Analyzing the past is something we do quite regularly when it comes to marketing-related data. But how many companies really do data-driven marketing where data guides decision making?
In my last blog post I presented the Analytics maturity model that covers four stages of marketing analytics maturity. The model starts with analyzing KPIs, which I covered in my earlier blog post, and moves towards more advanced analytics – predictive and prescriptive analytics.
When it comes to marketing, one very potential and critical application area for these last-mentioned analytic methods is customer churn detection and prevention. There are, of course, other interesting applications as well, as data and machine learning can be used quite widely to help make better decisions for the future.
There are a variety of business-related questions that can be answered with machine learning. In predictive analytics, you can predict classes and classification probabilities, or you can predict a value of something, e.g. profitability and conversion rates.
One could, for example, classify whether a customer becomes profitable or not or whether they will churn. It is also interesting to predict probabilities: how likely is the customer to become profitable or churn.
All the insights gained from descriptive and diagnostic analytics can be done by a machine and you can let algorithms discover insightful patterns. For example, what are the common characteristics of converted/most profitable customers?
Prescriptive analytics helps finding the best course of action for a given situation. Prescriptive analytics gives specific recommendations and helps optimize some measurable target value.
The first two stages of Analytics maturity model, descriptive and diagnostic analytics, are about understanding the data. These steps are common both in business analytics and predictive analytics. After understanding the data and finetuning the right questions, we can start to utilize the data in a more intelligent way. For example, simply calculating KPIs and statistical significances within different cohorts will give you a lot of useful information.
Before moving towards the last stage of the model, prescriptive analytics, predictions and automatic pattern finding are essential steps. Machine can learn which customers are most likely to convert and how likely they are to become profitable. Based on these predictive calculations you could for example decide whether to target the customer at all. Algorithm can recommend which customers you should target to achieve 20 % increase in conversion. However, these recommendations are based on the earlier done predictive calculations. Therefore, you need to apply descriptive and predictive modelling first in order to know what characterizes profitable customers and how much revenue will they bring.
All the above-mentioned questions can be answered with data when the four building blocks, descriptive statistics, diagnostic statistics, predictive analytics and prescriptive analytics, are in place. Would you like to utilize data more efficiently but don't know where to begin? We are here to help you start your journey towards more data-driven marketing.
Mirva is a marketing data professional in Tieto’s Customer Experience Management business unit. Mirva helps brands and organizations develop data-driven digital and customer engagement transformation that enables businesses to adapt to the era of the customer. Mirva has a strong understanding of business processes and visual analytics, combined with theoretical background in computer science and mathematics.