How broader analysis pushed SurLaTable`s response rates up

Analyzing e-commerce and catalog database activity optimizes customer segmentation to identify those most likely to buy, says Sightward, SurLaTable’s analytics vendor.

Kurt Peters

In a head-to-head comparison, cookware and tabletop gear retailer SurLaTable boosted its catalog response rate by 40% over the response rate forecast using a traditional recency, frequency, and monetary model by using predictive modeling technology from analytics vendor Sightward. Sightward technology selected a different segment of the customer base to receive fall and winter catalog mailings, which drove the higher response rate. Though response for the fall and holiday campaign was measured in terms of catalog response only, Sightward used both e-commerce and catalog activity from SurLaTable’s database to come up with its recommendations on customer segmentation for the campaign.

“We help companies identify the customers in their database who are most likely to respond to promotional offers,” says Carla McGrew, director of retail client development at Bellevue, WA-based Sightward. While many customer segmentation tools used by retailers today employ descriptive analytics based on an historical view of customers’ past behavior, Sightward’s predictive model forecasts what’s likely to happen, based on customer attributes that go deeper than RFM’s, she says.

“For example, if 35 year-old soccer moms are your best customers, our technology will help retailers find more customers in the database that act like those customers,” McGrew says. “If the 35-year-old soccer mom is at the top of the bell curve, and you’re directing your marketing message to her, you’ve also got customers that fall to the left and the right of her that are likely to respond to the same offers.”

To identify those likely customers in a retailer’s database, Sightward looks at historical data and runs test campaigns to create models for segmentation based on campaign results. The technology looks at every customer attribute available and the interaction among those attributes to determine who is most likely to respond to an offer. Alone, recency, frequency and monetary attributes don’t factor in other customer data that have predictive value, says McGrew. They don’t get at whether the customer bought on sale, for example, or at demographic and even psychographic data many retailers have in their databases.

Sightward technology bundles 18 possible statistical modeling paths into a single tool that runs the retailer`s customer data against all 18 simultaneously. The analysis results in a recommendation on which model will predict results most accurately. McGrew expects the company will shortly announce campaigns now in development that are segmenting retailer’s databases for e-mail marketing campaigns. “Catalogs are using our technology to decide who to contact, who not to contact and how often to contact them. For e-mail, it’s determining what to e-mail each customer who`s being messaged,” she says.


Customer analytics, Database marketing, Online shopping, Predictive modelling, RFM