Kevin Ertell is the SVP of digital at Sur La Table. His team defines ecommerce strategy, owns the cross-channel digital customer experience and oversees technology for the premium cookware retailer. With over 30 years of ecommerce experience, Ertell is an in-demand thought leader and serves on the Shop.org board of directors.
Personalization has been the go-to buzzword in retail for a couple years now. Is it overhyped or under-leveraged?
Yes to both. Partly because the term is overly ambiguous, partly because everyone says they do it and define it differently. The term has become so overused it’s not meaningful. That said, I don’t think many people are doing it well. A lot of retailers have tons of data, but accessing or analyzing it properly to get meaningful personalization is a hard problem.
Also, I think the ability to personalize with today’s technologies is really dependent on a couple of key factors, like the frequency of shopping and the types of products sold. Certain retailers are naturally in a better position than others because their categories are better positioned in these two factors. For example, when I was at Borders and Tower Records, we had lots of repeat business and we sold products that were character-defining. If a customer recently purchased Iron Maiden, Anthrax and Metallica, I can assume showing a page that features Britney Spears and Adele is probably not going to be all that effective.
Kitchenware is harder. We don’t have nearly as much repeat business. If someone buys a dutch oven, a baking pan and knife, what can we assume about them? Even if someone makes two consecutive bakeware purchases that doesn’t necessarily mean we shouldn’t show them knives. In our categories, we can’t as easily confine a segment only by looking at what they recently bought. Understanding customers’ shopping patterns across channels and devices is also complicated the issue.
When you think of personalization, what comes to mind?
I think of really local, traditional retail. I remember Sam Drucker in the show “Green Acres.” He knew every customer, hung out with customers, could recommend what they need or what they might want based on that relationship.
Of course, that model doesn’t really scale and it doesn’t work in the self-service environment of a web site. If we’re going to try to create a very relevant experience for large groups of visitors, scaling and self-service are important.
Can algorithms replace Sam Drucker? Sales associates in our stores benefit from a lot of signals they can use to almost instantly personalize service for customers. When a customer walks in the door, the associate is already armed with inherent knowledge of local culture, weather, etc. When she first sees the customer, she starts taking in more signals. Is she alone or with someone? How is she dressed? How does she carry herself? What is her general age? Service can get directed very quickly with a couple of quick interactions. So how can we get closer to that online? Does it requires a back and forth interaction (and identification) of the customer? Ultimately, personalization means how do you provide right service at right time in the right way.
What tech vendors come first to mind around the topic of personalization?
Monetate, BlueCore, Certona, Rich Relevance. All come at it from very different angles.
How do you expect personalization will help you this year?
BlueCore is good example. We have solid, recent data and are able to react to that with triggered emails. On site we are trying to do everything we can with Monetate, using data about that customer and macro data around circumstances (geo, weather, time of day). For example, on geography we can focus the home page on outdoor cooking for people in the south, while the east coast gets Dutch ovens.
At Borders we did an email called Borders Monthly. It was relatively low tech. We asked people when they signed up what book categories. There were 35 categories. When we created the email it was modular. We had “pods” for each of the categories. Buyers would give five books for each pod or category. Then we used CheetahMail to choose which pods went to which people. I calculated at the time that there were 4 billion possible permutations. I’d love to find more ways we can do that sort of thing at Sur La Table, both in emails and on the site.
It is dependent on implicit or explicit insight into the customer. It’s tough to figure out the implicit/explicit data in the kitchenware category. If we go on what they bought, for example two baking pans, are they a baker? A possible next step is get our customers to interact with something and/or give more information. If we get explicit data, we’ll map to merchandising or perhaps even change the merchandising strategy.
What are some challenges you face in pursuing personalization?
One obstacle is coming up with creative assets. We don’t have the resources to create exactly the right creative for every customer, of course. We’re trying to find ways to incorporate more modular designs so we can mix and match template creative.
Data availability and accuracy is also a challenge. Tags can be sketchy and are often missing data. As an offline retailer, getting that data and making it available to personalization tech is also hard. Online personalization tools are not always directly connected to POS. Then you have the mobile visitor which is difficult to connect across devices and channels. When you waterfall all that out, you may end up with small audiences.
We’re investing more on personalization using Monetate. However, we’re in a “throwing spaghetti at wall” approach. We try lots of stuff and ideas can come from anywhere.
In our testing program, a year ago we had one of our biggest finds. We we doubled font size of price it had a huge impact on conversion. However, there is only so much you can do to lift conversion across all audiences. That’s where personalization comes in.
Plus, it is part of our brand promise, to provide tailored help. We equate this to the in-store Sur La Table experience, where the sales associate interacts with a guest and can take in many signals from the customer. If we can create more interaction on our site, we will have a lot more information and get to an online-equivalent of tailored help.