image

4 Data Scientists

image

Informed and targeted marketing

image

Reliable value prediction of marketing campaign

Client

Ted Baker

Purpose

Ted Baker is one of the UK’s leading designer lifestyle brand. One of the ways that retail brands can increase customer engagement is through targeted advertising. However, to do this it is first necessary to identify who your customers are and their spending habitsin a process known as customer segmentation.

Approach

To conduct their investigation,a team of four data scientists were given access to sales and customer data to explore, spanning from 2007 to early 2017. Using machine learning algorithms, the team revealed five key segments in the market: low spenders, infrequent high spenders, frequent high spenders, discount hunters, and unsatisfied customers (characterised by the fact this group only returned items).

Using a predictive algorithm to classify new customers as one-time only customers or repeat customers, the team built an algorithm demonstrating that revenue generated by a targeted marketing campaign could be increased, when compared to a blind marketing strategy.

This is our first investment in data science within marketing and has helped us understand where to get the biggest return on investment from every £1 spent. It is clear that Pivigo has demonstrated the value of data by optimising our search ad spend to convert more new customers.

Claire Holden, Global CRM Manager, Ted Baker
The Outcome

One of the models used a 10% discount and a 4% conversion rate,targeting 60,000 customers. This model allowed the team to predict that a targeted marketing campaign would generate an increase of 1100 per cent (ex costs) in revenue compared to a blind marketing campaign. Following an analysis of purchase history,the team were able to provide recommendations to increase sales – including a list of customers who are most likely to buy a particular item given their purchase history. Further insights provided by the team, showed that by combining the sales data they used with web analytics data (for example how many times a customer visited the website before purchasing), revenues could be further boosted. They were also able to utilise association rules mined from customer purchase histories to generate marketing leads and a recommendation engine for products. Finally, the team identified that the value contributed by the frequent high spender group is not growing over time, unlike the other groups. This allows the marketing team to decrease their investment in targeted campaigns towards this group.

Connect with the Data Science Community

For Data Scientists & Businesses

TOP