A WIN FOR HUMANITY OVER AI.

Last night, in one of the most incredible nights of football ever, the San Francisco 49ers edged out the Detroit Lions to go to their second Super Bowl in four years. One of the main catalysts for the comeback, were three 4th and short situations where the Lions had the opportunity to either kick a field goal, or “go for it” to get a new set of downs, score a touchdown and potentially kill the game.

·      The first time, they kicked a field goal, made it, and went up by three scores.

·      The second time, they went for it, and the 49ers stopped them.

·      The third time, with an opportunity to tie the game, they went for it, and the 49ers stopped them.

Although I’m a 49ers fan, the game itself took a back seat for a split second to thoughts of my career and the role that AI and Data plays in today’s marketing and advertising world.

The Lions are known as a deeply analytical team, and they went for it on fourth down 34% of the time during the regular season, the highest rate of any team this century.

In this particular game, the first field-goal was kicked to make the game a three-score game, when the Lions were running THROUGH the 49ers. They couldn’t stop them. But the Lions made the decision to go up three scores.

In the second opportunity they were again up by 14th, but for some inexplicable reason, they didn’t make it a three-score game, and went for it. Surely, the data showed a significant increase of success by going for it? .3%

In the final opportunity, when the team had the opportunity to TIE THE GAME, they once again went for it, instead of kicking a field goal to tie it. Surely, by this time the percentages swayed in their favor? .2%

Momentum and humanity resulted in the Lions using BAD DATA. The .3% and .2% recommendations came from completely different environments (full season) from completely different teams (the 49ers are an elite team vs. some regular season teams that didn’t even make the playoffs.) Also, to completely ignore how the players playing the game were reacting and the way the game was playing out is a disservice to the model. 

In the marketing/media/advertising world, AI models are being built and developed to replace everything from planning & buying to creative. The move towards AI and machine learning makes sense from the POV that if you’re a brand spending hundreds of millions of dollars on advertising, you want to limit human mistakes, push creative towards the right targets, and you certainly want to cover your @$$ when you’re presenting your QBR to leadership and/or the board.

You should certainly have a strong data and analytics foundation for all your strategy development. However…humanity, context, computer error, and in the world of advertising…discovery, always need to be considered in the decision making process.

Marketers love programmatic, but in a recent SSRN paper showed that the technology doesn’t work. The programmatic platforms use third-party data, which is unverified, and when cross-referenced against actual audience data, they were able to define gender with only approximately 50% accuracy. When you layer against an additional attribute like age, the accuracy drops to 24%.

 Sources: https://papers.ssrn.com/sol3/Data_Integrity_Notice.cfm?abid=3203131

All of a sudden, the security blanket of programmatic data doesn’t feel so good anymore. And guess what, it costs significantly more to target in such ways.

So does this mean that marketers should forsake targeted media? Absolutely not. But prioritize first-party data, and layer in broad targeting media options as well. Broadly targeted video, out of home, and audio are specifically strong options to reach more audience and implement discovery at the top layer of the omnichannel experience.

Finally, creative. This hasn’t reached an AI tipping point yet. The technology is not quite there yet. However, it’s all coming, and it will be important for brands to bring humanity to the work. Because despite the best prompts, the work that resonates will have human emotion and even ::gasp:: some imperfections built into it.

Until then, thank you to the 49ers for winning a battle against the robots. And in two weeks? The real challenge…taking out the Swifties.