As Senior Director, Customer Insight at Best Buy, Matt Smith oversees Web analytics, one-to-one marketing, the Reward Zone customer-loyalty program, market-share measurement, in-store test-and-measurement and an analytics research and development team that serves multiple operating groups. Smith explains the retailer's move into cross-channel analysis, which he says is essential to transitioning from customer acquisition to relationship building.
Why is Best Buy moving into cross-channel analysis?
When you look at some of the traditional tools, either Web analytics or BI tools, they tend to be pointed very much at the business. It's about counting widgets or numbers of clicks and understanding behavior only in an aggregate form – how many bought X product, reported Y problem through the call center or looked at Z page on the Web site. The customer insight team is focused on beginning that analysis from the perspective of an individual consumer. A few years ago we built a customer database that houses [online and offline] transactional data at a consumer level. In the last couple of years we've started to bring non-transactional data into those databases.
Can you describe the data warehousing environment?
We have two primary data warehouses that sit in our IT space. One is engineered for rapid response and cycle time and it's meant to deliver customer information back to the point of sale in the stores very rapidly. The other customer database is engineered for much deeper queries and analytics. Both of these databases were originally built on Oracle. We've begun using Teradata in other places in the enterprise and we're examining whether there's value in transitioning those databases to Teradata. The second database has somewhere around a decade's worth of transactional data combined with third-party information. We're beginning to link non-transactional data into these stores as well. The two databases are also connected; we do some model scoring in the deeper analytics database and those scores out to the enterprise customer database.
What kinds of non-transactional data are you adding to the mix?
The first step we took [two years ago] was to connect the transactional data across channels, including Web site transaction data and in-store point-of-sale data. That lets us see customers who are single-channel and multi-channels, and we know how they behave across channels. We've added clickstream data for those who log in at the site or who use their Reward Zone number, and we can begin to understand that behavior. We're just now working through call center data as well, so if you've called Best Buy for service or a problem with a product or if you called to check on your Reward Zone account, we're just beginning to merge that data with our transactional data.
What will you learn by bringing all this information together?
Number one, we know from some of our early work that some of our best customers tend to be multi-channel customers. They have a large number of interactions with the brand on an annual basis, and the idea is to create a well-rounded view of the consumer. We also know that our best customers drive our business. As we've exposed Best Buy to more and more consumers, we're quickly moving out of the customer acquisition game and into relationship building, so understanding the health of the customer relationship across channels is crucial.
What tools do you have in place to perform cross-channel analysis?
There are two pieces to that. One piece is how you listen to customers, and that's what these databases are doing. The second piece is what you do with that insight and how you talk to customers. We use Visual Sciences [formerly WebSideStory] to do Web analytics and we also use its visualization tools to do analytics [across channels]. We use Unica's Affinium software to understand how to talk to customers. We'll set up a series of business rules around particular behaviors and the idea is to respond to customers with particular messages based on the behaviors we're seeing.
For example, if we see that you're doing research online in the home theater space or we see that you've gone to a store and made purchases in the home theater space, we want to begin tailoring our messages to you to help you build the best home theater experience.
How are you delivering those messages?
Right now we're using e-mail and direct mail. As we continue to redesign the Web site and open up more screen space, individual customer personalization on the Web will become a bigger tool for us. Where we think this will really have power is if we can inform the blue-shirt sales associates [in the store] with the same data we're using to make ink and pixels smarter.
How far out is that capability?
That's a very complicated strategy. We've already begun to do it in small ways, but it won't be fully integrated for a couple of years.
Just what would you communicate to the sales associates?
We've built robust personas for our customer segments, and our sales teams understand the attitudes and motivations of those customers very well. We've also begun to think about the way we sell product in a different way as a result of some of this behavioral work. A couple of year ago, if you came in to buy a digital camera, we thought about you as buying a digital camera. Now we understand, from watching customer behavior, that what you're really doing is trying to build a solution around sharing memories. That expands the relationship beyond just that product to a variety of products that will help you capture, edit, archive and share memories, such as editing software, printers, memory cards, accessories or off-site digital archival. It extends the way you think about what the customer is tying to do. We can take what we learn and apply that to our training for the blue shirts.
Best Buy's customer personas are well know, but can you share some examples?
"Jill" is the affluent suburban soccer mom archetype. Jill is very busy and very concerned about how her family consumes technology. She hasn't particularly embraced Best Buy as one of her favorite shopping destinations, but she understands that it's important for her family, so we help her navigate the experience. Another persona is "Barry," who is an affluent suburban male who enjoys home electronics and complete solutions. Because he's affluent, Barry tends to buy higher-end products.
We've thought about Barry and Jill for a long time, but now we can think about how to help them with more complete solutions. In the future, we'd like to be able to deliver a very particular message to a particular Jill that reflects where she is in filling out a solutions bundle. We would understand the context of the products she has already purchased and we'd know which product we have that would help her extend the solutions and which ones will work with the products she already owns. That's where you can get very personal. It's easier to do that through the mail and online right now, but it's much more challenging to deliver that to a sales associate in the store, so that's the piece that we're working on.
What kinds of personalized messages are you already sending via e-mail and direct mail?
If you bought a home theater solution from us with a plasma TV and some other products, you might get a direct mailing on home theater surround sound, clean power supply or universal remote controls. If you were to buy a notebook computer, you might get a piece on our Geek Squad brand talking about home networking options or flat-panel displays. We can tailor the message based on customer and geographic segment and the customer's value to the brand.
Can you share any examples of cross-channel differences?
[On the outbound side] we can vary the message by channel. We can see different preferences across channels and there are some messages that are more cost-effective to deliver electronically than in mail. That goes back to the rules in the [Unica] Affinium software.
On the Web, we use [Visual Sciences'] Web analytics product to understand behavior at very high levels with respect to how people are navigating through the site and how they're spending their time. We've brought all of our segmentation schemes – the personas and customer value scores - into our Web Analytics so we can understand if a Barry or a Jill is navigating a particular area of the site. We understand where our best customers are spending time on the site and we're able to translate that, because of our customer database, into an online and offline experience, but we don't use that to understand individual customer behavior.
The other side of the Visual Sciences tool box we use is the visualization tools. Those sit on top of the customer database and we use them to do very rapid analytics. It almost puts a GUI interface on a big customer database.
Any final thoughts on the importance of cross-channel analysis?
You're seeing the tools sets begin to come together. When I think about individual customer experiences, each of our channels represents a listening post. Whether it's a call center or a store or a Web page, I'm listening to the customer tell me something. Ultimately, as these relationships with the customer deepen and as the game changes from acquiring customers to building relationships with customers, you have to figure out how to be relevant to millions of customers at a time.
The Agile ArchiveWhen it comes to managing data, donít look at backup and archiving systems as burdens and cost centers. A well-designed archive can enhance data protection and restores, ease search and e-discovery efforts, and save money by intelligently moving data from expensive primary storage systems.
2014 Analytics, BI, and Information Management SurveyITís tried for years to simplify data analytics and business intelligence efforts. Have visual analysis tools and Hadoop and NoSQL databases helped? Respondents to our 2014 InformationWeek Analytics, Business Intelligence, and Information Management Survey have a mixed outlook.
InformationWeek Tech Digest, Nov. 10, 2014Just 30% of respondents to our new survey say their companies are very or extremely effective at identifying critical data and analyzing it to make decisions, down from 42% in 2013. What gives?