Is it possible to use Small Data to make marketing campaign performance prediction?
The definition of audience targeting is the practice of using data to segment consumers by demographics or interests in order to find the right person on the right device at the right moment. With audience targeting, marketing and sales managers are more likely to reach consumers interested in their products with relevant messaging. It also decreases time and money on uninterested people and help move potential customers down the funnel. Data management is key to audience targeting. And it happens that digital marketing allows experts to collect large amounts of data on their prospects.
What about using Data Science in order to better target prospects? In this case study, we build a predictive model which identifies prospects who will subscribe to a banking product (a bank term deposit) as a result of direct marketing campaign.
Marketing and communication experts are not data scientists. They may not have the required skills in machine learning nor in software coding to build predictive models. Typical data handled by these experts are prospects age, gender, social category, number of contacts, type of job. Moreover, most data handled by these professionals relate to a few thousands of prospects. As such they are considered Small Data, as opposed to Big Data (millions of Data). Traditional machine learning tools work well with Big Data but do not perform well on Small Data.
MyDataModels allows marketing and communication experts to build automatically predictive models from Small Data. No training is necessary. These domain experts can use directly their raw data: no normalization, neither outliers handling, nor feature engineering are required. Thanks to this absence of data preparation, the results from this specific dataset were obtained with a few clicks in less than 5 minutes on a standard laptop.
MyDataModels brings a self-service solution for the domain experts who have Small Data and no data scientists.
There are two main types of targeting – inventory targeting, which serves ads on sites that offer a specific type of content, and user targeting – which serves ads to individuals who have exhibited a particular behavior or interest. In this use case, we have focused on the latter one. Through the increase use of digital marketing, compagnies collect first hand data from their own audience and customers. This data is looked at as very valuable due to its quality.
Furthermore first hand data is safe and can be utilized by sales and marketing department. Using these data in order to fuel a prediction engine such as TADA can have a tremendous impact on the efficiency of a marketing campaign.
“Predictive model reached an 85% accuracy rate”
In this use case, the results from MyDataModels’ predictive model reached an 85% accuracy rate. In other words, in 85% of cases a future sale was rightfully predicted. In 85% of prospects approached by the salesforce after being targeted through TADA, an actual sale was closed.
Thanks to TADA, marketing and sales teams have a new means of qualifying prospects which is extremely efficient. Through the use of MyDataModels machine learning techniques, the sales teams will no longer waste precious time and money trying to close a sale with ‘low probability’ prospect and will focus their energy on the ‘high probability’ prospects thus closing more sales.