Transforming data lakes into predictive engines
If you're looking to outperform digitally, the possibilities are endless with technology that's becoming smarter by the day. Where once we often made business decisions using our common sense, today we increasingly rely on machine learning. If you've built a robust data lake, you can use it to predict the success of propositions.
How does machine learning work?
Here's an example to understand it: Imagine a travel agency wanting to predict which type of customers are likely to book a luxury cruise. We've gathered a large dataset with information about their customers and booking history. This dataset contains input variables that help predict which type of customer we can best target with marketing.
In the table above, the first 70% of the customers (for instance, customers 1 to 70) are the training set. We use this data to train the model. The model will learn which characteristics (like age, income, number of previous cruises, whether they're part of a loyalty program, and how many vacations they take per year) are key indicators of whether or not someone will book a luxury cruise.
Next, we use the remaining 30% of the customers (customers 71 to 100) as a control group to see how accurately our model can predict new data. We'll 'feed' the model all the input variables of the test set, but not the outcome of the last column. The model will make its predictions, and we can now compare these with the actual data to assess the model's accuracy.
The outcome here might be that a new customer with a high income, who has already booked several regular cruises and is a member of the loyalty program, will book a luxury cruise. This helps the travel agency in targeting marketing efforts and offering personalized deals.
Machine Learning Requires a Data Lake
Make no mistake, you need a variety of data to eventually learn from it. So, if you want to start with machine learning, begin by collecting customer data and developing your data lake.
In the development of your data lake, use First-Party data and techniques like server-side tagging. The more data you collect, the more opportunities you have to optimize that data. Consider, for instance, A/B testing based on Cialdini's 7 principles of persuasion. This way, your data lake grows and increasingly gains predictive value.
Why Choose New Story
Ready to dive into machine learning? Let's plunge into the data lake together. With the data-driven marketing team at New Story, you're set for next-level marketing. From our blended expertise - strategy, design, technology, and marketing - we always find room for improvement.
Let's chat without obligations
Schedule an appointment with New Story. We're eager to learn more about your organization, market, customers, and revenue streams. And of course, we want to understand how your digital landscape is currently structured to identify where the opportunities lie.
Please reach out to: