Though predictive analytics and data mining are particularly hot topics at the moment, they’ve been widely used in marketing analytics and research for well over 30 years. Marketing mix models using econometric algorithms were developed in the ’80s and ’90s, while other statistical approaches, such as cluster analysis for segmentation analysis, were already in use at that time.
When I started working in digital analytics about 15 years ago, I found that the worlds of digital and predictive analytics were largely separate. At the time, digital analytics was mainly focused on reporting. The analytics were mostly descriptive, emphasizing what happened and possibly, why it happened. Between analytics technologies not actually being very analytical and data stored in proprietary structures, it was difficult to get data into another technology to do more detailed or sophisticated analytics.
Digital analytics has gotten more predictive recently, with a greater emphasis on what might happen in the future, as opposed to just describing what happened in the past. Companies have made significant investments in getting their digital data into environments where it’s possible to leverage predictive analytics. Data scientists are also being naturally exposed to more and more digital data as their business becomes more and more digital. Additionally, some digital analytical technologies are building predictive capabilities into their platforms.
As a result, we can see predictive analytical techniques being used and deployed more widely to address such issues as understanding marketing ROI, scoring customers for the next best offer, and understanding which customers are at risk of churn. Data mining and machine-learning techniques are being used to create more powerful audience and customer segments. So where is all this going?
Predictive analytics is evolving into prescriptive analytics, a mash up between the worlds of predictive analytics, and simulation and optimization, which have traditionally been used to understand the best course of action given a series of constraints. Though these techniques, such as linear programming and Monte Carlo simulation, have their applications in marketing analytics, they’re deployed far more extensively in areas, including supply chain and logistics. So whereas descriptive analytics address what happened and predictive analytics address what might happen, prescriptive analytics answers the question, “What should I do?”
Let’s use a GPS navigation system to explain the difference. The descriptive analytics tell us which way to go, while the device’s predictive analytics tell us when we are likely to get there. If our GPS has prescriptive analytical capabilities, it will monitor that forecast in real time against my objective, which is usually to get somewhere as quickly as possible. If, on the basis of the feedback, it finds that another solution will be better, it will recommend that I change my current route and take a different route instead. It tells me what I should do.
In the marketing world, descriptive analytics would tell me that a customer has churned. Predictive analytics will tell me that a customer is likely to churn. Prescriptive analytics will tell me that a customer is likely to churn and what the appropriate intervention strategy should be, based upon my objectives and constraints at that time.
Some argue that there is no real difference between predictive and prescriptive analytics. From an analytical perspective, I can see where they are coming from. The algorithms are predominantly the same, with some additional optimization ingredients thrown into the mix. However, I think there is a real difference when one looks at the operational readiness required to successfully deploy a prescriptive analytical capability.
In terms of the data – which will be predominantly real time, consist of multiple and integrated data sources, and be both structured and unstructured – the analytics will be integrated into the technologies. The algorithms will also need to be adaptive, meaning that there needs to be a feedback mechanism in place. Finally, workflow and governance must exist around the data and technology to ensure that objectives and constraints are in place.
Few companies are at the stage where they’re deploying prescriptive analytics across the enterprise and for the foreseeable future, I think it what will define those who are able to be analytical competitors.
Marketers create personas to better understand their target audience and what it looks like. If marketers can understand potential buyer behaviors, and where they spend their time online, then content can be targeted more effectively.
What’s behind a successful data-driven marketing strategy?
One of the major challenges in the martech industry is getting the attention of prospects in a world where they are bombarded by content and emails on all sides.
Facebook is addressing one of the biggest missing pieces of its chatbot offering: analytics.