Predictive analytics approach focuses on customer needs and preferences which enables marketing departments to develop campaigns that generate high response rates. By selecting the right customer, channel, timing, and offer, predictive models give marketers the information they need to create compelling, profitable campaigns.
Selecting the right customer
This model is used to determine which customers or customer segments to target. The use of predictive models often significantly reduces the number of customers contacted, which results in measurable cost reductions. As a result of this first step alone, predictive analytics typically reduces campaign costs by 25 to 40 percent, while maintaining or even increasing response rates.
Selecting the right channel
This predictive model helps marketers determine the best way to contact each customer. By using each customer’s preferred channel, companies increase response rates. The channel scoring model enables marketing departments to optimize their outbound campaigns across channels, by selecting the best channel for each customer (based on channel preferences and predicted response), balancing expected profits with the cost of the channel, and taking channel constraints into account. If a channel reaches capacity, for example, predictive analytics suggests switching to a backup channel (second best option) to ensure completion of the campaign.
Selecting the right time
Consumers today have many choices for meeting their needs. That’s why it’s critical to reach customers in a timely manner when their behavior indicates an unmet need or a risk of defection or attrition. The tastk of AI assisted campaign automation is to continually scan customer databases for just such events, and trigger specific campaigns when a need or risk is detected. This event-marketing approach can result in as much as double the typical response rate.
Some companies increase the frequency of campaigns to improve the chances of reaching customers at an ideal time. Rather than offering a certain product once each year, they run campaigns for that product every month or week. These campaigns target fewer customers, but the customers they do target have a high likelihood of response. Machine learning enables marketers to schedule recurring campaigns, and to use predictive models and event triggers to select the appropriate customer targets.
Selecting the right offer
When companies increase the number of campaigns they run, they risk alienating their customers by overloading them with offers. Conventional campaign management tools are not designed to address the potential overlap. Using predictive analytics, however, reduces this risk. The process begins by making the customer—not the campaign—the focus. For each customer, a machine learning model evaluates all of the available campaigns and selects the
one that best balances the customer’s likelihood to respond with the profit potential of the campaigns. It also takes into account suppressions and contact restrictions, such as “do not call” or “do not contact more than once every two months.” This customer focus, combined with the ability to optimize campaigns around restrictions and preferences, has enabled companies to report a profit increase of between 25 and 50 percent