HomeInsightsThe Problem With...
Topical

The Problem With Personalisation And Programmatic Display Advertising

31st May 2021

If you’ve ever Googled something off hand and suddenly seen adverts for the same thing on Facebook, or noticed your Amazon searches influencing your recommendations, you’ve seen personalisation in action. The tracking of ‘cookies’ – tiny data files that log the websites you visit – is big business for marketers.

The worlds of personalised ads and programmatic see companies doing constant battle to earmark the perfect customer, and find them the perfect product. Companies are convinced that by distilling down the essence of someone’s browsing and shopping habits, they can appeal to them on a completely different level than picking a broad age or gender demographic.

But is this really true? To twist an old maxim, do people actually want what they know? Or is it advertising’s role to curate, tease and suggest products that the viewer might not ordinarily buy?

Knowing Me, Knowing You

Personalisation in the 21st century is driven by algorithms. Riding on the future tech coattails of AI, algorithms are complex calculations that allow computers to analyse user data, and apply appropriate solutions. The ability to collect and parse large datasets (‘big data’) with modern computers has allowed algorithms to be honed, providing much more accurate results.

The standard bearers for this approach are the driving force behind today’s most popular media software. Netflix bases its much-vaunted recommendations on algorithms which track not just what you watch, but how you watch it; assessing when you pause or skip, and the links between different shows. Spotify meanwhile has its Discover playlists, which automatically curate a weekly selection of music similar to the other things you listen to (with a few outliers thrown in).

These serve a noble purpose, and it’s likely they will only get better. They essentially make the quality of the service its selling point, which is never a bad thing. That’s particularly true in the case of Spotify, which seems to deliberately throw in some left-field choices from time to time.

But there’s also a danger that these recommendations become homogenised. They may start to only offer products that they know with absolute certainty you will like, rather than a range of things you might like. And that’s inarguably an issue.

Consider how TV used to work (and still does, in some cases). With a limited number of channels, there was also a limited number of shows to watch. You would often sit through programmes or movies you might not otherwise watch, simply because there was little else on offer. When you found out that you liked them, that could change your tastes forever. Giving people a chance to like new things was a way to expand beyond their segmented audience.

Personality > personalisation

This might not seem like a bad thing for the likes of Netflix. After all, sticking to what people know reduces the risk of annoying or offending them with poor recommendations. But at the same time, Netflix produces a huge array of diverse content. Spotify provides a broad library. If they could get more people to consume more of it, their platform would hold even more value.

It’s a tricky proposition, and one they are still navigating. But it highlights the issue with going overboard on personalisation. Yes, people may rather see adverts for the thrash metal they actually like, rather than the new Ed Sheeran album (nothing personal, Ed). And for many people, music and movies are only an occasional distraction, not something they feel the need to experiment in. But an equal number of people tend towards new and unique experiences.

We see this daily in the world of Instagram and Snapchat, platforms dedicated to documenting and sharing experiences: new food, new locations, new clothes. We even see it in the retro revival, and the increasing trend away from digital towards physical media, like vinyl. Part of the value of listening to new artists, seeing new movies or showing off your collection is the ‘hipster’ exclusivity; the ability to share your love with like-minded friends and fans.

This applies pretty equally to advertising across the board. From food to clothing to household goods, owning or liking something that few people know about can be a major drive in purchasing decisions. Some advertising will always be situational: an advert for a recognisable product on a banner ad may be more likely to be clicked, simply because the clickthrough rate is generally poor. But on the basis of history, there’s more value in exclusivity and the marketing of niches than there is in narrowing people’s cultural palette.

Be different

This doesn’t mean personalisation has no value. It’s clearly desirable in a lot of circumstances, and Spotify has shown that it can achieve a balance, where an element of unpredictability is introduced. But we shouldn’t lose sight of the potency of a unique product.

There’s a danger that the more people’s tastes are honed in on, the more homogenous content and products will become. This is arguably already happening with the music industry: radio setlists in the US are notoriously limited, with the effect that most people only hear the chart hits. This has arguably impacted the presence of alternative genres in the mainstream, potentially locking whole genres out of significant sales.

This is especially an issue in ecosystems which essentially monopolise this form of content delivery. Netflix has been extremely good for content creators so far, but it would not be surprising if in the face of future competition, they choose to hone in on their most profitable customers. You can see the beginnings of them reigning in resources in the recent spate of cancellations, such as the popular but divisive Wachowski vehicle Sense8.

It’s scarcely a trap for Netflix; they’ve exploited the demand for personalisation and unique content, and if they bring that to an end, it will be a wise business decision. But businesses of all stripes should be wary of following the same model. Not only is personalisation incredibly tricky – relying on algorithms for digital products, or painstaking work for physical ones – it’s also not necessary.

The traditional way to personalise is not to build a product that hones itself over time, but to create a bunch of great ones. Your product is unique up to the point when it becomes popular. Personalisation, above all else, is a way to make the most widely used services feel intimate. For everyone else, the ultimate angle on a fresh product is that not many people own it.

Share
Nick Huxsted
31st May 2021

Ready to get the conversation started?