http://feedproxy.google.com/~r/ad-exchange-news/~3/Gmg3pU8w1hA/

Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Nathan Brixius, vice president at 84.51°

Data and analytics are often used in dubious ways for dubious ends. Bob Hoffman and others have written about their frustrations of an industry that needs to start thinking and acting more purposefully.

Here’s a perfect example from a recent analysis that drove me nuts. After the Super Bowl, Salesforce used its social media marketing product to identify the most mentioned advertisers, trending hashtags and popular food choices. During the first half of the game, the top three brands mentioned across sponsored and organic content were Mexican avocados, Pepsi and Doritos.

But I believe ad tech loses credibility when it relies on superficial analyses of the wrong data. Ad tech matters when it is useful and takes advantage of data and computation to account for actual human behavior.

Response also matters. Salesforce’s study tells us “avocados from Mexico’s #GuacWorld were the most popular hashtag, gaining 83,132 mentions during the first two quarters.” Does that matter? The road between a hashtag and a sale (or increased customer lifetime value) is long and winding. Anyone involved in a marketing campaign should be squarely focused on the impact that the campaign has on its audience. The further away ad tech drifts from response measurement, the less reason we have to believe the claims it offers. Doubly so with channels like Twitter, where we know that as much of 50% of Twitter traffic is automated.

Measurement matters. Measuring mentions without measuring sentiment is inadvisable, because there is such a thing as bad press. People may be talking, but sometimes it’s not pretty. I have yet to see a response model that tells me that a tweet containing a barf emoji is a good thing. Discerning between types of online activity is important, too. Lumping sponsored and organic together is dangerous, because the former relates to spending and the latter to response. Giving advertisers credit for the number of sponsored mentions is like handing out medals for participation.

The long run matters. Who cares about the ephemeral? Can anyone really establish a relationship between the number of social media mentions in the first two quarters of a football game and the long-term health of a brand? Of course not. Does anyone care about #GuacWorld a week or two later? Look and see for yourself. There may never be a follow-up article that describes the impact of Super Bowl Twitter campaigns on creating and retaining brand-loyal customers because the results would be far too uncertain to be comfortable. That needs to change. The lack of studies on the long-term effects of advertising, particularly when advertisers increasingly emphasize customer lifetime value, is troubling.

Impact matters. The number of non-bot Twitter users in the US is somewhere between and 30 and 70 million, but only a fraction of them were actually using Twitter during the first half, and a fraction of those were exposed to Super Bowl–related advertising. I would guess that the total addressable audience was likely less than 10 million US adults. By contrast, 103 million people watched the Super Bowl in the US, all of whom saw ads whose messaging and resonance was far more impactful than 280 characters of text. When we speak of brands winning the online battle, we need to keep this context in mind.

I’m not saying advertising on Twitter, digital media or programmatic advertising is bad. There is enormous potential in targeting the audiences of interest in a cost-effective manner through automation and machine learning. We know the world is digital. The transformation isn’t coming; it’s happened.

The problem is that ad tech can lose its way when some in the field abandon science and the human element. We still need to measure what matters. We still need to understand our customers. We still need to write good copy. Advertisers can leverage what they know about their customers to sell them things they want, without annoying them.

When ad tech works for this end, it’s good, and when it doesn’t, it’s useless. Don’t be useless.

Follow Nathan Brixius (@natebrix), 84.51° (@8451group) and AdExchanger (@adexchanger) on Twitter.