The following is a guest post from Ed Keller, CEO of Engagement Labs. Opinions are the author's own.

Super Bowl LIV is a month away and Fox announced that all ad inventory has been sold out for more than a month at prices "north of $5 million to as much as $5.6 million." How will advertisers know if they've achieved ROI on this significant investment?

USA Today has measured the popularity of Super Bowl commercials for more than three decades with its Ad Meter system. It has become the most widely used indicator of advertising success in the big game — but it doesn't work, and it is destined to fail again next month.

Ad Meter invites thousands of football fans to evaluate the quality of each ad on a scale of one to 10 and publishes the results. In the current issue of the peer-reviewed Journal of Advertising Research, two colleagues and I compared five metrics of Super Bowl advertiser performance and discovered that Ad Meter is a clear outlier. It is the one metric totally unrelated to the others.

In last year's game, Bud Light launched a memorable series of ads thematically tied into the imminent finale series of "Game of Thrones," showing medieval characters at a jousting match and discussing which beer brands contain corn syrup.

Ad Meter proclaimed three spots for Bud Light to be ranked 16th, 17th, and 43rd among 58 measured. And yet among 23 Super Bowl advertiser brands, Bud Light generated the single biggest increase in Google Search activity and the fifth largest increase in YouTube views, according to an evaluation by Engagement Labs. Bud Light also performed well on two other metrics we measured — online and offline conversations. It saw the biggest increase in social media mentions and the fourth biggest increase in offline conversations. By all indications, Bud Light won Super Bowl 2019, but its success was completely missed by Ad Meter.

In fact, Ad Meter's results for all the advertisers were negatively correlated with each of the other four metrics studied — meaning the better the Ad Meter rating, the worse its performance on other metrics. In addition to Bud Light, brands such as Doritos, Pepsi, Olay, and TurboTax did much better on every other metric than they did for Ad Meter. At the opposite end, Amazon, Hulu, Microsoft, and Verizon all performed far better on Ad Meter than on the other metrics.

While the Super Bowl is largely perceived to be a "buzz generating" event, the study also shows that online performance is key, but is insufficient on its own. With their heavy focus on digital amplification, most advertisers see growth in online posts around the big game. However, as the most social TV viewing event of the year, water cooler buzz is also huge and not well reflected by what's happening online.

When both online and offline are factored in, the outcome is more mixed. Only seven of the advertisers enjoyed lifts of at least 10% in both online and offline conversations. Driving up social media only without a corresponding increase in offline conversation is like kicking a field goal instead of scoring a touchdown. Of these seven brands that achieved both, only two (Kia and the NFL) scored among the Ad Meter elite.

Why does the Ad Meter system fail to predict success as measured by the other metrics?

• Targeting matters. Although the Super Bowl draws a broad audience, many advertisers are focused on just a portion of the audience, perhaps just men (Budweiser), women (Olay) or young adults (Doritos). It doesn't matter if the ads earn poor marks from people outside those targets, so long as the respective target market is engaged.

• Branding matters. Some popular ads fail to generate a strong response for the brand because the narrative dominates at the brand's expense, as what likely occurred when Harrison Ford's indulgent pet dog used Alexa to order excessive amounts of food. Entertaining, for sure, but was there a clear brand message?

• Creative matters. To be sure, the quality and execution of a creative idea matters, provided it's used in tandem with other strategic imperatives. In 2019, Bud Light tapped into the cultural zeitgeist by partnering with "Game of Thrones," the most talked about TV program in more than a decade.

• Behavior matters. Self-reported opinions — such as evaluations of Super Bowl commercials — are generally less reliable as indicators than actual consumer behavior. YouTube views, Google searches, social media mentions and offline conversations are all behavioral metrics, whereas Ad Meter reflects opinions.

• Outcomes matter. Being popular is not always crucial to victory, as we've seen lately in politics. Last year, TurboTax ran a creepy commercial about the value having a real person — rather than a robot — on the other end of a customer support line. The spot drove a bigger lift in offline conversations about the brand than observed for any other advertiser, yet it ranked nearly last in Ad Meter's popularity contest.

More important than deciding who wins and loses the Super Bowl LIV ad war, though, are the lessons for what makes in-game ads effective. To win on Super Sunday, advertisers should concentrate less on the popularity of their ads and instead be laser-focused on driving engagement of their target audiences — both digitally and face-to-face in America's family rooms and sports bars.

Here's hoping for an exciting game and engaging ads. To the victor go the spoils!