When Bad Personalization Happens to Good Retailers


Retailers are regularly mocked about being terrible at personalization. Late last year, Bloomberg stroked brands with one hand while giving a slap with the other when it published a column titled “Personalization Helps Retailers; Too Bad They're Terrible at It." This was a blanket accusation. Others get more specific.

Every so often an article will come out featuring a story like this: Person looks at a pair of pants on, say, the J.Crew website. Person buys those pants in a J.Crew store. Person is then retargeted online with an ad for the same pair of pants. The article’s conclusion? J.Crew (or other retailer du jour) is terrible at knowing its customer.

Despite the glaring implications of these articles, retailers aren’t stupid; personalization is just really hard.

Knowing that a specific customer looked at something online and then bought it in a physical store is difficult enough to pull off on its own. But to then feed that information to an ad network so it can stop serving up retargeting ads featuring the item the customer just purchased? That’s no easy feat.

But it doesn’t mean retailers shouldn’t try.

Taking the personalization challenge

Currently, there are companies that attempt to solve this problem by working with retailers to track individual customers across every single touchpoint and channel with which they interact. It’s a noble task, but not one that’s easily — or even usually, successfully--pulled off.

The first, most basic step these companies might recommend is identifying each customer and prospective customer through data from customer relationship marketing (CRM) systems, data management platforms, the devices they use, the social media they participate in and a variety of other sources.

That’s a tall order, but we’re just getting started. The next steps involve knowing what customers buy, view and consume, why they make their decisions and who and where they are. Next, it’s time to make personalized recommendations based on their actions, preferences and interests and deliver these messages in the context of where they are, the recent events around them ― oh, and the time of day and year.

Retailers that are getting it right

Rather than mocking those who are doing it wrong, an easier task might be to look at who’s doing it right. And what ‘it’ even is.

When people talk about personalization, they’re usually referring to technologies that enable A/B testing or purchase recommendation engines. However, these activities are outcomes that offer tactical ways for brands to deliver distinct messages to individuals. They aren’t personalization. True personalization strategies come from a position of deep knowledge, and a brand’s deepest, most easily grasped knowledge is what it knows about its products.

Take the clothing shopping service Stitch Fix, which assigns each of its garments 100 or so different attributes (things such as material, color, season, garment type and so on) to get a deep understanding of the variables to which different people respond. Stitch Fix then combines this knowledge with feedback that customers give to their stylists about what items they like and don’t like. Data science then kicks in to understand patterns between things the customer likes across items and pinpoint the exact attributes to which they’re consistently drawn. The result is a dynamic recommendation capability that allows the company to present apparel more likely to please any given shopper.

That’s a very different strategy than, say, throwing products that are supposed to appeal to young professional males in a monthly package and hoping for the best.

Another highly effective personalization strategy comes from Netflix. Todd Yellin, vice president of product at Netflix, likes to say his company has a “three-legged stool” approach to helping people find shows and movies they’re likely to enjoy. According to Yellin, “The three legs of this stool would be: Netflix members, taggers who understand everything about the content, and our machine-learning algorithms that take all of the data and put things together.”

Netflix is in a unique position because its data, its communications, its product and the customer’s experience with all of these things reside in the same place. Retailers, on the other hand, don’t usually see their customers daily so they have to prioritize the personalization of outbound interactions, such as email or online ads, that bring customers back to engage and buy.

Email’s potential for personalization is particularly high for a couple of reasons: 1) shoppers have deliberately opted in to receive communications, and 2) it allows for a cohesive series of messages that retailers can use to create an ongoing narrative with customers over time.

Personalization becomes a lot more interesting and effective when brands start thinking of it in these terms rather than as a blunt instrument for re-selling the customer on an item s/he’s engaged with — or worse, an item the retailer simply wants to offload.

A new way of thinking that’s actually an old way of thinking

What seems like a new approach is actually in line with how marketing teams were structured before the digital revolution. For people who worked in the pre-Internet era, marketing channels are just that ― channels. They weren’t strategies.

Take marketers who wanted to sell, say, Cocoa Puffs cereal (and you thought those chocolatey poofs sold themselves!). They wouldn’t talk about a television strategy or a magazine strategy. They would start off by asking, “Who buys Cocoa Puffs?” and answering, “Moms who have busy days.” Based on that, they’d advertise in women’s magazines or on daytime television during soap operas, all while talking about getting kids to eat a good breakfast.

Then they’d ask “Who influences the purchase?” The answer would be kids, so they’d talk about how delicious Cocoa Puffs are and they’d go out with a memorable commercial that has a crazy bird who gets coocoo for Cocoa Puffs. They’d run that spot during cartoons and have ads in kids’ magazines or around schools and rec centers.

But that’s not the way advertising works today. Today, rather than following a top-down strategy where all channels are working toward a common and unified thought, retailers seem to take a bottom-up approach where each channel has its own rules and those rules don’t necessarily influence or get affected by other channels.

The rules of the road

An email team, then, is limited by some arbitrary rules around how often someone should receive emails — rules that someone truly believes are the right rules for all prospects. They might mean well, but that’s not good enough.

Steve Madden is a good example of what can happen when a brand rethinks its personalized customer contact strategy and unchains itself from arbitrary email rules. Before we began working with the brand a couple years ago, Steve Madden’s strongest personalized efforts were triggered cart abandonment reminder emails. But even those had limitations: the system only allowed these messages to be sent once a day, and then only to site visitors who were logged in at the time they abandoned their cart.

Since then, Steve Madden has worked to reconfigure its cart abandonment emails to send a designated time after the activity ― not just once a day. But that was still just the beginning: the marketing team tested things like which product categories customers had the highest affinity for and algorithms that could predict the likelihood of conversions and unsubscribes.

The impact of these efforts became clear when the Steve Madden team decided to run a test on the effectiveness of these models. The team sent the same email featuring its line of Freebird shoes to two different groups: an audience of past purchasers and an audience who had a high-predicted affinity for the line of shoes despite not having purchased them in the past.

To the surprise of everyone, the group of customers with a predicted affinity for the shoes spent twice as much as the group of past purchasers. The Steve Madden initiative demonstrated that personalization can go beyond triggers to reflect consumer interactions with a specific product. By pairing product attributes with customer affinity insights, the brand was able to deliver the right messages to an audience that needed and wanted Freebird products ― an audience a traditional marketing team would have overlooked.

Personalization is powerful, but that power can be used for good or evil. Done well, it will boost engagement, responses and sales. Done poorly, or without the right data, it can give a brand a bad rap for not knowing their customers and challenge its hard-won reputation as a reliable source of information on what consumers will like. Fortunately, marketers have exactly what they need to do it well right at their fingertips: knowledge of their products’ attributes, an understanding of their customers and the ability to determine where those two intersect.

Jared is the SVP of Marketing and Insights at Bluecore, a high-growth marketing technology company based in New York City. Bluecore’s AI-driven retail marketing platform is used by over 400 retailers to bring together website, customer and live product insights to match customers with the products they love.

This ad will auto-close in 10 seconds