You are working hard to provide your customers with relevant information. The response rate of your marketing activities is largely positive and the ROI meets the expectations of the management. So, everything’s fine? No, not necessarily. Presumably, you are giving away valuable potential in customer communication. Have you ever thought about the fact that some customers might be turned away by a marketing measure? Or that customers and prospects must be incentivized differently to make them perform the desire action?
This is where uplift modeling comes into the game, a method for modelling consumer behaviors that predicts the net impact a treatment such as a marketing measure has on individual behavior. Put simply: What impact does a marketing measure have on a customer compared to no communication? Uplift Modeling has proven especially useful for customer retention, mailings, cross- and upselling as well as discount campaigns.
Predictive analytics vs. uplift modeling
Traditional predictive methods may well be able to predict the response of a customer on a marketing activity, however they don’t answer the question how a customer would have behaved without the treatment. Possibly, a customer would have purchased the advertised product anyway and your marketing expenses were a waste of time and money, at least for this individual customer. Uplift modeling tells you which customers you should contact and which you should not because otherwise they may turn away. After all, you don’t want to wake sleeping dogs.
Sinking campaign costs and rising ROI
The main advantage that results from uplift modeling is that you contact only those customers who want to be contacted thus saving contact costs. This, in return, decreases your campaign cost and boosts your ROI – which the finance department will be more than happy about. For discount campaigns, uplift modeling has also great potential, predicting for example what amount of discount a customer needs to purchase your product. Instead of giving each customer a discount of, let’s say, 30% (for some by far too little, for others more than they really need), you can adjust the amount of discount individually to your customers, thus not offering more than necessary. Of course, this will also make a significant impact on your campaign cost.
Not everyone wants to be approached
Using predictive scores, uplift modeling creates four customer segments:
Customers who perform the desired action only when you address them (Persuadables)
Customers who always buy, independently of your marketing measures (Sure Thing)
Customers who never buy no matter if you contact them or not (Lost Causes)
Customers who won’t make a purchase when you contact them (Do-Not-Disturbs)
Until now, you may have addressed these different customer segments the same way. However, you will have noticed that only approaching the Persuabdables really makes sense. Contacting the Sure Things or the Lost Causes has no impact at all. In any case, you should exclude Do-Not-Disturbs from any marketing activity because otherwise you might lose them.
By applying uplift modeling, you can benefit in multiple ways:
You approach only those who want to be contacted, thus saving contact costs and increasing your ROI.
The conversion rate of your campaign increases.
Approaching only Persuadables increases the share of customers that develop brand loyalty.
Save rebate campaign costs by offering the customers only as much discount as necessary to make a purchase.
We help you to organize and use your customer as well as business data for a smart business and happy customers.