Why build models for marketing campaigns? One reason is to help rank and select the right prospects that correspond to your program objectives — those who are most likely to take action on your offers
- For an acquisition program, who is most likely to buy from you?
- For a retention program, who is at risk of leaving you?
It is critical that you strive to integrate a strategic mindset throughout your marketing analytics modeling project, from beginning to end, and then beyond:
- At the beginning — the Discovery Phase — You know what you want to accomplish with the campaign — you should dig a little deeper when gathering the data to go into the model to understand what the data means in order to apply it most effectively. Don’t just take the available variables for granted – what other questions about your marketing problem might the data answer? This is often best accomplished by involving key stakeholders and owners of the data.
- In the middle — the Data Build and Modeling Phases — create variables and develop models that perform well and are stable. Because time was invested during discovery to better understand the data, you are able to focus in on defining and creating variables that are more likely to produce good results. But think a little harder about the models – look beyond the scores, can profiles of your audience at the high end and low end of the scores tell you something that could lead to stronger messaging?
- At the end — the Results and Validation Phases — prepare your scored datasets and share the results of the project. This is where everything ties together to help explain why specific variables are drivers of a predicted outcome. But don’t just present the scores or a gains chart – think graphically about your data and explain your findings to non-statisticians in a way they can make better decisions about their campaigns. Use maps of the highest scoring customers, or create matrices to show how different groups might deliver higher results so you can build response scenarios and fine tune the size and frequency of the messaging.
Once models have been built, delivered and are being used for the campaigns, the focus often shifts to tactical — as implementation takes hold. Is this really the right time to stop thinking strategically? A little more time spent on strategy following implementation may lead to new learning and potentially alter your implementation plans.
Applying a strategic mindset to the post-implementation phase might include the following:
- Test the impact of the model scores against controls to make sure they perform how you believed they would.
- Continuously monitor how your customer base is changing: a sudden influx of a brand new customer segment may be missed by a scoring methodology that only looks back at history to make future predictions.
Finally, given that the marketplace is continuously changing at an ever increasing rate, it’s a good idea to plan for ongoing assessments of programs and results with key stakeholders. Ask yourself open-ended “why” questions to maintain the flow of strategic insight and new ideas.
About the Author:
Paul Raca is the Vice President of Marketing Analytics at SIGMA Marketing Group. Follow Paul on or connect with him on .