A beauty retailer approached us with a unique challenge: they wanted to increase customer retention by using their historical transactional data and psychology to establish what drives people to change - switch from buying one product to another and similarly what drives someone to start purchasing a new item.
The majority of academic and industry projects have hoped to predict one’s behaviour by heavily focusing on purchasing attitudes. But attitudes are not a very good predictor of customer behaviour - which is why we decided to turn this problem on its head: we worked backwards and using the previous purchasing behaviour and customer psychology we were able to predict one’s propensity to start a new pattern of purchases. Simply said - we wanted to predict one’s probability to change based on what they did before.
"We want to predict which customers will change their behaviour based on what they bought before and based on their personality traits."
Although much is understood about human behaviour, there is relatively little knowledge about what directly leads to behavioural change, especially in terms of their purchasing behaviour. The first assumption we made was that the change of customers purchasing behaviour can be observed as the proxy of the change in the level of purchase of a specific product of a brand between two sequential periods in time, on the specific market. We also designed a questionnaire that refer to specific psychological traits and demographics as additional features in our model. Our task was executed as a classification task, using the machine learning approach and we were able to develop the Change Readiness Index, as a “likelihood that someone will increase purchasing a product (or category) of a brand ‘a’ over a given time period x”.
We then extended this work using a novel machine learning technique called Model Class Reliance, which enabled us to also understand which features were the most predictive of one’s behaviour change, with a goal of changing the level of sales by adjusting these factors. This also has implications for marketing communications: by recognising which factors lead the sales, it would be possible to adjust marketing campaigns and push the “buy button” in the mind of consumers.
In the case of a this particular makeup brand, we analysed their lipstick product in this work: we found that the driving features that predict customer change were:
Total spent on the product category,
Based on this work we developed a strategy for the brand that would help them engage with customers with a more personalised approach (tone, message, colours and strategy) - with a focus to nudge them to change. Among many suggestions we recommended starting with the practice of lipstick refills - since conscientious people would feel more comfortable with the idea of being responsible, whereas the ethical approach (and less land waste) would appeal to agreeable people.
There were several benefits of the particular approach used in this work.
We were able to:
For each customer, identify the propensity to change, which enabled our client to approach them in a timely manner and increase the customer retention
Identify factors that influence change in one’s purchasing behaviour and create personalised recommendations based on these factors and using the science of nudging: we made suggestions about how to talk and engage with customers
The overall approach enabled our clients to have a better idea about the future sales and adjust their approach - whilst working with us on numerous following projects
Discovering what drives changes in purchasing behaviour was a game changer for sales conversion. Customer data is now at the heart of the business as our client continues to offer gold standard products and service.
If you would like to find out more about this case study please contact us.