Is targeting a form of discrimination?

Victor Condogeorges lays out the ethical issues posed by targeting and ponders if a robot would be any better at it. Spoiler: probably not.

When a campaign is planned or set up, there’s probably not that much thought given to the notion that it might be discriminating unintentionally.

But many targeting and optimisation methods may now be considered forms of discrimination, as we shift towards a more politically correct world. Here are a few examples of perspective checks to think about, with a little help from Road Runner.

Discrimination masqueraded as targeting

Imagine setting up a campaign for Acme Corporation’s e-commerce store, advertising dehydrated boulders. They’re on sale at the moment for 20% off. The target is hard to achieve, so to aid with marketing, the redemption code is shown in the banner.

Your team sets up the programmatic campaign in the system, does all the trafficking and are at the stage where they’re about to add targeting. They search for ‘Coyote’, logically, and when they find what they think is the most relevant segment and add it.

Great! Now you’re on your way to having a targeted programmatic campaign. Only targeting those that are relevant to the product and campaign – provided all the research has already been done as to which segments are actually the right ones to target.

This is the only channel running for this campaign as some idiot who thinks programmatic only is a good strategy planned it.

Stop, hammer time.

So let’s think about what we have on the table here.

  1. We have a product we’re selling, and it’s on sale now at 20%.
  2. This is an e-commerce only campaign thus it is not available in store, the redemption code is also only in the banner.
  3. The targeting is exclusive to Coyotes as everyone knows 3rd party data is 100% accurate
  4. The campaign is only being advertised on a channel that allows us to segregate the audience.

So what about the meep meep?

Now, considering that coyotes are being targeted exclusively with this campaign, that means roadrunners aren’t eligible to take advantage of the promotion. They don’t even know that it exists. If they visit the site, they pay the regular price because they haven’t seen the promotion or the coupon code that was in the millions of impressions going to all those coyotes out there.

You have probably ended up at the logical next question. Is this campaign discriminating against an audience or is it just a targeted campaign?

Let’s look at some real world examples that might be a bit more relatable.

  1. Advertising baby products. Do you target mothers or fathers?
  2. Advertising car tools. Do you target male users with an interest in cars?
  3. Advertising male grooming products, do you target men or women or both?

Excellent, now you should be in some sort of marketing existentialist crisis. Let’s move onto the next challenging thought experiment.

Optimise the creative

It turns out the marketing initiatives that exclusively targeted coyotes were picked up by the wrong crowd, resulting in Acme Corporation being lambasted and losing market share. The shareholders followed suit in the shadow of fear and nearly bankrupted the company – and just when you thought you were doing the right thing.

So you think to yourself, “For the next campaign, jet-propelled pogo sticks, we’ll target EVERYONE to avoid discrimination and simply optimise the creative instead. Make it more relevant to the intended audience then let the their interactions dictate the optimisations.”

The help of a highly awarded, heavyweight creative agency has been enlisted to this campaign, one by the name of POW.erpoint. Using POW.erpoint’s creative finesse, several variants of the creative are developed. One with the most commonly recognised coyote colour and others with different coloured coyotes. There are enough variants in your creative that it looks like an Andy Warhol painting.

Some coloured, some black and white, some green, some pink, etc. Just different coloured coyotes. The campaign is activated and overall it goes well. Majority of the audience don’t get offended, interact with the creative and everyone has a great time.

Back to the table. What do we have here?

  1. A tailored campaign featuring a diverse range of coyotes.
  2. A diverse creative set to garner adequate learnings.
  3. A personalised experience for the audience.

The question here is, did you do the right thing by tailoring your creative to the audience or were you discriminating? After all, personalised experiences are where it’s at, right?

Algorithmic discrimination

Unfortunately the last campaign landed Acme Corporation in some hot water. Again.

Personalising the creative ended up offending all the pink coyotes because ‘pink coyotes are under-represented in cartoons, and they never catch the roadrunner. Because they’re pink.’ or something along those lines.

You think about it and land on the conclusion that every time a human gets involved in the marketing optimisations or planning, the company takes a hit. So you say bugger it, I’m done with this shit! Except that there’s a new campaign to run and you’re a dedicated, passionate employee of Acme Corporation so you just get on with it.

This campaign is selling Tornado kits. The do-it yourself kind. Taking the learnings from previous campaigns and after being roasted for trying to do anything optimally from creative to targeting, you decide to let the damn machine do it all.

Technology at Acme Corporation is sophisticated, much like their products. Images can be generated from scratch. Because of this, and the learnings from the jet-propelled pogo stick, your team set up the campaign with no targeting and no creative. You add some machine learning magic spice and let the tech run itself. After millions of iterations, it turns out the technology, after going through every iteration possible, has found an optimal creative.

Acme Corporation’s Optimal Creative Optimisation Engine has landed on an image of a type of coyote that reflects the audience which delivers the greatest sales yield. The algorithm has found that by featuring green coyotes with long hair you will get the most sales. As a business representative responsible for maximising profit and achieving targets, you take advantage of this technology and let the machine do its thing – in turn, you get the most sales of DIY Tornado kits ever! But still manage to piss people off in the process.

In summary, some uncomfortable questions

If you set up a campaign with no parameters at all and the platform found these paths being optimal, would some audiences still be upset about your marketing initiatives? Who is responsible here? Society, the brand, the machine, or the person setting it up?

Is it ethical to use targeting in campaigns and deny an audience from seeing a message? Who determines which products or services are mandated for equal opportunity advertising?

Should creative be optimised to suit the audience and personalise their experience? What if an audience feels objectified, stereotyped, or discriminated against? Would it be different if there was a machine coming to that conclusion compared to a human coming to that same conclusion?

Machines aren’t irrational, we are. Some times we just need to understand that nothing can be perfect and that an audience will always be offended.

After all, if a machine comes to a conclusion that a specific race and gender sells kitchens better by being featured in the creative, can it be classified as being sexist and racist? Or are these just a reflection of our intrinsic biases as irrational humans?

Maybe we should look within and accept responsibility for behaviour before we search outward to blame everything else but ourselves.

Victor Condogeorges is a digital transformation expert.


Get the latest media and marketing industry news (and views) direct to your inbox.

Sign up to the free Mumbrella newsletter now.



Sign up to our free daily update to get the latest in media and marketing.