Blogs

Post a Comment

Are you really who I think you are?

December 8, 2011

The promise of higher online sales on Black Friday, Cyber Monday and throughout the holiday season creates a real incentive for retailers to improve their understanding of consumers. “If I can really narrow down and profile my customers better, I should be able to make them an offer they can’t refuse.”

With online websites claiming to know who each customer is and providing visitor profiles and cookie information back to advertisers and online sellers, holiday promotions could reasonably be expected to deliver high response rates. The problem though is that online profiles can be inaccurate.

Cookies are often hard to use to identify people. According to Comscore an average of 56% of all cookies point to multiple people, so only 44% of the cookies can actually be used to identify a unique person. Profiles for people who are using the same cookie can often be very divergent. My interests in sports equipment and electronics mean that I shop quite differently than my wife, who shops for women's accessories, school supplies, travel deals.  Our son tends to visit the Nickelodeon website to play his favorite online games. Yet on our home lap top we have one cookie that combines all three of these profiles.

Finding unique identifiers and linking behaviors to individuals is the real challenge. At my house we use the same credit card for lots of purchases – from a new cell phone charger at BestBuy.com, to a new travel guide from Barnes & Noble, and a new pair of boy’s soccer shoes from the adidas website. Just knowing what was purchased doesn’t offer enough clues to tell marketers who the technophile (me), reader (my wife) or sport enthusiast (my son) are in our house. Linking behavior to individuals requires more complex analytic modeling, especially when that behavior occurs online and off line, over time, across devices (tablets, laptops and phones) and with different payment methods.

What is clear though is that when marketers do manage to identify me individually and successfully connect my in-store activity to my online activity, they will have a better chance of giving me a promotional offer that matters to me…and that I might respond to.

Making sure that application providers and websites appropriately reveal user information, and adhere to privacy policies is certainly a valid concern for most of us. Just look at Mark Zuckerberg’s recent apology for inappropriately revealing Facebook users’ information. But an equally strong case can be made that online marketers (including retailers) are dramatically underutilizing the consumer data that is legally and fairly available online. For example According to Exact Target 54% of people unsubscribe from permission emails because they received too many emails and 49% unsubscribed because the content is repetitive or boring.

Do retailers have the tools and solutions to really understand online and offline behavior models and bring them together under one umbrella, rather than keeping them in multiple silos? Email marketing response rates will be low when you know little about your customers’ needs and behavior and fail to create meaningful customer segments and deliver tailored offers.

In order to understand customers, retailers need to collect enormous volumes of information on customers’ behaviors, identify each customer individually, and then perform the necessary association and linkage to tie together each customer’s transactions, clicks, tweets and facebook update into a single view. Retailers need to have the ability to organize the multiple sources of information they get in a way that is flexible and efficient, and then analyze that information to make appropriate connections and correlations. Finally retailers need to be able to conduct this process in tools that can return answers quickly and support advanced analytics and segmentation - a challenge that will only grow as the mountains of available digital breadcrumbs on customers’ behavior, interests and desires continue to grow ever larger.