How Machine Learning Can Improve Direct Marketing Results
Posted by Francel Mitchell
February 26, 2018
With direct marketing, you likely have a benchmark success rate from previous similar campaigns that you base your goals on, and you try to optimise it through various strategies, whether by trying to offer the best deals or by using behavioural tactics.
One of the most successful ways to optimise your direct marketing results is with machine learning. Implementing a machine learning approach in your direct marketing efforts is sure to result in improved conversion rates. In this blog, we discuss using machine learning for direct marketing in finer detail.
How can machine learning be applied to direct marketing?
When using a large database for a direct marketing campaign, a key objective is to identify and remove prospects not likely to respond, or that represent a high risk to your organisation, to minimise time and resources spent on leads that won't result in ROI. By removing those not interested, you will improve your response rates: making your campaign more profitable.
You also want to identify the leads who are most likely to become good, high-value customers, to focus your efforts on the leads that will improve your company's bottom line – and the longer they stick around, the better.
With a data-driven approach like machine learning, you can identify which leads will respond to offers and which are high-risk and unlikely to respond. By segmenting leads into categories, you know which leads to make offers to and which to exclude from your direct marketing effort. This allows your focus to be on leads who are likely to convert, resulting in a higher ROI.
Does it work?
A data-driven approach has resulted in huge lifts for some of our clients. When comparing the results of campaigns using data-driven lead selections to the random sample (or holdout), one of our Insurance clients experienced a 5.5 improvement in Quotation Rate and Activation Rate. Challenging Machine Learning models against more traditional scorecards also provided an average lift of 25% over time.
It's always a good idea to implement new strategies on sample size, and not to the entire universe, to measure lift provided by these new strategies. This will allow you to motivate the investment in optimising your data-driven selection approach but also enables you to understand the benefits and ensure you are making improvements and moving forward.
How to get started?
Implementing machine learning can be a daunting prospect, with a lot of infrastructure commitments and you need to ensure you have the right people with the right skills to develop the machine learning models. Principa has created an end-to-end solution that allows our clients to circumvent the need to invest in additional infrastructure, set up a team of data scientists and train machine learning models. We develop bespoke machine learning models for you based on your previous campaigns and apply those models to the data universe to assist you in selecting better customers. The selection will be made both from a risk and a propensity-to-respond perspective.
You pay a monthly subscription, and you don't have to worry about the ongoing maintenance, hardware, service and security upgrades. When you don't need machine learning for your lead selection anymore, you stop paying the monthly sum. There are multiple benefits of following this approach, especially if hiring a team of data scientists isn't part of your business strategy.
Francel Mitchell is the Head of Decision Analytics at Principa. Francel’s team has a winning track record using descriptive, predictive and prescriptive analytical techniques within the financial services, marketing and loyalty sectors. Utilising available data and through the application of advanced analytical techniques, the team takes pride in their ability to predict human behaviour that can be used to assist business in making profitable decisions.