Ryan Steelberg

Data, analytics and attribution are increasingly becoming table stakes in today’s advertising world, as marketers look to their media partners for empirical evidence that their campaigns delivered quantifiable results. That starts with affidavits and attribution, which Veritone’s artificial intelligence platform now provides to a growing number of radio groups, including Beasley, Bonneville, Cumulus, Entercom, iHeartMedia and Townsquare.

Veritone’s aiWARE platform ingests, indexes and packages audio and video clips, which can then be analyzed by both broadcasters and advertisers looking to measure interactions and exposures. Inside Radio spoke with Veritone president and co-founder Ryan Steelberg  about how radio stations are using this technology, how it can help them exploit the exploding podcast opportunity, and how repackaging and redistributing their content can deliver a more personalized, interactive listening experience on smart speakers. An edited transcript follows.

Veritone works with many of radio’s largest companies. What are some of the most popular ways your clients are using the aiWARE platform to grow sales?  

We now have 500 stations that are live on aiWARE. For the majority, one of the primary use cases is for search and discovery and the ability to pull real-time airchecks for pre-produced spots and live reads. It starts with verification and the development of more real time affidavits so they can be more competitive and on par with other forms of advertising, specifically digital. The way aiWARE is deployed within these radio groups greatly improves the efficiency of programmatically facilitating airchecks of live reads. So you now can get a 360-degree view of all spots that aired and also track the audience size and the ad efficacy, using Nielsen data or other audience-based metrics.

How are clients using the technology for organizing and redistributing content?

Each group is a little different. We all imagine a world where all of this content is going to be indexed, packaged and potentially redistributed on a host of new and exciting platforms. Our technology makes it very efficient and turnkey for both broadcast groups and their affiliated partners to greatly enhance search and discovery of their content. Radio and TV historically have not done a lot of pre-indexing and archiving for the purpose of redistribution.

The majority of podcast consumption is original programing, not traditional radio-based programming. We think that trend is going to change. AI [artificial intelligence] makes it easier and more cost effective to index mountains of radio content and prepare it for redistribution, whether it’s on-demand streaming, podcasting, and the emerging and very exiting platform of IOT [internet of things] with Google Assistant and Amazon Alexa.

What insights can aiWARE provide radio sellers to help them grow ad sales? 

AI is just a tool to create effective data that we correlate with Nielsen or other proprietary datasets. We want to give real-time visibility, down to the second, of everything that’s happening on a radio broadcast: What they’re talking about, the context of the conversation, what ads and programing are airing – all of that is geo- and time-stamped. You can look at all those different elements with the same resolution you can for a digital campaign, whether you’re using first person data or different forms of data to overlay on top of Nielsen-based data.

That allows you to answer the questions: Did my campaign increase website traffic? What was the total number of spots and cume audience reached by zipcode? And then benchmark that against the performance of sales: Is my advertising working? How can I compare it with increased search engine and website traffic and call volume that I have historically been able to do with other mediums? Both our broadcasting and advertising clients will attest to the fact that radio works. This is the next evolution at scale to cost effectively build efficacy models to empower radio groups and individual sales reps with the tools and insights to prove to their clients that their campaigns are working.

Most major groups are using several different attribution models so we can do seamless binding of those datasets. So I can see when and where campaigns are airing and correlate it with the audience size and benchmark it against Google analytics information or proprietary web logs or call volume.  

How else is the technology being used?

Everyone believes the future is not just telling Alexa to tune into the radio station. They believe the [smart speaker] format warrants a different type of consumer engagement since it is interactive. Now that we can index the content, you will see a richer content experience of almost infinitely personalized content, packaged and delivered just for you. The big insight here is maximizing content and yield. Radio is sitting on a tremendous amount of untapped yield from its primary content.

We like to think aiWARE will touch every aspect of a broadcaster’s business, from maximizing their current yield for advertising, which we’re doing today, to affording them the ability to start analyzing their content through a different lens and looking at it on a granular basis. For example, as compared to the norm, how are we trending in mentions of the recent tariff law put into effect? You’ll be able to see those insights a lot faster. Most importantly, it allows you to take advantage of the opportunity to repackage and redistribute content for somebody who only has five minutes to engage with the show. AI will help optimize not just the advertising side of the ecosystem but also fundamentally the content itself.

Can broadcasters use aiWARE to monitor what the competition is talking about?

If you are looking for chatter or trending topics on a competitive station, if you’re leveraging our technology to give you an analysis of what they’re talking about, if that is the end goal, we can do that for almost everybody right now. If you intend to redistribute, share or post content that you do not own, we do not facilitate that and you need a license. That’s where the line gets drawn.

AI is relatively new to broadcasters. How do you see its use evolving in the future?

People are going to start allocating more and more budgets to AI machine learning, potentially indexing all of their content and programming. That prepares them to redistribute and optimize the content of their choice for IOT playback. Secondly, radio has woken up in a big way to the real opportunity of podcasting, to effectively compete and dominate this new emerging market. We’re very bullish on podcasting. Take that to a whole next level with IOT and Alexa and Google Assistant.

I think radio will be stepping up in a big way, which will be a huge advancement and growth curve. But that only works with AI and cognition, because there’s so much content and you have to be able to package and organize it. You can’t create an infinite array of playlists for every single individual. But by analyzing what individuals are listening to, you can automate that through AI.

It’s not pie in the sky to do this cost effectively at scale. Most TV station affiliate groups have put in some form of recommendation engines on their websites that helps improve time spent listening and engagement. I think traditional radio broadcast groups are going to follow suit. And because of the emergence of podcasting and IOT, there will be real dollars waiting for them.

Are any of your radio clients making their audio searchable by keywords, a sort of Google for audio?

All of them are internally but it’s up to them whether they want to open it up to a third party search engine. People are invoking our API and putting some semblance of search on their websites. But if you’re referencing a consumer front-end for search and discovery, I don’t think any groups have made the decision to do that.

Is that the next step?

I think it’s coming. Every station built mobile apps but not too many of those got singular traction, independent of the larger gateway apps. You’re seeing the same kind of the same thing with Alexa skills. There’s a mad rush to build skills. But to take advantage of the opportunity, the experience needs to be different than just telling Alexa to turn on my favorite radio station. That's something that radio groups are going to need to rally behind. And that leads up to search, meaning that groups will have to play together.

To provide a large enough corpus of data to make it valuable for somebody to search against, you can’t have two radio groups be the only ones indexing their content. Imagine Google indexing only 10% of their content – it would be a very shallow and hollow experience. Radio and larger groups have their own agenda and their own proprietary content, but consumers need enough scale to satisfy them and give them a reason to come back. For a group to realize the vision that you’re talking about will require investment from almost every broadcaster that’s willing to say, “Here’s our content, here’s a very rich and vast and dynamic experience.” And I don’t see that happening – at least in the next few months

How is Veritone helping radio keep pace with advertiser demand for data and analytics and to make radio more accountable? 

Every radio group is going to have to be ready to deliver to their clients a report that’s discernible and not too confusing. And if the brand demands a data feed because they’ve invested in their own data farm and they want to ingest it and do their own hyper-analytics, broadcasters also need to be prepared to do that.  It’s not just: Hey, we ran your campaign and we’re going to send you a log file at the end of the month. The passion for data analytics is in our DNA and we have afforded the ability to provide that level of insight in any format the broadcaster or advertiser is looking for.