Most marketing professionals know that predictive marketing can produce dramatic improvements in business performance: reduction in customer acquisition cost (CAC), increased customer lifetime value (CLV) and lower churn. But still, they hesitate to take it seriously. Why?
We’ve been wondering about this for some time, so we started asking marketing directors about it whenever we got the chance. Here are some anecdotal concerns we have gathered:
- “Doesn’t predictive marketing require software coding and math skills? I’m afraid my team and I don’t have the right competence to do things like data visualisation and machine learning.”
- “I got into marketing because I wanted to do creative work. This sounds like BI.”
- “Our data is bad and there probably isn’t enough of it.”
- “Predictive marketing sounds expensive. Only big companies can afford to do stuff like that.”
- “Isn’t AI bad? I’m afraid of the robot uprising.”
As predictive marketing specialists, we hear you. These are valid concerns, but some of them are clear misunderstandings and others have simple remedies. Here are our responses.
COMPETENCE GAP. Setting up the data and analytics environment for predictive marketing is hard and it does require mathematical, statistical and programming skills. But the ability to run predictive marketing programs and utilise the analysis that results is well within the skill set of any typical marketing team. Most marketing departments use either their own BI staff or 3rd party partners to set up and maintain the analytics system but run all the predictive marketing activities themselves. In fact, they are much better suited to take advantage of the system than those who set it up.
LACK OF CREATIVITY. Predictive marketing will not reduce the need for persuasive content - quite the opposite. Since predictive marketing enables more detailed targeting and personalised messaging, even more, creative content is needed.
VOLUME AND QUALITY OF DATA. Quite decent analytics can be done with relatively small datasets and less than perfect data quality. Predictive marketing uses machine learning, and machine learning has mechanisms to fill empty data fields. Good predictions can be made with just a few thousand rows of data. The quality of the analysis that results depends more on how well your data reflects reality, so it’s worth investigating even if you think your data might be inadequate.
IS PREDICTIVE MARKETING EXPENSIVE? This is definitely debatable. While predictive marketing can be expensive, it doesn’t need to be. There is plenty of easy-to-use, low-cost or even free open-source tools to make machine learning feasible for even small companies. And a lot of money can be saved simply by learning effective ways of working.
AI TAKEOVER. We added this as a joke, but it is a real concern some employees have. Most AI is “narrow” in nature, meaning it is designed to solve one specific problem. What it can do is relieve you of some of your more menial tasks so that you can concentrate on creative and interesting work. While today’s AI might be able to beat us at chess and GO, it won’t be able to take over your job. At least not anytime soon.
Predictive marketing is happening, and many of your competitors are already using it to improve their business performance. If they can do it, why can’t you? If you haven’t started a predictive marketing program yet because you have some of these concerns, maybe it’s time to take another look.
More on the topic:
- Take this test to see if you are ready for predictive marketing
- BLOG: Predictive marketing adds value to many business processes
- Webinar series on predictive marketing
About the writer:
Matti Airas is a consultant in predictive data-driven marketing and customer experience. He has previously worked for the customer experience feedback analysis company Etuma and for Nokia, in the U.S. His passion is figuring out how to use data to solve business problems.
Matti enjoys writing, podcasts (especially on U.S. politics), golf, long walks with his wife and Jack Russell Terrier, and any kind of skiing.