AgilOne has rolled out an enhanced version of its predictive marketing platform with new tools, as well as improvements to fine-tune its forecasting capabilities, said Chief Marketing Officer Dominique Levin.
The focus of the upgrade was to automate even more of the analytics and subsequent decision-making by users, she told CRM Buyer. “Ultimately, our goal with everything we do is to make predictive analytics and the use of big customer data accessible to all marketers.”
One of the new tools is Revenue Finder, which uses predictive analytics to uncover underserved customer segments. The other new offering is Turnkey Actions, which automatically generates relevant campaigns in response to the insights uncovered by Revenue Finder, as well as in other scenarios.
“For us, it is not just about shifting through data for customer insights, but also about how to make those insights actionable,” Levin said.
It was not by accident that the company chose the “Revenue Finder” name. Its guiding motivation is to help users mine their customer base for further revenue, either by upselling or cross-selling to users, or otherwise making the right offer at the right time.
Revenue Finder is a machine-learning powered set of reports that show marketers which customer segments require which types of treatment or campaigns.
“It works to uncover the hidden gems among a customer base — such as, for example, customers that purchased during a sale and might be amenable to purchasing again if the offer is relevant,” Levin explained.
Cluster algorithms do a lot of the heavy lifting, she said. Reports are generated by customer lifecycle to better highlight the opportunities for additional selling.
Each revenue report is linked to a Turnkey Action. These are pre-built automated campaigns that align with the reports in terms of acquisition, growth and retention strategies.
The campaigns can be launched automatically when the report delivers. For instance, if a report identifies a group of high-risk customers — say, customers who show signs of leaving for a competitor — then the system might send a promotional offer. To one-time buyers who never have made an additional purchase, the system might offer a coupon for a product that is complementary to their original purchase.
“These all can be launched with a couple of clicks,” Levin said.
Under the Hood
AgilOne enhanced its predictive analytics capabilities in general, Levin said. “These upgrades were basically enhancements done under the hood, so to speak — basically, the user doesn’t even know they are there, but in the background they are helping to make better decisions.”
For example, enhancements made to the data-cleansing engine let marketers make better matches between customers even when an ID is not a 100 percent match. A customer might use two or three email addresses, Levin said. The system is advanced enough to make certain assumptions that various email addresses belong to the same person — without disrupting the customer transaction.
These enhancements were seeded throughout the predictive process. AgilOne makes roughly three types of predictions about customers: clusters, or segmentation; product preferences, or brand affinity categories; and collaborative filters that connect a customer’s purchase to other items that might be of interest.
The system also can predict how likely a customer is to make a purchase based on past behavior — a useful data point when applied to abandoned carts, Levin said. “If a customer with a high likelihood to make a purchase abandons a cart, the system might decide it is counterproductive to offer her a discount to come back, since she probably is planning to anyway.”