BY CURT BERGMANN – SR. DATA SCIENTIST
There is a lot of insight hidden in your data. With a little imagination you can extract more than you might have thought about a customer’s behavior. But, if you’re not careful, you can easily draw the wrong conclusions about a customer, waste money by including them in the wrong marketing campaign, and worse, miss your revenue goals. We’ll show how some common ways of looking at loyalty points can cause problems and show that explaining them in different ways will reveal better insight.
BALANCE
Let’s take a loyalty point system where customers earn points through purchases and then redeem them for free products. At the most simple level you have a series of credits and debits resulting in today’s balance for two customers named Laura and Liam.
Hidden within this one value, of course, is the behavior that resulted in both customers having the same balance. We can instead look at the series of credits and debits:
Here it is obvious that Laura has been much more engaged than Liam since Liam has only had 1 credit of 50 points versus Laura that has 7 credits and 3 debits. If we were only to use Balance when determining how to market to loyalty customers we would miss out on a lot of information about Laura’s behavior and likely treat them the same despite their difference in engagement.
POINTS EARNED
Rather than using a single value of Balance we can then separate this into two values – the Points Earned versus the Points Redeemed. Here is the single value and a chart of how that looks for each transaction.
In this view, it is easier to see that Laura is more engaged with the loyalty program than Liam. We can see that her balance has been accumulating over time.
RECENCY
Speaking of time, we have only been showing the loyalty transactions from first to last and haven’t shown them by date. If we don’t consider the date of the transaction we can run into a problem similar to our first. Here are two customers that appear to have about the same loyalty behavior:
But instead of looking at it simply as a sequence of transactions, adding dates to the x axis gives you a much different story:
Laura hasn’t engaged with the loyalty program since last year and has possibly lapsed.
We will want to market to these two customers quite differently. In order to do so we can create several new attributes at the customer level to be used as input into our marketing list selection strategy.
Now with five different attributes from a simple series of Loyalty Points we have much more insight about a customer’s behavior and can segment these customers in order to make much better marketing decisions.
When to Dig Deeper
BY CURT BERGMANN – SR. DATA SCIENTIST
There is a lot of insight hidden in your data. With a little imagination you can extract more than you might have thought about a customer’s behavior. But, if you’re not careful, you can easily draw the wrong conclusions about a customer, waste money by including them in the wrong marketing campaign, and worse, miss your revenue goals. We’ll show how some common ways of looking at loyalty points can cause problems and show that explaining them in different ways will reveal better insight.
BALANCE
Let’s take a loyalty point system where customers earn points through purchases and then redeem them for free products. At the most simple level you have a series of credits and debits resulting in today’s balance for two customers named Laura and Liam.
Hidden within this one value, of course, is the behavior that resulted in both customers having the same balance. We can instead look at the series of credits and debits:
Here it is obvious that Laura has been much more engaged than Liam since Liam has only had 1 credit of 50 points versus Laura that has 7 credits and 3 debits. If we were only to use Balance when determining how to market to loyalty customers we would miss out on a lot of information about Laura’s behavior and likely treat them the same despite their difference in engagement.
POINTS EARNED
Rather than using a single value of Balance we can then separate this into two values – the Points Earned versus the Points Redeemed. Here is the single value and a chart of how that looks for each transaction.
In this view, it is easier to see that Laura is more engaged with the loyalty program than Liam. We can see that her balance has been accumulating over time.
RECENCY
Speaking of time, we have only been showing the loyalty transactions from first to last and haven’t shown them by date. If we don’t consider the date of the transaction we can run into a problem similar to our first. Here are two customers that appear to have about the same loyalty behavior:
But instead of looking at it simply as a sequence of transactions, adding dates to the x axis gives you a much different story:
Laura hasn’t engaged with the loyalty program since last year and has possibly lapsed.
We will want to market to these two customers quite differently. In order to do so we can create several new attributes at the customer level to be used as input into our marketing list selection strategy.
Now with five different attributes from a simple series of Loyalty Points we have much more insight about a customer’s behavior and can segment these customers in order to make much better marketing decisions.
Tags: