Morning Brew just started monetizing our first podcast. Turns out, it’s low key very different than selling newsletter ads. Newsletter ads are fun! We write the ad, the partner approves it, we send the newsletter to an exact number of people and we show success via open rates, click throughs and brand lift surveys. Podcast ads are… not so easy. In order for publishers to make the best experience for listeners and profit, we have some work to do. So here’s some baseline knowledge we have recently collected and a few of my initial observations.
Let’s lay the groundwork for the context of this piece:
Assumption #1: We are not at peak enterprise podcasting. You have an incredible in-house brand partnerships team. Extra points for relationships with partners that trust you with new products. Hobbyist podcasting has a different monetization landscape.
Assumption #2: the podcast advertising market is still very confusing.
Let’s get into it.
This space is confusing:
- There’s a lot of jargon
- Listening platforms (Apple and Spotify) provide limited data
- There are a lot of advertising strategies
- Dynamic insertion is new and attribution is hard
Podcast-specific jargon exists because of the RSS feed. RSS is a type of web feed that allows users and applications to access website updates. It was created in the late 90s by Netscape and is basically the same today. The feed design allows the “push” of new articles and new podcasts to news feeds and listening platforms.
The RSS feed defines the URLs where you can download an mp3 file. Once the actual URL for a particular episode is downloaded, the host platform can officially count a download (Thanks Mike and Jeremy from RedCircle for the definition). It’s not like Netflix where the application can see user information, clicks, show consumption or other shows also watched. Well, the listening platforms own this data but don’t share it. We’ll get there. I also made a chart.
Podcast data is messy because RSS is “simple” and the listening platforms own the real data but don’t share it. Today, the Interactive Advertising Bureau (IAB) certifies the downloads, listeners, ad delivery and client-confirmed ad plays. Host platforms have been adopting the certification, but listening platforms provide zero transparency.
We sell on the certified download metric. So when we show metrics to partners, our host platform (Art19) has the certification to back us up.
But downloads do not directly translate to listens. When the Apple podcast app automatically downloads a new episode from a show you are subscribed to but you never actually listen to, it still can count as a download. This is an issue. But it gets worse.
Our hosting platform shows us the number of listens on each listening platform. Spotify’s data showed 43% more listens than our hosting platform showed on Spotify. Apple showed 80% fewer listens than our hosting platform showed on Apple. Maybe there’s a delayed update window. Maybe I’m looking at it wrong. I hope I’m looking it wrong, please email/tweet/holler @ me if you know how to reconcile the numbers. Anyway, they do provide data that I want to list out:
I definitely want a consistent download number. And originally, I wanted even more data about our listeners. But in retrospect we wouldn’t significantly change our actions by knowing granular information. But we do need more ways to engage with the audience. Basic features like customized push notification to followers, enhanced capabilities in the show notes and A/B testing episode metadata would all allow creators to engage more and constantly improve the product.
Before getting to advertising jargon, publishers need to know how many downloads are available to advertisers. We created a model for projecting impressions and managing sold inventory (happy to send you my template) that takes into account long tail downloads since old episodes continue to gain fresh listens. It’s still early, but we can get a good enough estimate of the total impressions we’ll generate next season and sell against that total.
OK, now more jargon and tables.
All of these except for baked-in ads leverage dynamic insertion. That’s the process of customizing each individual download. Operationally, here’s what happens:
- During the recording, the host (let’s call her Kinsey) precedes a question during the interview with a “before we get to that, let’s break for a quick message from our sponsor”
- Kinsey pauses and picks back up with “and now back to our conversation with [this week’s awesome guest]. Here’s my next insightful question…”
- In production, we create an insertion point with that timecode and the right ad placement label (pre/mid/end roll).
- Advertisements placed by our ad ops team then automatically trigger at the insertion point.
Dynamic insertion is… pretty dynamic. Publishers can tie ad units together in sequential campaigns. For example, we run a pre-roll that mentions partner A and is always tied to a mid-roll advertisement for partner A. Our episodes have 2 sequential campaigns, each with a pre and a mid roll that are host-read, native advertisements.
Publishers can also use dynamic insertion for creative content experiments like intros based on the time of day or rotating calls to action (Subscribe! Share! Follow on insta! Send us episode ideas!).
A lot of host platforms are adding support for dynamic insertion and are building programmatic ad platforms in hopes of helping monetize inventory. The programmatic bet does not align with aspiring publishers like Morning Brew that plan to keep monetization in-house. Rather than taking over our sales relationships, we want tools tools that improve our ad operations. For example, a feature that exports audio clips (30s before the ad, the ad, 30s after the ad), transcripts, timecodes and number of impressions served with that campaign over time to partners would be incredibly helpful.
Now you’re probably wondering how we measure the performance of these ads. Stop wondering. Just stop. There’s no real solution here. Listeners skip ads, ignore unique links/promo codes and may be influenced by a podcast ad but only make a purchase when the clickable, visual ad is in front of them. That’s a lot of leakage!
There are helpful studies like this one that observed 22% of podcast listeners have bought products based on advertising, but there’s no true attribution like we have in newsletters or web. If we keep creating great content the rest of the market will mature and create better tools. In the meantime, we are trying to advertise the same brand partner on other show assets such as newsletters (which have better attribution and can prove that our audience responds well to the brand).
One day a listening platform will make this much better. Imagine actually enjoying a well-produced ad, opening up the app while listening and having a product image right there to click on. Then when the advertisement ends the main image is flipped back to the episode art.
We are going to see a number of different monetization options grow, like selling exclusive content to listening platforms, charging listeners via paywalls, selling merch, ticketing live events, syndicating on radio and profiting off of YouTube traffic. We’re monitoring all opportunities and can’t wait to see which pan out.