Retention marketing campaigns the predictive way

Aug, 15 2013

Recently I wrote a blog about some of the best retention marketing campaigns. All of these campaigns are customer event driven, which makes them so popular with customers. You can really surprise and delight your customers with any and all of these campaigns.

The table below summarizes the incremental conversion you can get from using predictions in your campaigns.

Screen Shot 2013-08-15 at 11.07.29 AM

In this blog, I want to elaborate on how you can use predictive analytics algorithms to make some of these tried and true campaigns even better and achieve these advantages.

Predictive abandoned cart (or search)

The idea of a predictive abandoned cart (or search) program is that you don’t just remind people of their abandoned cart (or abandoned search) items; but that you use individually sized incentives like free shipping, a free sample, bonus points or discounts to get customers to come back and complete their transaction. You only use incentives, if the predicted lifetime value of the customer is high, and/or if their likelihood to purchase is low (in which case you basically have nothing to lose and might as well give away an incentive).

Ideeli abandoned cart campaign

Predictive new customer program

A new customer program is for customers placing their first order, not to be confused with customers signing up for your email list for the first time (though you should do something special for them too). In the case of a predictive new customer program, you don’t just welcome and thank your new customers for their business, you make individualized next best product recommendations using predictive algorithms (30% more accurate). Moreover, for customers with a predicted lifetime value that is high you add an individually sized incentive. This can be, but doesn’t have to be, a discount. Customers are delighted with free gifts such as a free sample. It’s especially important to have your high value customers come back a second time. Two time buyers are significantly more likely to remain loyal than one time buyers so it’s super important to close that second transaction quickly.

Predictive cross sell program

Hyper-personalized recommendations that make use of modern, predictive algorithms (rather than more simple rules based recommendations) are 30% more successful (on average, sometimes a lot more than that) in soliciting customer engagement.  Predictions also allow you to tie an individually sized incentive such as free shipping, a free sample, bonus points or a discount to the cross sell offer. Don’t apply incentives across the board: for customers with a low predicted lifetime value it could be a money-losing proposition but for high value customers it’s a smart investment. Also, be consistent with your recommendation across all channels including email, web, Facebook and phone.

Predictive replenishment reminders

It’s great to send replenishment reminders, but it would be even better to predict exactly when you should send this reminder for a specific customers. In addition, or alternatively, you can increase replenishment response rates by adding an individually sized incentive like free shipping, a free sample, bonus points or a small discount, to the reminder. You will see the trend, but we recommend you don’t use incentives for everybody. Increase your profits by using incentives only for those who have a high predicted lifetime value or a low likelihood to buy. The accuracy of predictive algorithms really shines when it comes to offers. Alternate rules might be 30% less accurate, which could mean sending 30% more discounts to people who do not really deserve them.

Win back campaigns the predictive marketing way

Don’t just tell customers who haven’t bought for a while that you miss them; add a personalized product recommendation using predictive algorithms (remember: 30% more accurate). Plus, if predicted lifetime value is high (they deserve it) or likelihood to purchase is low (you have nothing to lose), then add an individually sized incentive.

Predictive appreciation program

Last but not least, let’s look at how predictions can improve your customer appreciation programs. It is surprising how few companies actually have a customer appreciation program. I am not speaking about a loyalty program here, but about identifying your top 10% of customers and surprising them with a gift, invitation, free sample or some other privilege. Sometimes just recognizing a customer as part of your elite can have a big impact. Predictions can help here because now you cannot just identify your most valuable customers based on past purchases, but you can also predict who could become your most valuable customers (if you treat them right). Now send these customers in individually sized gift or reward based on the individual (not segment) value.