The Scottish Knowledge Exchange Awards 2018 held in Edinburgh on February 22nd. We presented the work we did on machine learning in marketing with Abertay University, Dundee. The audience gave us great feedback and some things to think about going forward. What problem are we trying to solve and how can we use machine learning in marketing as a solution?
What marketing challenges do SMEs have?
Most companies know that marketing has some impact on their business performance. But very few know the marketing activities have the most impact. The result is usually one of two extremes. Either the company spends on as many marketing activities as possible or they conclude that they don’t need marketing after all. Therefore, the challenges are two-fold: What works for my business? How do I prioritise these activities based on a tight budget?
What solution did YO! Marketing propose?
Machine learning in marketing is not widely used yet. It is a newish and exciting way of learning from the past and in real-time. We collaborated with Abertay University to device a model that could identify patterns in data in a supervised way. Using experience and data gathered from 35 companies, we identified critical relationships in the data that could predict the impact (or ROI) of specific marketing activities on overall business outcomes. This means that SMEs can quickly discover what is working to grow their business and focus marketing investment on that. They optimise how they allocate resources and time, and make effective use of a limited budget. Did our solution work?
Was Machine Learning in Marketing a Good Idea?
Abertay University has extensive in cyber-security and data analytics. By working with one of its lecturers, Dr Xavier Bellekens, we combined our marketing experience with machine learning expertise. That was a great idea with many benefits.
We successfully built a model that works, an achievement that is a first in our industry. However, we are limited by data. For a model like this to provide the cutting-edge capabilities that we envisage, we need lots of data. Our current model is a start, and we have a handful of companies working with us to gather more data. If you are interested in what we have built and you would like to support us, contact us for an informal chat.