Big Oil, Big Coal, Big Data
Last week we discussed carpet bombing in the context of marketing campaigns and came to the conclusion that smart targeting is always better. This week we continue by looking at how we can make smart decisions with data. In other words, let the data decide.
Set clear targets To run successful campaigns, you need to be specific about which group of customers you are targeting and also what exactly you hope to achieve. For example, a campaign might be seeking to reduce churn amongst a student customer population by 5%. Another could be advertising your mortgage product to newlyweds with an expected take-up of $10m over a clearly defined period. “Because we ran it last year” isn’t a good enough reason to run a campaign. A campaign should always have a clear goal and a well defined target audience.
Take time to describe your customers An important aspect of marketing is defining your customer. As a start, customers are often described by criteria that look rather like this:
/Demographics - age, sex, income level, race, employment, location
/Transactions - what, where, when, how often, how much
/Behavior - interests, hobbies, likes
/Psychographics - preferences, personality types
Like a miner, pursuing a seam deeper and deeper into the earth, the quest to understand one’s customer can extend well beyond the basic attributes described above. For example, how up to date your computer’s software is, is reportedly indicative of your willingness to repay debt.
Some online lenders use this, alongside other attributes, to decide whether to grant credit or not. Likewise the proportion of names in your mobile phone address book that have both first names and last names (Jospeh Smith as opposed to Joe), is reportedly a tell-tale sign of your likelihood to miss payments or default.
As mentioned in a previous blog post, one of the foundational applications of machine learning is cluster analyses. This set of tools is used widely for segmentation and divides a population into clusters such that the members of each cluster have similar characteristics. In other words, they look alike.
But too many companies stop at income based segmentation. They herd their customers like sheep into a handful of purpose built stockades with signs reading the designated income brackets hanging above the entrance. You probably live on a street where your neighbours earn roughly the same amount as you, and therefore for most companies, you’re all the same. Yet you just have to be invited into your neighbours’ homes and look at their decor, taste their food or page through their photo album to realize how different you are.
Profiling your customers provides deeper understanding into who they are and how they use your - or your competitor’s - products. Good customer profiling helps you decide how to talk to your customers, when to talk to them and even what words and images work best in the campaign.
So set clear targets and try to know in as much detail as necessary, who you’re talking to. It pays off in the end.
If you balk at broad brush customer segmentation, give us a call. We’ll help you do a better job of understanding who your customers really are.