Erik Lundberg is Chief Revenue Officer at ArtsAI — this interview has been lightly edited for style and readability
EL: My name is Eric Lundberg, and I’m the Chief Revenue officer at ArtsAI. I’ve been with the company for eight and a half years, and in the in the broader digital marketing industry for about three decades.
JC: So - what is Arts AI?
EL: We’re leaders in three aspects of the podcast industry.
The first is measurement. There’s a large number of podcast advertisers who want to run an ad to drive someone to their website, to make a purchase, or a free sign up: and we measure that. We also measure when someone hears a podcast ad and goes to an advertiser’s mobile app - maybe it’s making a purchase at McDonald’s, or signing up for DraftKings or something like that.
We also have in-depth analytics. We can help you by targeting time of day, day and week, shows and episode - targeting that gives you the highest return on investment, or the lowest cost per conversion.
And lastly, we have AI personalization technology that will personalize and optimize your ad creative. Should it be a younger male, or an older female announcer? Should it have an Australian accent or an American accent? What should the offer be? What should the call to action be? We personalize and optimize the podcast ad creative, to typically increase conversion rates from 50-100%.
So we have a full spectrum of offerings to make podcast advertising more effective; and with our partnership with Barometer we can do brand safety and brand integrity for podcast advertising. So if you’re an agency or a marketer and you don’t want to be in profane content or violent content, or you want to stay away from certain sensitive social issues, you can set those brand safety preferences. It’s the first in-campaign solution that actually will flag those for you, daily or weekly.
So ArtsAI offers a full suite of advanced technology tools to help make your podcast marketing more effective.
JC: The big news is the merger that you announced earlier this week with Claritas. That must be a a tremendous piece of of news for you and the team.
EL: Yes - tremendous news. Claritas is one of the leading marketing data measurement companies in the world; they have the leading identity graph in the industry. There are billions of data points - they have had PRISM segments for decades, which is one of the leading targeting segments available. So it’s very big news.
They have a very robust podcast measurement business, and are one of the leaders in that area. We’re bringing our technologies and two very talented teams together. And yeah, we’re thrilled.
JC: Claritas must have a lot of the same tools that you do, but you’re presumably bringing something new to the table?
EL: Their measurement is based on their identity graph, and the device needs to be in their identity graph. That leads to very precise measurement, and some amazing audience insights where they can tell you the PRISM segments that work best.
We come at it with some very advanced AI technology. That’s the name of our company, ArtsAI after all! We use a probabilistic approach for our measurement. So, we can measure a conversion even if the user is not in our device database. We have our own proprietary device graph, and we can measure the conversion if the impression or the conversion occurs on a cellular internet connection. Claritas does amazing work in the US, but their identity graph is US focused - with our probabilistic approach, it allows us to measure globally.
When you start thinking about the possibilities of bringing our advanced AI personalization with their identity and data graph, it’s really phenomenal. Our AI personalization can take any data input to make that personalization and the optimization even more powerful. There’s some, some very unique capabilities that come into the equation from each company.
JC: Now, the Podcast Business Journal is read across the world; and whenever people talk about measurement in Europe, the four initials GDPR appear. What do you do in Europe?
EL: We have not rolled out in Europe, or in the UK, yet. We’re going through an in-depth GDPR compliance audit. We expect that to be completed in the coming months.
The way that our probabilistic measurement works is we get a bunch of input signals from the impression we have. Our impression tracker tracks the ad being served. We have a prefix URL redirect solution for baked in ads, and then the advertiser typically places a conversion pixel on our website. From that conversion and from the impression, we typically get user IP address, and from that the city, the user, and we can know a rough geo-circle location.
We can know some general information about the device, we can know the timezone, what their ISP is, and so on. So, the reason we believe will be GDPR compliant and rolling out there is that we don’t need all those signals.
We have a partner here in the US, for example, who won’t share IP address with us. So we’re already doing some great attribution without using an IP address. Our machine learning probabilistic approach is very flexible and robust and we’re looking forward to rolling out in Europe in the near future.
JC: In terms of the the automated creative, you can produce different bits of creative for different people. How does that work in practice?
EL: We don’t do automated production of the creative. We sit down and consult with the marketer and the publisher or agency. We spoke about this when we met at Podcast Movement a few weeks ago - a lot of marketers, especially a lot of the very big ones, are taking a radio commercial; and that’s not the optimal creative for a podcast.
The example I like to give is my mother - she lives in Charlottesville, Virginia. She’s a senior citizen, and if she’s on an Android phone on Sunday lunchtime, and she’s listening to a cooking podcast and the advertiser is a leading exercise bicycle, the ad creative that’s going to convince my mom is different to the ad creative that will convince me - on my iPhone at 6pm on Wednesday evening, listening to a sports car podcast. The audio and creative that’s going to convince my mom is very different for me. We sit down with the marketer and we say: would you like to try different voiceovers, a male or female, older or younger? Would you like different music, like rock, classical, jazz - should we try different sound effects? What’s the copy? What’s the script? What aspects of that product of your product you want to talk about? What’s the call to action? What’s the offer? Then the creative is produced, and then our patented AI technology takes all available input signals. Where is the user, what type of the device are they on, what’s their content? It figures out, for that specific user, what the optimal combination of those ad creative messages is. That decision is made in real time, and the optimal combination of ad creative messages for that specific user cohort impression is served. It increases conversion rates typically anywhere from from 50 to 100%.
JC: Are you mixing the audio on the fly then?
EL: We pre-produce them because you can’t create a broadcast quality audio ad with all those combinations in milliseconds and serve it. Pre-producing works better, in other words.
Host-read ads are inherently contextual. If it is a cooking podcast, the host is talking about the product or the service in the context of their podcast for their listeners; it’s a sports podcast, the host is doing the same thing. And then you have DAI ads, where people run the same ad creative everywhere. So this AI personalization technology can bridge the gap between host-reads and DAI.
JC: One final question: you’ve obviously got an awful lot of data around how well the ads on your service work and how efficient and effective they are. Do you have any tips for advertisers in terms of what to do with that ad creative to make it as effective as possible?
EL: One of the great things about the merger with Claritas is that there is a very good chance we’ll be putting out a lot of research on these benchmarks. As far as tips for today, you do need to be a fairly large advertiser to employ that personalization technology. If you’re a smaller advertiser or you just aren’t ready for this technology, do an A-B test. So try male versus female. Try different copy, try a different music background. If you’re doing host-read ads, try a few different scripts, and then just like learn and optimize.
Let’s just say you try male versus female voice over, and the male voice just gives you a 25% increase in response rate. So, then try different male voices, older, younger, British accent, American accent. That continuous evolution: test and learn, test and learn, test and learn. That would be my advice for your readers.
JC: Thank you so much for spending the the time with us.