In any concern that deals with marketing, the term ‘lift’ should be the Holy Grail. What is ‘lift’?
You’re going to do direct mail for a car dealership. You randomly select 14 postal routes around your dealership which total about 10,000 households. If you are able to moniter these 10,000 households over a period of, say, 6 weeks, to see how many of them bought automobiles similar to what you are selling, you have the ‘control’ number. Now do a list created with some form of targeting involved in its creation and come up with 10,000 households on this list. The ‘lift’ (if there is one) is the UPWARDS PERCENTAGE of buyers on the ‘targeted’ list over the same period of time for the same products.
Lack of ‘lift’ has caused many large data companies to eventually scrap many of their ‘in-the-market’ models that they spent years and money creating.
Quantifying “lift’ requires data sources that can verify a purchase having taken place that matches the manifest (marketing list). There are two ways to look at this. Overall lift based on the purchasing of a consumer and lift based on that consumer purchasing from the entity (dealer and marketer) that produced the campaign.
Both are important. Quantifying the higher percentage of buyers in a market purchasing a product similar to what the dealer sells, let’s us know if the list is good. Quantifying the number of buyers that bought from the dealer let’s us know two things: Was it good for the dealer and did the dealer perform well in responding to the inquiries generated by the marketing list.
Good ‘lift’ has been more attributed to ‘behavioral’ (hand raisers) marketing than marketing based on ‘models’ in recent years. Some of the stories relating to this phenomenon might surprise you. One of these stories involves the largest data compiler company in the world. Would be happy to share it with you if you are interested.