about 5 million unique
visitors a month. "Yesterday" is at the far left of the chart
– 1 day ago; the right of the chart is 90 days ago.
Customers are plotted by the number of days since
their last visit (page view). For example, all the customers
who last visited
9 days ago are counted and the total is represented on
the graphed line above the "9" at the bottom left of this
graph. Look at it for a minute. What does this data speak to you?
If your answer is these guys have a great business,
you’re right. I mean, they have 5 million monthly uniques TOTAL, and
most of them visited YESTERDAY. Not only that, but virtually ALL of them
have visited in the last 10 days! This is a smokin’ business.
wouldn’t know that without looking at Recency, would you? You have
to know this stuff. It means something very important, not only to
marketing, but also to the value of your business in the future.
take a look at this second graph below.
Once again, a graph of customer Recency, based on last visit
(page view) date.
Unique visitors are on the left (y-axis), and days since last
visit (page view) on the bottom.
Yesterday (1 day ago) on the far left, 90 days ago on the far
This is a smaller site. They launched with about 60,000 uniques
90 days ago, but only about 3000 a day come back now.
What’s the future of this business?
What’s the data say to you about it?
It’s toast, man.
Absolutely moribund. They’ll
be lucky to keep the visitors they have. They
need to make major changes, and I hope they have the e-mail addresses of those
people who never came back so they can promote their new offering.
If they don’t, there’s not much hope.
These two businesses didn’t just magically get
to where they are overnight; there was a process; customer behavior
changed over time. Wouldn’t it be
great if there was a way you could track this process, a way to know the
direction your business is headed? There is, and you can learn about this
method and many others for tracking and managing customer value in the Drilling