Radio data and the future of broadcasters: Using attribution analysis to measure consumer behavior

Video: How AI, the cloud, and Big Data are transforming the pharmaceutical industry

Radio stations have a business model traditionally based on advertising, and this is somewhat problematic. At a time when advertisers have many options in terms of digital media that can let them work out ad placement strategies and ROI based on troves of data, over the air radio seems antiquated in comparison.

Also: The digital transformation of advertising, from print to mobile devices TechRepublic

Is there a way for radio to catch up? This question may be fundamental for its survival, and could be restated as follows: is there a reliable way to measure the effectiveness of radio advertisement?

Attribution analysis

Attribution analysis is a quantitative method used to estimate the impact of certain actions by comparing data against a benchmark. It is mostly known as being applied to determine the performance of fund managers. TagStation, however, is applying similar techniques for radio advertising.

TagStation has developed a cloud service that provides radio station data to the NextRadio App and the Dial Report portal. TagStation just signed a deal with WideOrbit, a leader in premium broadcast technology, and ZDNet had a Q&A with Paul Brenner, President of Dial Report, to figure out how it all works.

Our first question had to do with whether there is any relationship between the attribution analysis used in funds and in radio stations. Brenner says that the concept is the same: both try to directly attribute a brand's radio spend or a fund's performance to the actual cause. The methods, however, are quite different:

Dial Report foot traffic attribution analysis looks at radio listeners who have heard a particular radio spot and analyzes offline behavior around the time of exposure to the ad. We compare the audience that heard the ad with a similar control group that never heard the spot. We also compare what the audience does before and after exposure to the ad.

The use of control groups sounds similar to techniques traditionally used to approximate advertisement effectiveness. But Dial Report also incorporates data collected directly from radio station logs, so we were curious to know what kind of data that might be and how they are used.

Integrating radio data

Brenner says that at Dial Report they use radio station playout data from TagStation direct integration at radio stations, radio station traffic logs, Ad-ID metadata about spots, radio consumption data from the NextRadio app, location data from NextRadio and other mobile apps, and place information to perform foot traffic attribution:

With radio station group consent, Dial Report can draw data via API from station logs in WideOrbit's WO Traffic platform which radio stations use to schedule radio spots. The more information we can learn about the spots playing the better, but we are not hindered in any way by the current quality of data.

More than 75 percent of U.S. TV stations are using WideOrbit software for traffic management today including many of the nation's largest radio station groups: Entercom, Cumulus and Townsquare Media.

WideOrbit and Dial Report worked together to determine which fields from WideOrbit's Traffic software were most relevant to Dial Report to enhance data about spots played. Dial Report uses WideOrbit logs to enhance what they already know about radio spots that are playing.

Brenner explains that integration with stations provides real-time information about songs and spots that are playing, but with a more limited set of metadata. So they match their own real-time play logs with WideOrbit traffic logs to fill in the dataset:

For example, we may know from the real-time feed that an ad for Home Depot aired. Using the log data from WideOrbit, we can add information about the spot - most importantly advertising IDs and product names. Instead of knowing simply that a Home Depot spot aired, with the additional information we can know which particular Home Depot spot aired (paint, lumber, garden, etc).

As far as processing goes, Brenner explains that many of their data points come in real time, but most data processing is done nightly into reporting warehouses. Attribution analysis is typically performed at the end of a radio campaign so it can analyze the effect of the entire radio spend.

It is a procedural approach and combines a number of data sources (direct integration at station, log consumption, Ad-ID metadata, radio consumption data from the NextRadio app, location data from mobile apps, etc).

Also: Facebook and Google, beware: Amazon is building a massive ad business

Volumes of location data can range from millions to tens of millions of records based on a given campaign period and market reach of those campaigns. That data is ingested in bulk loads, and records are matched with spots and exposure metadata.

The biggest challenge according to Brenner is combining and normalizing the various sources from legacy radio systems: "This is a big reason why we value the integration with WideOrbit -- it makes linking extended metadata to radio playout data simpler."

The future of broadcasters

Data collection for over-the-air radio helps to better identify spots, but how does the analytics that can be done with attribution analysis compare to what digital media can do? Brenner says:

Once you are able to measure exposure to an advertisement and tie "listen" to a particular user, there isn't much of a difference between doing attribution for web radio and broadcast radio. Dial Report is running our analysis on actual, measured listening tied to mobile users - just like any other mobile advertising. Historically, the challenge has been measurement.

Advertisers expect digital-like measurement and attribution reporting, which has been historically difficult for broadcast radio to provide. Dial Report is helping radio stations prove to their customers that broadcast radio works - and that helps them compete against any and all other media.

Speaking of other media, Brenner believes that similar methods could be used for TV stations, and adds that in TV, there are already measurement and attribution players:

Various companies are working in the TV sector -- SambaTV, Placed, iSpot.tv, Data+Math. Each company offers some specific form of attribution around TV viewing. The method used by TagStation is specific to integration with station-aired events plus consumer exposure, both combined and used to show all forms of attribution.

TagStation has a strong network of real-time data flowing in from almost 4,000 radio stations in the U.S. Our NextRadio app is the only internet-connected mobile FM broadcast radio receiver -- so our listening data is quite unique.

The combination of these two aspects of our business is the core of our attribution measurement capability. There is no other company currently that can measure radio this way and perform this type of analysis about radio consumption. Our methods for radio could be applied to TV.

But overall, how do digital media compare to traditional broadcasters? Brenner believes that over some amount of time, audiences are less and less conscious of how they received the content, with the exception of cost:

Over-the-air remains the most efficient distribution. While digital or internet-based distribution will change and grow, we don't think consumers at the end of the day will want to lose a truly free option for media.

We see radio audiences using over-the-air and digital in different scenarios. TV is also experiencing the same as their audiences access content throughout the day on different devices. It's also evident with the growth of cord-cutting and people moving back to antenna for certain uses like local programming.

Regardless of distribution, the buy-side expectations for ROI or ROA will continue to shift. And with that, broadcasters will need to account for all forms of distribution in attribution and planning models.

Internet of Things

Who really owns your Internet of Things data?

Read now