Personalized marketing: Three key ways to avoid the pitfalls

The end of the year is approaching, and wherever you looked in 2018, it was hard to avoid one theme: Unintended Consequences.

For companies large and small — but particularly in the tech space — there seemed to be new and remarkable ways to screw up lurking around every corner.

Take the world’s biggest social networks. It became only too clear for them this year that a mission to connect the world comes with downsides as well as positives — downsides like foreign actors trying their utmost to mess with the course of democracy.

In the world of hardware, we saw the rise of connected home appliances which were designed to be simple to set up and operate — but turned out to be equally simple to hack and turn into botnets. Oops.

And in the field of machine learning, there were reports that an automated HR system in development by one tech giant had to be junked — because of its tendency to simply regurgitate the biases of the past when making recruitment decisions. (Understandable, yes. Disturbing: also yes.)

Innovation can, to state the obvious, be less than straightforward to achieve. Innovation in a highly networked, constantly changing era can be even less so, as 2018 proved.

And innovation in digital marketing is no exception to these rules.

Let’s agree that personalized marketing can be a terrific thing. If my company can show you a product specifically picked for you, based on what we know from your consumption habits, in the right channels at the right time — then chances are you’d rather see this than being advertised something entirely random.

Better still — the creative we use to promote this product could be specifically tailored to your habits and interests. Double the personalization. Why would anyone have a problem with that?

Well…

In case you missed the story, black Netflix users recently reported seeing their Netflix queues dotted with films whose thumbnails implied black cast members were central to the movie. The reality was often quite different. Frustration and annoyance, unsurprisingly, ensued.

“If something is black, I take no offense in being catered to,” filmmaker Tobi Aremu told The Guardian. “I am black, give me black entertainment, give me more — but don’t take something that isn’t and try to present like it is.”

The lesson: clunky or misguided personalization can be way worse than no personalization at all.

I don’t want to rag on Netflix too much. They’ve been public for some time on how they show different users different creative assets to promote the same film. Enjoy comedies? You might see a more light-hearted poster image. Fan of Robin Williams? If he’s in the movie, he might be front and centre in your personalized thumbnail.

This approach is, broadly, very smart. And obviously, harvesting signals in the background from consumers’ consumption patterns feels like a slick way to offer personalization at scale: just like the company uses uses automation to crop, reversion and optimize imagery for a mindboggling number of placements, languages and territories.

Netflix themselves clearly didn’t set out to disappoint or anger their subscribers via these tactics. And the company was explicit, when challenged, that users’ ethnicity is not directly employed as a targeting factor: “We don’t ask members for their race, gender or ethnicity,” their statement read, “so we cannot use this information to personalize their individual Netflix experience. The only information we use is a member’s viewing history.”

And yet. The offence and annoyance caused to some loyal consumers was no less real, and no less justified.

Push the personalization line just one notch too far, through the use of data — how ever legitimately gathered — and consumers’ perceptions can dramatically swing from delight… to intense negativity.

How to avoid similar unexpected consequences when it comes to mixing data and creative? Three pointers from me.

• Think laterally, and remain alert. You might not be aiming to create particular segments within your consumer base, or target certain groups in a special way — but through combinations of data points and the coinciding of particular characteristics, that’s exactly what you might end up doing. Be alert to the risks, and listen to your audience if they spot issues or tonal mis-steps before you do.
• Avoid a company/team monoculture. Nurture a diverse workforce who’ll be more likely to spot potential traps in your tactics around tailored creative and/or targeting, and who will see things a homogenous team won’t. There is a huge risk in too much uniformity when it comes to gender, ethnicity, background, sexuality, styles of thinking and more.
• Personalized marketing should never over-sell. Just because I can sell you something by staking a claim for personal relevance, doesn’t mean I always should — especially if that supposed “relevance” is a stretch (like, say, a minor movie character being portrayed as the star of the show). As a consumer, your brand experience and perception of me will suck as a result: if not now, then tomorrow.

At Beamly we love to push the limits of personalized experiences by teaming creative with data in new ways — but we always stay conscious of the potential pitfalls too. Sounds interesting? We’d love to hear from you.