So what makes the technology different? For one thing, traditional recommendation engines operate in batch mode. “Often by the time product recommendations have been processed, the user is no longer in the market for that product or service,” Jivox CEO Diaz Nesamoney told Real-Time Daily via email. “Machine learning technology uncovers insights about consumer behavior in real time by processing vast amounts of data in milliseconds. It can determine behavioral and contextual triggers and correlate patterns instantaneously, something humans cannot do (or if we try, it would simply take too long),” he said.
The principle behind machine learning is that it learns continuously and the system gets "smarter" and better at predicting what product or message is likely to drive a consumer to engage.
“Recommendation technology, as an application of machine learning, ensures that a brand is able to predict and offer products that individual consumers are most likely to purchase based on correlations between products (similarity) and/or behaviors of the user or similar users. Machine learning-based recommendations take away the guesswork,” Nesamoney added.
Consumers can be influenced by many types of triggers — the time of day, specific offers, friends’ posts on social media, and more. This suggests that personalization technology needs to very quickly interpret what the triggers mean in terms of consumer intent. “Real-time analytics feed the machine-learning algorithms. They are critical to aggregating data that enable machine-learning algorithms to yield the best performance,” according to Nesamoney.
Jivox will demonstrate the technology March 10-19 at SxSW in Austin. The company said it will also share case studies that extend to the retail and e-commerce, CPG, travel and hospitality, automotive, entertainment, and financial services sectors.