A version of this article was published in Advertising Week.
Since the birth of programmatic, laser-focused audience targeting has been the name of the game. But as the cliché goes, with great power comes responsibility. And from environmental impacts to user privacy, we are arguably only now starting to take full account of that. Meanwhile, further potential side effects of people-based targeting are still coming to light.
Movements such as #metoo have shone a light on discrimination and misconduct across several industries – advertising among them. There have also been moves to clamp down on lingering stereotypes in creative, most prominently by the ASA. But what about planning and buying? Have we seen the same scrutiny around audience assumptions, many of which date back for decades? And have machine learning and algorithms exacerbated the problem, or are they simply a reflection of the status quo, in a shiny new context?
Facebook Ads and ‘Legal Skirmishes’
It is perhaps less well known in the UK than the US, but stateside, discrimination cases around Facebook ads date back to 2016 – specifically around its jobs and properties offerings. In that year, ProPublica revealed that it allowed advertisers to exclude people from such ads, based on “gender, age, or other protected characteristics.” It was only in 2019, and “several legal skirmishes” later that this feature was removed.
In 2021 NGO Global Witness research found that 96% of impressions for a UK Facebook ad for mechanics were served to men, 95% of one for nurses to women. Employment lawyer Schona Jolly QC stated a “strong suspicion that Facebook has acted, and continues to act, in violation of the Equality Act 2010.” Similar research from the University of Southern California found the same bias, though this was not reflected in campaigns run on LinkedIn. Which puts the lie to the argument that technology and automation are the root of the problem here.
In yet another example, just a few weeks ago Northeastern’s Khoury College of Computer Sciences found even more issues with how Facebook ads are algorithmically targeted, depending on people depicted in images used in the creative. Including the following: “ads featuring young women were delivered disproportionately to older men.” Clearly, this problem is not going away anytime soon.
30% Less Impact Among Women
However, before we assume that the problem of gender discrimination in media buying is limited to tech giants alone, we should have a good look in the wider display and agency mirror.
In agency land, the newest research on this topic is already a few years old – which tells a story in itself. But from what we do know, Kantar suggests that, despite both genders being equal ‘main buyer’ for households across most product categories, almost all (99%) of UK ads for laundry products are targeted at women, while 70% of ads for toiletries and food products are aimed at women.
Not only does this betray obviously outdated views – as Kantar points out, they’re also just simply inaccurate. Such decisions also unavoidably and unquestionably also mean less success with consumers. Little surprise then that in the year the research was produced, ads were said to generate almost 30% less impact among women than men.
On a wider level, gender as a targeting proxy is rarely a true representation of someone’s spending habits. This is truer now than ever: for example, Harry Styles has launched gender-neutral make-up and nail varnish brand, Pleasing. And Brighton and Hove Albion experienced a 55% boom in girls signing up for football sessions following the Lionesses’ Euros success.
In short, preconceived ideas of gendered purchasing habits have never been less useful for marketers. Kantar says that ‘gender balanced’ brands are worth more than female-skewed brands, and much more than male-skewed ones. Also, for those still unconvinced, we have the practical limitations on gender- and other types of demographic targeting, via cookies.
As most know at this point, structural change is the order of the day, with an estimated 40-60% of all impressions no longer targetable, for the simple reason they appear on Apple devices. To add to this, MIT research has shown that cookies reportedly only recognise user gender correctly 50% of the time. So, even in the face of all the above evidence, if you still believe gender targeting based on unproven audience assumptions has a role to play, you must also accept the tools to enact it are no longer fit for purpose.
Of course, beyond the cookie itself, there is an abundance of cookie-like post-cookie solutions to choose from – whether based on email addresses, fingerprinting or other forms of profiling. But even they, for all their innovation, fail to address the elephant in the room – just as audience targeting and the advertising and agency world at large have also failed to do.
The very act of binary gender targeting is hopelessly out of date. Will the ID solutions and CDPs ever offer a ‘they/them’ audience attribute alongside he and she? If they aren’t already, it’s probably a fair sign they will never catch up with the present.
Right there is the killer argument – if more were needed – were gender targeting to disappear completely tomorrow, none of us should mourn. We would be better off as an industry.