Business intelligence tools have long been important to helping companies spot buying patterns, product return rates and other information useful in product development and support.
But the information is by definition old because the analysis often happens days or weeks after the purchase information has been collected.
In this age of social media, the popularity of products can rise and fall with astonishing quickness, so it’s become more important than ever for companies to respond in a flash to what customers want to spot trends that can help those companies quickly avoid market missteps.
Enter Syntasa, a Hadoop-based platform that has honed the broader category of predictive analytics to what it calls marketing analytics as a service. Syntasa runs on the Cloudera implementation of Hadoop.
“We’re taking clickstream data [and] social media feeds and combining that with other data sources, including mobile, to really help companies predict what the outcome [of a customer interaction] is going to be,” Jay Marwaha, CEO of Syntasa, told eWEEK.
The approach has attracted big enterprise customers such as Lenovo and Lowe’s, which have added Syntasa’s platform to their own Hadoop-based database systems to boost online sales by analyzing customer behavior and tailoring their online interaction.
Marwaha said the data can show who among current customers are most likely to return a product even before they’ve purchased it or, conversely, which customers are most likely to be open to special offers and incentives, or “upsell” in sales jargon.
A secondary benefit is security features. The software can identify a buyer’s geo-location so if, for example, a regular buyer in New York is suddenly ordering from Boise, Idaho, that is a signal that someone could be using the account illegally. It can also identify the IP address to see if it links with one known to be used to conduct fraudulent activity.
It’s common these days for credit card companies to call customers when they see what appears to be an irregular transaction, such as a purchase out of their home country, but that check is made after the purchase has been made. Syntasa operates in real time.
Enterprise customers license Syntasa software and can decide how many of the features they want to use and how they will apply them. For example, a sketchy online presence might simply generate an automated message to call an 800-number to complete the transaction where the company could ask the buyer for more identification or they might simply disallow the transaction.
“Location wouldn’t be the only factor. There are two to three hundred data points we can look at to figure out the signal through machine learning,” said Marwaha.
Roots in the Intelligence Community
Before joining Syntasa in 2012, Marwaha was CEO of Absolute Business Solutions for six years, a federal consulting firm that provides services in big data analytics, cyber-security, CIO support and intelligence services. He said Syntasa’s technology came from work done in the intelligence community, where typically the goal is to find security threats, the proverbial needle in the haystack.
“Now we can identify security issues, but it’s mainly used to find customers,” he said.
Syntasa has also added services designed to augment traditional databases and analytics software, including data adapters for Adobe Analytics and Salesforce.com shops.
While Marwaha prefers to position Syntasa as a service that enhances the big data systems enterprises already have, he notes it can either augment or outright replace certain business intelligence and reporting tools companies have been using.
“It’s really a leap for many companies because they’ve never had behavioral information before,” and the tools they’re using aren’t really predictive, he said. “Most enterprises are really using tools to report data and not doing anything to predict.”
The Syntasa platform has open APIs letting data scientists build their own models and plug them right into Syntasa applications. While retail and financial service customers have been a primary focus, Marwaha said the company is pursuing other opportunities, such as health care.
Currently, the typical patient interaction that’s recorded is between the doctor and the patient. Marwaha said Syntasa would be able to analyze a broader set of data, including the interaction the patient has across the hospital system. “When you can see that from a patient’s point of view and understand the pain points, that’s where the hospital or institution can save costs from an operations point of view. That’s all new—we haven’t touched that yet.”