The host-read ad dominates podcasting, but ad tech is rapidly changing that

Back in the early days of podcast advertising, which means two hours ago, things were pretty simple. The host read an ad, someone clicked to a website, they activated a coupon code, and a mattress arrived at their home. Now as more brands are engaging with podcasts, we are hearing about programmatic insertion, geo-targeting, lift studies, pixels, attribution, conversion, engagement, user-agents, caching of audio and more. Or to put it another way, our little baby is growing up so fast.

eMarketer forecasts a doubling of programmatic ad spend this year. Right now it’s tiny but will ramp up sharply over the next few years.

Nothing may be more emblematic of the rapid evolution of the sector than a start-up newsletter focused exclusively on advertising technology for podcasting. Bryan Barletta who has spent time in the world of ad tech has launched Sounds Profitable, a newsletter affiliated with the same fine people who bring us Podnews.

We sat down to dumb down all that is going on with podcasting and ad technology.

This interview has been edited for clarity.

(S) An ad tech newsletter for podcasting means we must be moving beyond the simple host-read ad ….. How did you get into all of this?

(B) I was part of the initial team at Barometric and helped build that out.  We were successful creating campaign attribution which big advertisers like and decided to pivot the company to focus on podcasting, and then Claritas bought us. I worked with Megaphone running data and monetization. I spent much of my time breaking down these topics and conversations for salespeople or account managers.

(S) I assume it's big brands and the dollars they bring that are pushing us beyond the baked-in ad. 

(B) Right, I think it's the people in digital who said “we’ve got all this money to spend and are super happy to try podcasting, but if it doesn't fit into our framework, and can't match columns in a report, it won’t work.” So in came third party tracking and attribution, that led to more ad dollars.  

(S) We hear the term “attribution” an awful lot these days. Lets crack that open a bit.

(B) It’s the ability to match an exposure of an advertisement to the desired action like purchasing a product (conversion) or to gain interaction with a potential customer (engagement). On the digital side, they also capture clicks.  

Advertisers place pixels on their website for everything they’re looking to track

Bryan with his editor-in-chief Theo

(S) Just because I hear an ad for something doesn’t mean I click right away, any more than I see a billboard on the road for beer and pull right off and buy a six pack.  How does that fit with attribution? 

(B) So that gets us into multitouch attribution or weighted attribution. You have a billboard, you a have digital campaign, your podcast, TV, and everything else. It takes a lot of effort and coordination to properly tag each of those channels, but you can see over the next 30, 60, 90 days if the person converted. Podcast attribution focuses heavily on “Lift Reports,” which take into account all the traffic driven from non-podcast channels without the complex setup.

(S) So that's what the advertiser is doing. They're taking this volume of data and looking for cause and effect.  Let's put that into plain English.

(B) The advertiser places pixels on their website for everything they’re looking to track. When you hear a podcast advertisement, the third party is able to identify your IP address and user-agent. Then, over the next 30-90 days, the advertiser is looking to see if someone from your household has either interacted with the site at all (engagement) or taken the specific action they were aiming for (conversion).

(S) What are “lift studies” --- and what are they measuring? 

(B) The advertiser places a pixel on their site for two weeks before a campaign starts. In that time, they’re identifying everything they can about every visitor - IP address, cookies, mobile device ID’s, and identity. 

Again, the name of the game is engagement and conversion. 

For a brand lift study, they look at the data in that two week period and identify the baseline visitors the site has before the campaign even begins. Then they look for people who haven't been exposed to the ad but match the demographics of the exposed group. The result is a report showing the incremental “lift” your campaign provided.

Brand lift surveys are slightly different from the web studies in that they form two groups and send them both surveys.  Or they play the ad for the exposed group but not the control group. The goal is to compare the effectiveness of the campaign.

 It’s really three different ways to approach lift, each costs more than the last.

(S) Back to pixels for a moment. Companies like Podtrac and Chartable are using them. Where are the pixels inserted?

