Google's Real-Time Final Four Ads Tried to Predict the Games' Outcomes. So How Did They Do?

A lot of companies like to flaunt their analytics after a big sporting event, showing how they crunched the numbers or analyzed the trends. But releasing your data in advance as a way to sell people on your software’s predictive abilities? Now that’s bold.

And that’s exactly what Google did this weekend with its ads that ran during the Final Four matchups on TBS.

Ads for Google Cloud ran before the Kansas-Villanova and Michigan-Loyola Chicago games, with a second set of ads running during halftime of each game.

While the commercials, created with San Francisco agency Eleven, didn’t predict the final scores, they did make data-educated guesses on stats such as rebounds and shot attempts. Making the real-time ads was a scramble, but did they pay off? Let’s take a look:

Kansas vs. Villanova:

Prediction: At least 26 assists.
Final tally: 28 assists

The prediction was off by 8 percent, but admirably close.

Prediction: 55 shot attempts in second half
Final tally: 64 shot attempts in second half

Off by 16 percent, this was Google Cloud’s least-accurate call.

While not too shabby, the Kansas-Villanova predictions were definitely outdone by Google Cloud’s prophetic commercials about the weekend’s other game:

Michigan vs. Loyola Chicago

Prediction: 37 three-point attempts
Final tally: 38 three-point attempts

Not bad! Only off by one, for a mere 3 percent difference miss.

Prediction: 29 rebounds in second half
Final tally: 29 rebounds in second half

Nailed it!

The spots were created live from the Alamodome by a team of Google data scientists and data analysts, who used a real-time rendering platform developed by Eleven and postproduction firm Cloneless.

Google hopes the ads will encourage businesses and nonprofits of all sizes to look into Google Cloud as a way of harnessing large amounts of data, such as, in this case, decades of NCAA records alongside real-time game stats.