I guess I represent all the axes of evil on this subject – I’m a qualified Conservative Party Agent, a professional marketer specialising in direct marketing and a politician.
I’m going to tell you that big data didn’t win it. That Cambridge Analytica is selling a bog-standard set of analysis tools rebadged as ‘political’.
Database marketing is brilliant. We can analyse large data sets and apply them to useable population geography – in the UK postcode sectors contain about 2,500 to 3,000 people.
When Cambridge Analytica tells you they have a profile for every community in the USA, they are simply describing the use of these well-established and widely used geodemographics.
It’s what they sell – it’s what we were selling to UK financial services companies, mail order business and retailers back in 1990.
Better still, we can then match this profiling information to our own data – in the case of a political campaign this is likely to be voting intention information: VI Data.
What this match will tell us is where our voters live and, more importantly, where voters with similar characteristics to our voters live. We might have lots of VI Data for Tory voters across the country but none in Walsall – what the profile does is tell us, at a very localised geography, where we’re most likely to find those voters in Walsall.
This is pretty much all that Cambridge Analytica do.
The big change Cambridge Analytica offer from base geodemographics is the use of social media information.
Overlaying this data onto that geodemographics perhaps (I’m not convinced) makes that granular geographical targeting more effective.
The bit of software Cambridge Analytica uses claims to be able to draw a phychographic profile from facebook likes and text – I’m not convinced (especially since most folks activity on face book involves saying LOL a lot, going aww at cat pictures and liking your friends baby photos).
Here’s an example of the problem from a review in technology blog, The Register:
A version of this test is online here, with another that analyses language. The first 1,000 words of Conservative Prime Minister Theresa May's Brexit speech from January generates a 67 per cent openness rating, making her "liberal and artistic" rather than "conservative and traditional", and a 99 per cent score for her being a man.The truth is that the data from social media is very dirty – what these clever techie folk claim is a long way from the reality. And, regardless of quantity, having rubbish data will always give you rubbish outcomes.
What Facebook does, however, is allow us to target advertising at that same granular level – I only need to buy advertising within a few miles of high profile score communities in target areas making the advertising more efficient and perhaps more effective. It also means that if you are outside my profile you won’t see much of my advertising.
What we have here isn’t a sinister conspiracy but just a set of marketing tools applied to politics. Does it work? Yes – but it’s not a silver bullet. We’d reckon on uplift in response of around 2X or maybe 3X compared to a random selection. Great until you realise that the response to random was around 0.2% - all that clever technology means that, instead of getting ignored by 998 out of 1000 people, you only get ignored by 994.
In every other respect this is little different from what we were doing nearly thirty years ago – digging into data to make our marketing better.
And it was not the main reason that Vote Leave and Donald Trump won. I see five things as mattering in effective political campaigns, only one of which is targeting.
1. Brand. Vote Leave got lucky because some journalist coined the term ‘Brexit’ giving a simple and memorable brand for the campaign. One that opponents of leaving repeated again and again. The same goes for Hillary Clinton – almost all Democrat campaign material talked about Trump – and Theresa May where the Conservatives pumped out millions of messages saying ‘May or Corbyn’
2. Call to Action. Again Vote Leave got lucky – “take back control” they said, giving people a simple action, voting to leave, that would make that possible. “Drain the swamp” said Trump. And the same goes for Corbyn (find quote)
3. Enhanced Word-of-Mouth – or ‘going viral’. Thousands of little green frogs – with all the memes and comments that went with those frogs - were probably more important to Trump’s campaign than the money he spent with Cambridge Analytica. And the same was true of Vote Leave and Corbyn – thousands of people creating and sharing memes, doing things unasked and without the campaign’s say so
4. Good targeting – yes it matters and using geography is a great way to do it in a system based on people voting locally. But, as Corbyn showed, targeting a demographic – young people – can also be pretty effective.
5. Finally Brexit and Trump won because their opponents ran dreadful campaigns. Theresa May did too but started too far ahead for Corbyn to catch her.
Yes targeting matters. Absolutely the generation of ‘fake news’ matters. But no, ‘Big Data’ didn’t win these elections. There is no sinister conspiracy, just a set of circumstances – good branding, viral social media, strong calls to action and awful opposing campaigns – that allowed the results we got. Targeting helps but, if we’d not seen the brand, message and memes – and the useless campaigns from Stronger In and Hillary Clinton, the results would have been different.