Bad Use of Data + Incompetent Bots = Changes to Marketing Spend

No one questions that big data has had a massive impact on the decisions of today’s marketers. But whether that impact is a good thing or a bad thing depends on your perspective.

For marketers and politicians who can better target their audiences, it’s a good thing. But for organizations trying to protect their brand image and reputation, the reliance on programmatic ad placements can be a very bad thing.

With programmatic advertising, an algorithm decides where to place ads, based on a person’s social media history as well as their demographic and psychographic data. This takes control out of the hands of brands. As it happens, those algorithms have been placing ads in some pretty weird places, specifically adjacent to fake news, hate speech, and other objectionable content. The result is unfavorable brand associations and, ultimately, advertisers pulling back from programmatic placements.

Before you turn major portions of your budget over to a computer, make sure there is a smart human available to check the results.

Witness Procter & Gamble which, despite putting pressure on Facebook and Google to fix the problem, continued to find its ads placed adjacent to undesirable fake news. Last month, P & G decided that persuasion wasn’t working and pulled most of its programmatic digital spending. Finance chief Jon Moeller said it had cut between $100m and $140m from its digital ad budget last quarter because of brand safety concerns and “ineffective” ads. “Clearly we don’t need to be spending money that is seen by a bot and not a person… or on ads that are placed in inappropriate places, and that’s why you see a significant reduction.”

What if $140 million in ads were cut and nobody noticed?

What’s really interesting is that the cut had no impact on P & G’s financial results, which beat forecasts as well as the results of its competitors. There was no loss in marketshare or sales after they pulled the ads. See a good discussion of why in Nigel Hollis’ post on the Kantar Millward Brown site.

It might also have had something to do with the fact that adjacency to objectionable material on YouTube and other digital platforms has resulted in consumer backlash for a number of brands. Earlier this year, YouTube took even more drastic measures, and eliminated the PewDiePie channel from all of its advertising platforms and canceled Scare PewDiePie Season 2.

SNCR in pursuit of fake news solutions

These anecdotes are perhaps an indicator of a trend, but, given that this is a special Measurement Advisor data issue, obviously what we’d like for this discussion is actual data. Which is why we are pleased that the Society for New Communications Research (@SNCR) and its corporate parent The Conference Board are tackling the topic head-on. They will survey their members to determine their perceptions of the fake news problem, the extent to which they might be rethinking programmatic advertising, and any solutions that they might have come up with.

We’ll be sharing the results when they’re available later this year, so stay tuned. In the meantime, my fellow SNCR advisory board member, Jeff Pundyk, former Senior VP Global Integrated Content Solutions for The Economist, sums up the initiative in “SNCR Takes On Fake News.” As Pundyck suggests, the data already shows that fake news in all its forms is eroding trust, the question is what are people doing about it?

Lessons about robots and money

There are two major lessons in all of of this. The first is one you’ve read in these pages before: Don’t trust algorithms! As good as an algorithm might be, the data it uses may well be bad. Which is why you need smart humans to check up on them. Computers and their calculations alone do not have the context and judgement to understand all the implications of data-driven decision making. Before you turn major portions of your budget over to a computer, make sure there is a smart human available to check the results. In Procter & Gamble’s case, they had two forms of smart humans. One that discovered the brand’s ads in the wrong places, the other in finance to see if they were actually bringing in revenue.

The second lesson concerns the use of data in decision making. P&G and others have pulled programmatic ads, not just due to potential negative impact on brand reputation, but also because they weren’t working financially. Again, the combination of smart humans and good use of data led to even better decision making. ∞

Thanks to DARPA’s video for the image above.

About Author

Katie Paine

I've been called The Queen Of Measurement, but I prefer Seshat, the Goddess.