Technology means we’re getting ever more personal with our customers. Or so we think… Helene Dancer asks: Do too many algorithms mean we’re actually beginning to lose the human touch?

The advert for your favorite band’s new album is following you around the internet like an overenthusiastic fan backstage at a concert. Wherever you click, it appears, along with ads for a new t-shirt range from a brand you bought clothes from last week. It’s like someone’s been watching you… which is precisely what’s going on here. 

For, hard at work behind the scenes, sets of algorithms decide which adverts are best suited to you. These algorithms are powered by data sets detailing your behavior – where you shop, what music you listen to, where you live, and even your political preferences. All this information is collected and then used to recommend products and services deemed relevant to you. 

It may sound clever, but this trend of hyper-personalized marketing has divided opinion. Many welcome the approach with open arms – research suggests that 75% of what people watch on Netflix and 35% of purchases on Amazon comes as a result of product recommendations based on algorithms. But others are decrying the lack of human touch in the equation. Can it be that the more ‘personal’ we’re getting with customers, the less we really know about them? 

Up close and personal

“The human element is always going to be there because humans make the decisions,” reckons freelance social media campaign manager Tom Paisley. “I feel like people still rely on their gut instinct. This guides research and decisions. It has always been at the core of what marketing is, regardless of how data-driven things have become.”

In his experience, too much personalization can make people uncomfortable, unless the consumer absolutely trusts the brand. Paisley recently worked on a retargeting campaign that involved giving people options for further information based on the city in which they lived. He found many responded negatively to having their locations in the public domain.

“It’s about finding a balance. Test the water first – see what type of response you get, and take it from there.”

“We didn’t even go that granular and people found it too creepy,” he says. “You can go postcode specific if you want to, so it’s about finding a balance. If Amazon were to tell you that eight people on your street bought a particular item, so perhaps you want it too, people would find that disturbing. Test the water first – see what type of response you get, and take it from there.”

But some platforms, like Spotify, base their entire operations on recommendations according to what people are consuming. So how do they manage not to freak out their customers? For Warren Ali, engineering lead at digital product design agency, Ostmodern, it’s because you know what your data is being used for. “You’re giving them your preferences and your likes but it’s in a very limited way – and you get something in return,” he says.

“Part of the reason Spotify is cheap is because they make their product based on everyone’s data – their recommendations. They monetize the data but they’re ploughing it back into the platform. It’s pretty symmetrical. You know what’s happening with the data.”

Facebook, on the other hand, isn’t as transparent, as the recent Cambridge Analytica furore shows. “Facebook builds up a pretty detailed picture about you and you don’t know what they’re doing with that picture,” says Ali. Of course, the targeted advertising on the platform gives some clue as to what the collected data is being used for – but he cites the example of Cambridge Analytica targeting black people to dissuade them from voting for Hillary Clinton in the US elections as a more insidious example. “By people not understanding what Facebook is doing with their data, they can’t make an informed decision.”

Much like Spotify, Netflix has also traditionally worked off recommendations, but they’re moving away from this model because, according to Ali, they’ve got to the edge of how hard algorithms can work for them. Netflix used to run the Netflix Prize, a competition for programmers to build the best collaborative filtering algorithm to predict user ratings for films. “People would try to beat the Netflix algorithms but if they did, it wouldn’t be by much,” he says. The company shelved the competition and is instead now focusing on pushing its own Netflix-produced films and series.

Changing the game

Technology is also changing the world of traditional market research from within. Magdalena Depta is client operations manager at Lightspeed, a digital data collection agency that’s part of market research giant Kantar. “Before we only relied on the survey data but now we’re focused on the data side a lot more,” she explains. “If you think about consumer and old-school marketing, we’d only rely on the survey and what a consumer tells us. Whereas now you have access to their behavior and attitudes so you have a much richer understanding of who they are and what they like.”

For Magda, any backlash towards hyper-targeted advertising comes from a generational difference. “I grew up thinking of myself as a private entity and it feels difficult to share, but younger generations growing up in a social media world find it much easier to share.”

Convenience is also a major factor that sweetens the pill. “If you were to ask me six years ago if I would ever consider doing my shopping online, I’d say no, never. But now I only buy online because of the convenience. I don’t need to go into a shop and have the human experience,” she says. “As a consumer I’m bombarded with ads anyway, regardless of the algorithms. Now they’re just more relevant to me – and more likely to tempt me.”

She argues that the algorithms may not be able to predict behavior – yet – but can certainly apply an enormous amount of influence. “They can understand what consumers like and what their triggers are, and then influence them,” says Depta.

So you’ll certainly be tempted – but likely by more of the same. “Algorithms are successful when assuming that people want things they’ve historically always wanted,” says Ali. “But this means you’ll buy less diverse things. The recommendations are very surface – not substantive.”

“Algorithms are adept at building echo chambers that rely on the most obvious features evident in the data on which they’re modeled. Data deemed irrelevant will be discarded, which means a far more homogenous vision.”

More of the same gives rise to filter bubbles that reinforce existing patterns. Algorithms are very adept at building these echo chambers that rely on only the most obvious features evident in the data on which they’re modeled. Data deemed irrelevant will be discarded, which means a far more homogenous vision. So if you only listen to hip-hop, then you’ll only be recommended hip-hop if that’s the most obvious selling point. 

“You’re being monitored and being manipulated – removed from another way of seeing things,” says Depta. “We just assume because we’re surrounded by likeminded people.”

It beings to feel a little nefarious when you consider the amount of data platforms like Facebook are collecting and crunching, and then consider what types of behavior are being influenced. Which is why, for Paisley, transparency is essential in this current state of affairs. “Brands need to be open with what they’re doing with people’s data. Transparency definitely brings people on side.”

For Ali, it’s an awareness of what the internet has become. “Everything on the internet is intrinsically linked to advertising,” he says. “Advertising is becoming the main economic model for how you do things on the internet, which was set up to be a utopian shared space where information was supposed to be free.” 

Don’t believe the hype

So how can brands make use of the technology available to them, and maintain the human touch? “I may recommend you a book that isn’t like anything you normally read, but it’s because I know you personally, and I know you’ll like it. If you believe the hype, the internet would’ve killed the high street,” says Ali. ‘“The human touch is really important.”

It’s for this reason brands are increasingly folding an element of surprise into their campaigns – one-off emails offering discount codes to loyal customers, for example, so they’re not just relying on hyper-targeted ads to make sales. Exclusive clothing boutique You Must Create sends its ‘favored’ customers surprise discount codes – but only on the odd occasion, so the brand maintains its premium sheen.

Other companies sponsor events, such as gigs, festivals and exhibitions, to facilitate a meeting of likeminds within a physical context. Barcelona’s Estrella Damm positioned itself wisely as the main sponsor of the city’s multimedia and experimental music festival, Sonar, and BP continues to support art exhibitions around the world, although this hasn’t always been welcome.

It’s a balancing act and one that needs to be handled carefully in order to maintain trust with consumers. “The technology is not going to be retracted,” says Paisley. “And the future will be more of the same. To throw all your eggs in one basket of specifically trying to predict what people want to buy is missing the point. People just want to trust that your brand sells good things.”

Ready to rock?

Fancy a chat about this? Or maybe we can help with our advisory services or get in touch!


Join hundreds of your peers and get your weekly fix of marketing street knowledge from the Rockstar CMO community.

Share this article