Can the Cambridge Analytica story teach marketers something?
If you live in the UK or US, your prime time news might have been temporarily hijacked by the story of how a small company, “deviously” named Cambridge Analytica, “hacked” Facebook data, used “brain science” to manipulate it and then used that to target people with “fake news” in a widespread act of “psychological warfare” that resulted in all the wrong people getting elected.
That paragraph, of course, is a ridiculous escalation of what is currently being said BUT what is being said is a rather distasteful escalation of what actually happened.
I am quite jealous that a very objective and step by step account of what actually happened has already been put together HERE by Chris Kavanagh. The facts, as you might assume if you’re not prone to hyperbole, are rather pedestrian: a highly unregulated industry, people working against no clear standards, the pressure of accelerated growth, a careless (I choose to believe they were more careless than evil) management team. I could just as well be describing the early days of the tobacco industry. Head on over to Chris K’s blog to read the step-by-step account.
What concerns me, as a marketing professional, is his bullet point no. 4 from below:
“That Steve Bannon wanted to weaponize big data… No difficulty believing.
That Cambridge Analytica claims to be able to provide effective tools for psychological targeting and manipulation… Certainly true.
That Chris Wylie, himself, was involved with some shady business and views himself as partly responsible… Sure.
That the self-promotional claims of Cambridge Analytica actually equate to how effective the services they provide are… Hmmmm.”
What Chris is raising is a very valid point: did it work? If you listen to the reporting, one of the many implication is that Cambridge Analytica won some elections for its clients BECAUSE of the way they manipulated the data they were in possession of. And I think, from a marketers perspective, some things are interesting to note.
- Having data and combining it in interesting ways is not new and NOT the answer to everything.
This is what the whole field of media has been busy-ing itself with since we started having marketing, and the idea of targeting, as opposed to blanketing, was thought of. Today, the concept of delivering messages to everyone while ignoring the info you might know about them, is ridiculous. Targeting is an integral part of broadcast advertising (not just digital advertising). We’ve all read the occasional outranged article when a retargeted message gets a bit too detailed with the specifics and it feels a bit intrusive. Just like we’ve read the annoyed moan on social to having seen an ad for a pair of jeans you’ve already bought last week. The fact that you’re only seeing a few “creepy” ads and some badly retargeted ones is down to the amount of info advertisers have about you and how they choose to use it. But for targeting, we’d all see all the ads, all the time.
Did Cambridge Analytica have “special” types of data that they were combining? Not so much. Two very good articles have been written HERE and HERE, about how the scripts that scraped the data and clustered it were created and what type of information they relied. Chris Kavanagh also makes an essential point about internal Facebook info vs information people put on Facebook.
The data collected was not internal Facebook data. It was data that developers scraped from the profiles of people who downloaded their apps (and their friends)
If you read the two articles, you will see that a lot of the data used is what we would call “social listening”, scraping available public posts on all possible platforms and combining it with equally public information scraped through a 2013 Facebook API loophole.
So while they may have had some information that was legally unobtainable by a regular media agency, it was not the bulk of what they were using.
2. Successful targeting alone does not equal success.
There’s a clear distinction that we need to make between targeting a message, the message itself and then the two coming together. The questions to ask are plenty: was CA successful because of their targeting strategy? Does targeting ALONE guarantee the success of your campaign? If you serve a crap ad to the right people does that ALONE make them go and buy your product? Clearly not. So combining special data in special ways does not immediately result in success.
Having a smart data strategy is a component of successful campaigns but not the only factor.
3. We [kindof] know emotion works but there is no industry-wide data to say “personality-based” messaging does.
The other marketing-relevant intimation of the reporting is that CA used some nefarious, “psychological” messaging. It’s been suggested that they targeted messages written to appeal to the neurotic or the conscientious parts of the audience’s personality.
I used “psychological” in inverted commas because I am still trying to figure this claim out myself. When working with audience segmentations in marketing, we use demographic and psychographic traits to cluster people into groups. Psychographic traits refer to personality, values, beliefs, etc. CA seems to have used the Big 5 framework of personality traits to sub segment their core data and the big question is: does messaging based on that work to persuade people?
We know, from wide studies on brand communication, that emotional messaging works to some extent better in the upper part of the marketing funnel. And that, to my taste, is still way to “loose” a piece of knowledge. But I have yet to see a study that says comms based on people’s personality traits delivers the same type of certainty of success.
So, finally, the messaging strategy itself seems to have been built on a false, or at least, unverified premise.
What I’d hate to have happen after this is a complete misreading of what one can do with data, how much and where it plays a role in campaigns and how we need to craft messages to persuade. Targeting is not new, it is not a guarantee of success and messaging strategies can be built on way more solid theoretical basis. It is however, interesting to see how little is known about how digital media works and how little people understand about persuasion through comms.
PS: If you are less worried about marketing and more worried about your next vote, please read this article by Privacy International which includes a lot more info on that aspect of the story.