Discussions on that day developed around the topics of data in general, technology, data-driven design – and even how to make whiskey with computers.
Marketing as a science goes beyond utilising data to build a predictive model that will tell marketers when the right time is to connect with customers, or what should be offered next. This information is definitely valuable, and it takes companies one step ahead, but knowledge alone won’t translate directly to customer engagement and customer satisfaction.
Customers need to be delighted with the ways we approach them, and marketers are meant to provide best-class experiences to them. This means that in any marketing plan for customers, there’s also space for imagination, creativity, and why not, intuition.
Here are four things you need to take into consideration in order to go from merely building predictive marketing models to starting to design customer experiences with them.
It is common to associate customer data only to the data collected by different technologies such as Marketing Automation and CRM. It is often ignored that humans intrinsically collect data through observation, conversations and any type of interactions. The broader concept of data collection implies involving both qualitative and quantitative data to get a more accurate picture of the situation to be analysed. Both research methods are crucial in the design of relevant and more accurate predictive marketing models.
With which method to start first? It depends. Qualitative research is helpful in cases where the variables are unknown, which in simpler words means, you don’t know where to start first. On the other hand, quantitative research is really helpful when the variables are known, as it enables finding patterns and correlations. Either way, they both should be part of the process and complement and support each other’s findings.
Customers can be one of the best sources of information. Involving customers helps discover their goals, hear about their perceptions and learn about their needs. However, customers typically struggle to translate their needs or challenges into a solution. Therefore, it is essential to know how to turn those into something relevant. Innovation plays a main role here, and this leads to the third consideration of involving different perspectives to find “out-of-the-box” solutions.
At this point, it should be clear that data are not only the numbers generated by machines. If this is true, then, why is it often thought that predictive marketing models only concern a group of tech experts centralizing the data for data scientists to model it?
That is the case for building the algorithm, but designing the model requires other disciplines that can translate business requirements, give context to the numerical data and consider other aspects, such as human behaviour and cultural differences. The point here is to allow other points of view and opinions to come to the table because diversity, if managed correctly, can result in greater innovation.
Aiming at designing a perfect predictive marketing model may bias people against the model itself. It is important to understand that there are limitations to the model, such as who collected the data, how the data was collected, what was left out, etc.
Recognizing those limitations is the first step to design better and better models every time. As it was said by one of the panellists, be humble when approaching data. Allow trying new things, using creativity and allow things to go wrong. Test things with smaller samples first to minimize the impact of mistakes and start optimizing from there.
To sum up, building a predictive marketing model is a step within the design process. Design is a more comprehensive process as it also covers the steps that occur before building the actual algorithm and the steps after companies start using it. After all, as one of the speakers pointed out, “data doesn’t say anything, it is rather what you say about the data” and what you do with it.
Juliana Tobon is a Solution Consultant with broad experience in the domains of marketing strategy, marketing automation, and marketing and sales technologies. She has supported several companies in the development and redefinition of their marketing and sales processes, ways of working, and technology architectures.