Last week, Marketing Week columnist Mark Ritson set a challenge to the marketing industry: to prove whether the final result of the UK’s referendum on leaving the European Union was decisively influenced by Vote Leave’s illegal overspend.
The official Leave campaign spent £449,079.34 more than the legal limit of £7m, due to its donation of £675,000 to a second campaign, BeLeave. Both were fined by the Electoral Commission after the regulator found they acted under a “common plan”.
While Ritson complemented the effectiveness of Leave’s emotionally driven message and its finely honed targeting, he asked marketers to determine how crucial the rule breach was: “Did the illegal 6.4% overspend enable Vote Leave to win the day? What marketers – or rather, what a particular marketer with advanced analytics training and nothing better to do for the next few days – need to do urgently is answer this question.”
Here are the responses of those who took up the gauntlet.
Andrew Willshire, founder, Diametrical
I believe that any attempt to “prove” the impact of the overspend is doomed to failure.
1. It is wrong for Ritson to say “the [marketing] campaign changed people’s opinion”. More accurately, people’s opinion changed during the period that the campaign was running. We know from other elections that most people don’t really tune in until close to the vote, so the poll movement occurred as might have been expected irrespective of the marketing campaigns.
Polls also shift radically due to more important factors than marketing and we have no idea if the people targeted were the ones who changed their minds. Furthermore, the simple in/out question is too simplistic. There has been a majority for either leaving or staying in but reducing the EU’s powers since at least 1992. As reducing powers wasn’t an option, leaving might always have been the most likely outcome.
2. We don’t have robust data. There is a single accurate measurement – the referendum. The rest of any putative time series comes from opinion polling, which is noisy and impossible to calibrate against a ‘true’ value.
The 2015 election polling inquest identified that samples were skewed and incorrectly weighted – but companies couldn’t know that until after the election, nor whether their adjustments were right until after the subsequent election in 2017. We also lack access to accurate media data and measurements of the impact of major campaign events, news coverage or, most critically, social influence.
3. Modellers also need to calibrate their work. It’s one thing to estimate how many packets of cornflakes were sold due to advertising when you’ve built dozens of similar models in the past on robust data, but quite another to model a one-off event with questionable data and many overlapping and conflicting hypotheses.
Ultimately it comes down to whether just over £449,000 of targeted advertising can sway 634,751 people – 71p of advertising each – to vote the other way on a matter of the utmost national and personal importance. There’s no way of knowing for sure, but I doubt it.
George Gloyn, associate director of business science, Mediacom
From an econometrics perspective, if we purely look at benchmarks rather than actual modelling, I don’t think the overspend would have been the deciding factor.
Across the brands I’ve looked at, the amount of sales driven by media can range from 5% to 40% dependent on the category. For the sake of stress testing, let’s assume that voting is even more sensitive and that 50% of the votes for Leave were driven by the spend. So 8.7 million votes were media-driven. If we reduce this by the percentage of overspend, we would see a reduction of approximately 500,000 votes.
Let’s be generous and assume that all these votes would still have existed and gone across to Remain rather than not happened. This would leave us with 16.9 million votes for Leave and 16.6 million for Remain – a split of 50.4% to 49.6%, a knife edge vote for Leave.
So even with a number of assumptions that definitely look favorably on the impact of media – and not even taking into account any impact of diminishing returns (a reduction in spend of 6.4% wouldn’t necessarily lead to a reduction in impact of 6.4%) – it is potentially far more down to the effectiveness of the Leave campaign and lack of effectiveness of Remain activity as Ritson discussed, than anything to do with 6.4% more spend.
A fair challenge would be that we would obviously have a very different result if media contributed more than 50% of the vote, but given the stability of voting intention in 2015 and the beginning of 2016, I believe this would be unlikely.
Billy Ryan, global marketing effectiveness manager, HSBC
If you look at diminishing returns versus the campaign limit of £7m, it’s unlikely that the Vote Leave campaign’s overspend of £449,000 tipped the scales in a 52% to 48% vote.
A £7m campaign over nine weeks – with a media buy focused on digital display and Facebook – is heavyweight by any standards. Notwithstanding wall-to-wall national media coverage of the Leave message and considerable owned media exposure via the fake news real estate, the paid element of the campaign would have reached saturation point well before the additional £449,000 was spent.
Even if we assume Vote Leave built a marketable universe as large as two thirds of the voting population that are active on Facebook, the one billion ad impressions claimed by campaign director Dominic Cummings would imply 100% coverage at an average frequency of 40 impressions per person. It’s unlikely that it was the last 6% of these impressions alone that drove the Leave campaign to victory.
The success of the Leave campaign was a simple case of better pound-for-pound marketing effectiveness. A more emotive message to ‘Take Back Control’ was delivered at scale via the earned media coverage guaranteed by electoral law. And this message was amplified to carefully curated and identifiably susceptible sub-groups via tailored content in paid and owned channels.