A video from the TED Talks series has been spreading through my digital marketing circles like a summer cold: Eli Pariser: Beware online "filter bubbles." The video is often shared with a comment to the effect of, “You must watch this!” And, yes, it’s quite good.The upshot? That personalization algorithms for online content are shaping what we consume (and, to some extent, what we are able to consume) so that we are less and less exposed to divergent ideas.
Pariser’s talk goes beyond esoterica, though, and cites how Netflix, among other commerce examples, applies personalization. That’s what got me thinking about how this algorithmic content evolution relates to online marketing, e-commerce, and our ethical responsibilities as marketers. Oddly, even though many of my friends who shared this video are themselves thinkers about digital marketing and online social sharing, there seems to be very little introspection about what the “filter bubble” effect means in terms of online marketing and ethics. So if you’ll pardon the navel-gazing, I’d like to examine this a bit and would encourage you to do the same; let me know your thoughts in response.
Certainly at [meta]marketer, we encourage our e-commerce clients to test behavioral targeting. There's typically a great deal of convenience that this kind of targeting affords the customer. For example, Amazon knows that I tend to buy raw vegan cookbooks, and so it tends to show me the latest and best-rated related books in my browse path. I welcome this because I get exposed to books I somehow might have missed but will almost certainly like. On Netflix, too, there's a good chance that showing me personalized suggestions will save me time and delight me, even as it reinforces my longevity with the site and ensures my subscription payments for months to come.
This selfish/selfless balance is the new normal in marketing optimization. It's what I personally am passionate about: using data to create better customer experiences and, simultaneously, generate incremental profits. It's what we do with our clients, and their KPIs speak for themselves.
But are we contributing to this insular and narcissistic phenomenon where the more time individuals spend online, the more they start to have mirrors set up around them so that they can no longer see diverse behavior, but rather increasingly similar likenesses of themselves? Perhaps. After all, one of the keys to the work we do is an emphasis on relevance. As I think of it, relevance is a form of respect. It shows customers that we respect their time and effort enough not to make them scour the site for what they're after.
Chris Brogan and other digital thought leaders have spoken about social news as a serendipity engine. (Serendipity, incidentally, has long been my favorite word and a beloved concept.) In earlier iterations of social news, you got what you got. So, too, in early e-commerce. As the availability of information has accelerated, though, and personalization algorithms have evolved, some of that serendipity has been traded for distillation and, yes, relevance. So, sure, from an editorial perspective, in the video Pariser is justified in saying that Mark Zuckerburg’s example of “a squirrel dying in front of your house” is not as important as “people dying in Africa.” But in commerce, the dilemma of moral or ethical priority is not nearly so clear-cut. Perhaps the personalization of search and social news makes it less likely that you’ll happen upon something random and wonderful, but the continued explosion of long-tail content and commerce means there’s randomness even within niches. While the “filter bubbles” Pariser describes might obscure your view of the randomness and chaos of the Web, in general, personalization does help uncover hidden gems within customers’ interests.
Because the other side of all this tailoring and customization is that the long tail is getting longer in every area, and the realization that we’re not going to be able to see most of what’s out there is starting to sink in. So personalized content and merchandising is as much a response to information overload as it is to data availability. Going back to my earlier example, if I landed on Amazon's home page and it made no effort to customize the content for me, it’s likely I’d have little idea of the breadth and depth of its catalog as it related to the semi-obscure offerings that appeal to me. Would I think to search for chia seeds, one of my recent purchases at the site, if it wasn't made clear to me that Amazon carried food stuff as well as books (and tools and shoes and sporting equipment. . . and, and, and)?
After all, relevance and targeting are not new phenomena in marketing. We study demographic and psychographic information to understand customer profiles so that we can tailor our advertising placements, our message, and our follow-through for optimal results. What’s newer is the ability to adjust whole experiences on the fly based on behavioral performance. Imagine if you walked into a store--let’s use Nordstrom as an example, since it’s famous for its quality concierge service. As you looked around and your attention landed on an object, the other objects around you shifted. Would you feel more catered to or more pandered to? Or perhaps both? In the context of Nordstrom, where it has been established that it’s trying to improve your shopping experience, perhaps it would seem just another level of superior customer service. If you had the same experience in, say, Wal-Mart, chances are a savvy shopper might feel manipulated.
As a marketer, I see my job as creating meaningful connections between company and customer. (Note that I don’t say that my job is to convert customers: Read my previous CMO.com article on empathy-oriented optimization for why conversion as a single KPI is short-sighted.) As a data-driven, technology-savvy marketer, I know that behavioral similarities among visitors and, ultimately, customers often lead to clues, validated through analysis and testing, that can improve the customer experience overall--and, in turn, increases profit. That this also occasionally means limiting a customer’s view of the site and creating an insulated experience is not only an acceptable side effect, it’s intentional. Because that’s what customer behavior dictates. Customers become overwhelmed when presented with too much choice, and since niche options abound online, that means that if I’m HomeDepot.com and a customer comes in from a search for power tools, I’d best show top-selling power tools and not home appliances or ladders or whatever else.
But perhaps a lesson to take away from this is that there might be opportunity in exposing the customer experience to a little randomness where it doesn’t interfere with the customers’ intentions. A little unexpected cross-sell of something charming, a quirky-but-fun site feature, something surprising and fresh--these types of experiments with commercial randomness might be worth trying in your environment and seeing how customers respond. Because with all of the filtering we’re presented with, the savvy shoppers out there might be picking up on the sometimes heavy-handed crafting of custom-tailored experiences. And maybe, just maybe, we’re all overdue for a little serendipity anyway.