In episode 7 of On the Mic with Ad Results Media, Ad Results' CEO Marshall Williams and Sr. Account Strategist Andrea Schwarzbach dive into topics related to the proliferation of podcast attribution platforms, the advertiser's strong desire to understand and capitalize on what the data is telling us and the comparison to other methods of tracking, such as vanity URLs and surveys.
Be sure to subscribe to the "On the Mic with Ad Results Media" podcast through iTunes. It's a podcast about the business of podcasting and audio advertising. We aim to educate, enlighten and push the industry forward.
Podcast Transcription:
Andrea Schwarzbach:
Today's episode of On the Mic takes a new direction, as partner and CEO Marshall Williams and I, Andrea Schwarzbach, take the hosting reins to discuss the evolution of podcast attribution and the part that our agency has played in it.
Speaker 2:
That's the great thing about stamps.com. They grow with you.
Speaker 3:
As much fun as I had, I couldn't wait to get back to my Sleep Number bed.
Speaker 4:
Yep.
Speaker 5:
I love my Third Love bras. They're hands down the most comfortable bras I've ever owned.
Speaker 6:
I love making Blue Apron.
Speaker 7:
Mm-hmm (affirmative).
Speaker 6:
I love it. It's my me time.
Andrea Schwarzbach:
So the first question that I want to ask today, and really just kind of discuss the history of podcast attribution, where did we come from, what did we start looking at in terms of attribution with podcasts and audio in general?
Marshall Williams:
Okay, so great question, and thanks. I'm happy to be here today.
Andrea Schwarzbach:
Yeah.
Marshall Williams:
Ad Results started out as a radio company, and radio obviously, before digital audio, was purely terrestrial, meaning AM and FM radio. And local market advertisers wanted to find out how their ads are working. So they would either drive people to their brick-and-mortar stores, or they would use a vanity phone number, which is the term for some phone number that's very memorable, “777-7777. Call us today. Tell them you heard it on Jack Snow at KDKA”.
Marshall Williams:
As we moved into the national advertising space, you had to use toll-free numbers. And so the company I worked for was a company named Allergy Free Air Filters, so we secured the number 1-800-ALLERGY. We would say, "Tell them you heard it on Paul Harvey and call this phone number." And those were the very primitive days of audio tracking. The difficulty in the audio landscape in terms of attribution, is that people consume audio in one context, perhaps they're driving, out walking the baby, walking the dog what have you, and they respond in a different context. It's always been a very difficult thing to track.
Andrea Schwarzbach:
Right.
Marshall Williams:
As the internet became much more ubiquitous with how everybody conducted business we had to create new metrics for ... And you couldn't say, "Call this number and use, 'Tell them I sent you.'" You would say, "Go to my website." And what we did was say, "Y- Enter this media code," or provide them an incentive, a financial incentive, perhaps a discount, or what have you to say, "Use Jack Snow from KDKA Radio and tell them that's how you heard about us." Again, the difficulty in reading audio is the contextual elements of it. You tend to listen outside of where you respond. And so that makes it difficult. People oftentimes don't remember media codes, so on and so forth. So we needed a better option.
Podcast Advertising Attribution Method #: 1 Media Codes
Andrea Schwarzbach:
Can you just give a quick example of what a media code is for some of our listeners?
Marshall Williams:
Of course. Of course. Good question. The design is to allow somebody who's a fan of this audio content, whether it be radio or podcast, podcast has become a much more ever present part of our existence a code so that the person who listens to that content can get a discount or what have you. And it also allows our clients to figure out where the dollars that are being spent in those particular avenues are working.
Marshall Williams:
So if you're trying to drive somebody to subscribe, or not subscribe, but purchase a particular product, you might say, "Okay go to the website abc.com and enter the media code snow and you'll receive 20% off your first order." And that would either be p- at the promo code box, which most people know where they are, or it might be at the beginning of your checkout window or something like that. But again, the design is to, one, quantify who sent that person to the, the commercial site, the brand site.
Andrea Schwarzbach:
Mm-hmm (affirmative).
Marshall Williams:
... if they made a transaction, and to offer the listener an incentive to do so. it's relatively easy, but you'd be surprised how few people actually respond with that media code.
Andrea Schwarzbach:
Well, that actually leads me to my next question, which is what are some of the challenges that we've really seen in [podcast advertising] tracking this way so far?
Marshall Williams:
Again, as mentioned earlier, there's a contextual difference between where people consume the media and where they respond. It's very difficult if you're driving in your car and, and not as safe as well to pick up your phone and go to a particular client's website to try and purchase while you're driving, so most ... or walking the baby or walking the dog or at the grocery store, or whatever it may be. And so what happens is that people will wait.
Marshall Williams:
And so if you say, my previous example was Jack Snow, "Enter my code, Jack Snow, at checkout, and you'll get 20% off your first order," or whatever it may be people won't remember that. Okay? They simply won't use it or what have you. So the attribution elements that we're looking for are pretty blurry. I mean, oftentimes we will look at you know, maybe two to five out of a hundred who will actually use a media code, but we're getting to a place where we can read how many people actually responded in a, in a much more effective fashion right now, utilizing some, some digital techniques.
