customer analytics

Everyone is talking about big data. People are excited to store it, process it, analyse and monetise it. Data analytics has been mainstream for some time now, but it is now becoming much more prevalent. Most businesses start their analytics journey by analysing data related to their customers. Most businesses have customers and beyond a certain scale, all generate some sort of data. It’s no mystery that great customer engagement can result in increased revenue over time. Turns out that data has a role to play here. By analysing data generated by customers in various formats, businesses can gain better understanding about their customers and in turn take more informed decisions. In past few years, marketing functions specifically are becoming increasingly data driven.

Analysing customer data and deriving meaningful patterns from it is a science in itself and volumes can be written on it. But most of the analysis and modelling can be summarised in few broad buckets listed below. Service companies have been providing insights through these models for many years now, and product startups are mushrooming world over to standardise and scale these models. Here’s a quick look into these models :

1. Customer segmentation — These models clusters the customers on basis of behaviour over time. Behaviour includes aspects like frequency, value spend, recency etc. While statistics has a key role to play here, it’s domain knowledge that counts more.

2. Cross-sell model and up-sell model — Based on customer’s purchase history and profile, these models recommends what products he / she can buy in future. Companies run campaigns to cross-sell or up-sell those items to customers.

3. Churn prediction model — Predicts which customers are about to leave and presents timely data to take appropriate action

4. Customer Lifetime value — These models estimate how much worth is the customer to business. This is the foundation for quantifying ROI in marketing campaigns.

5. Next Best action — A difficult to implement predictive model, which estimates what is the next best action business can take to engage with every customer. Is it to send a feedback message, a promotional message, a sales pitch, a discount etc ?

The first two are examples of descriptive models, the next two of predictive models and the last one is example of prescriptive model. My suggestion to businesses which are just starting out is to start with simple segmentation models. Based on understanding of the customers, they can classify the customers into various buckets — like frequent, high paying, dormant customers etc. The segments change over time and customers move from one segment to another. This approach will provide businesses a great starting point to explore data and understand broad behaviour of their customers in more detail. Going ahead, they can invest in building more sophisticated models for predicting churn and customer lifetime value.

In future posts, I will talk in more detail about these models. Stay tuned and please give your feedback. I would love to hear from you.

And thanks for reading : ) !