(B) If you use a “prefix URL” like Podtrac does, it is added right before the URL where the podcast episode is.  It means that when your device goes to play that episode or download it, it first pings Podtrac and sends the information they need, then they redirect back to your device to get the episode. That’s how they know how many downloads have occured.

(S) NPR proposed a technology a couple of years ago that would send info back to a server as people listened to certain parts of a podcast – so they would know if the ads were heard and how far people listened through a particular episode. 

(B) It might happen in the NPR One app but Spotify and Apple, for example, are not in any way motivated or incentivized to provide additional data.  It will never happen unless everyone is on board. 

A big thing I like to point out is that Spotify is not a podcast advertising platform. It is a mobile application with sign-in data at the scale and granularity that Facebook has and it has its own private ecosystem. The content they happen to use is podcasts.

(S) Let’s unpack the whole privacy issue.  You wrote in your newsletter ‘digital will look more like podcasting at the end of the privacy debate’ - what does that mean?  

(B) First, being pro ad tech doesn't mean anti privacy. In a pre-privacy world, people logged into Facebook and they have your name, phone number, address, email address, and all of your habits. Then if you log into Facebook from your phone on the mobile app, it as your mobile identifier.  From your computer, it has your IP address for your home. 

And basically, if you stay logged into Facebook or you log into Facebook on all these different apps and experiences, it follows you throughout the web.   So now the cookie which enabled all of this, is going away.  

(S) And podcasts?

 (B) Lower tech.  IP and user-agent, are the only things podcasts have been dealing with.

(S) Lets talk about geo-targeting – ads running in a particular geographic area like Chicago. NPR made an announcement a few weeks ago about customized versions of their afternoon news podcast with specific content for an area such as Boston. 

(B) Buying on a national level, or local is really powerful. State is a good area to focus on if it's a cellular connection, but more specific may be tough. If I check my geolocation right now by IP address, it tells me I'm either in San Antonio or Houston and I'm in Austin. 

(S) So selling metros would be tough right now. 

B) Everyone, Art19, Megaphone, they can all do dynamic ad insertion, but I don't think any of them will reliably say they can just hit say Tampa because of the differences between WIFI and cellular.

The empire that Sirius XM is building is going to be one of the most competitive things that we're going to see in this space for Spotify

(S) What is programmatic ad insertion versus what a company like Megaphone is doing with their marketplace? 

(B) When a lot of people think of programmatic right now, they think of something like The Trade Desk on the video and digital side. I put my ad in and it can go everywhere because everybody is connected to The Trade Desk. We need to hurry up and get there. There is some movement in podcasting already, but not a lot.  It’s mostly manual.  What a lot of people refer to as a programmatic solution currently, isn't that, yet. 

Publishers are getting access to programmatic-like experiences by using VAST tags for each programmatic partner, like DAX, Triton, or AdsWizz. Pandora and AdsWizz can actually do true programmatic. I think that the empire that Sirius XM is building is going to be one of the most competitive things that we're going to see in this space for Spotify. If they can wrangle it all together, they have a lot of really talented people and a lot of great tech there.

 Watch out for Amazon in this space for podcasts and smart speakers.   

(S) It feels like programmatic is a solution beyond the elite 1% of podcasts making money now.

(B) Absolutely. I think that even in that top percentage, we're only filling 50% of inventory every month. 

____________________________

Certainly the host-read ad will continue to be an effective tool. But advertisers are looking for better measurement and tech to facilitate bigger campaigns. And still there are the issues of compatibility. Apple makes targeting and measurement more difficult, Spotify has tremendous profile capacity and Sirius/XM is poised to have a big impact.

Thanks Bryan. Really helpful. We’ll be following the newsletter and watching the rapid evolution of ad tech for the podcast business.

 9/30 added note - As mentioned above, the host-read ad will most certainly endure. It has a unique magic hard to find in other media forms. In most instances host-read ads, like most podcast ads, are now dynamically served and thus measurable in the same manner as the other ad types we discussed.

While the business was largely built on host-read ads, the march to ad tech has introduced other ad forms which enable more choice and flexibility which advertisers like.