Podcast Advertising Attribution Method #2: Point of Sale Surveys
Andrea Schwarzbach:
Right. So along with that, I think there's a lot of brands out there that have come up with a semi-solution to tracking to account for those that maybe forget the code in that time span of them not taking action, and one of those solutions is a survey on the back end. So can you talk a little bit more about that and how we've seen those results influence our campaigns?
Marshall Williams:
Sure. As mentioned people will go to a website being driven by a piece of audio content, whether it be podcasting or radio, or what have you, and they simply won't remember the code, okay? they'll go through the transaction process. And what we tried to do is give them a supplemental kind of memory device thing so that somebody can say, "Oh, that's right. I did hear it on the podcast." So, as you go through the checkout, and you don't necessarily use the, the promo code, 'cause you just want to buy the product that this host recommended to you, you, um ... a survey shows up. It's interstitial, meaning it's in the flow of the transaction. And it's usually a single-page things that say, "How'd you hear about us?" Most people are so familiar with this, that it becomes s- simple for them and it won't chase them off or what have you.
Marshall Williams:
And they'll add a box in there or something that says, "Podcasting," or what have you. And we use that as a proxy for how many people actually signed up. So if we get three or five out of a hundred, the survey may indicate that it was higher than that. So let's say that we give somebody a multiple of three X or four X on top of those three or five. So it actually gives us a better, more accurate read on how our attribution truly is and how efficient the investment was on that media outlet, on that audio channel.
Podcast Advertising Attribution Method #3: Pixel-based Attribution
Andrea Schwarzbach:
So now that we are in this ever-evolving industry, which is one of my favorite things about working in the podcast landscape, it's just always changing, something new is always coming up, one of the newest trends and something that we as an agency have been very focused on testing for our clients has been pixel-based attribution. So let's talk a little bit about that, kind of explain to our listeners how that works, and what we've been seeing with some of our clients.
Marshall Williams:
Fantastic. So great question. In an effort to try and get a more accurate read on what ... You know, we spend millions of dollars for our clients and in these audio channels and trying to get a performance read is really critically important. I've always felt that because of the aforementioned disconnect between consumption and when people respond, there's a gap. I mean, I, I think ... I truly believe it works better than it does, and we were looking for a tool to be able to prove that.
Marshall Williams:
Media code is mentioned. Even the adjunct survey is good, but it's still directional at best. There's a lot we're missing. So, in diving into how digital marketers track discovered that in, in social media posts, Facebook digital display ads, et cetera, they can embed a pixel. And in the context of those display ads, et cetera, when somebody clicks on an ad that pixel fires at a corresponding pixel is placed on the brand's webpage and in a few other places in that brand's webpage so that they can track what happens when somebody responds to that digital ad.
Marshall Williams:
I thought to myself, "Man, we gotta be able to do this in the audio space. Is it, is it doable?" And we had talked to a number of people who said, "Well, we're not quite there yet," but a couple of smart, savvy, tech software people said, "Yeah, I think we can do that. What we'll do is we'll place a corresponding pixel in the RSS feed." And the RSS feed is when people hit download on a podcast, it's the, it's the format that that podcast sits in, and it starts to play as soon as people download that podcast. Okay?
Marshall Williams:
So they would place a pixel in that RSS feed. So when somebody downloaded the podcast that pixel would fire. Now there's a lot of permutations that take place from there, but what happens is we can do the same thing the digital marketers do, and put a corresponding pixel on the brand's webpage, their transaction page, whatever it may be, and track that listener in a much more linear and direct fashion than we've ever been able to do before.
Andrea Schwarzbach:
Right.
Marshall Williams:
Right. And so it's still early in the game, but we're excited about what we're seeing so far because it's showing tremendous lift in response, which we assumed all along.
Andrea Schwarzbach:
Right.
Marshall Williams:
Right, yeah.
Andrea Schwarzbach:
So it is showing tremendous lift?
Marshall Williams:
Tremendous lift.
Andrea Schwarzbach:
Can you talk a little more about that?
Marshall Williams:
Sure. Okay. So what we're finding is that as this pixel fires, it goes through sort of an ingestion process to determine anonymously who this individual or who this digital fingerprint person is. Somebody consumed a podcast on their phone, laptop, whatever, all of that has kind of a, a digital thumbprint. It's anonymous to that person, but you can still figure out who that is. And that the corresponding pixel as mentioned is on the brand's webpage and that kind of thing.
Marshall Williams:
It allows us a much more direct linear connection between the podcast listener and a response. And so, if you look at just our media code attribution, which might be three or five out of a hundred, maybe slightly higher even with the adjunct survey, I mentioned earlier where it might be three X on top of that, we're seeing ratios as high as 10, 12, 14 X on top of that.
Andrea Schwarzbach:
Wow.
Marshall Williams:
Right. It's really impressive. And the closer the shows are to alignment in terms of the context of the show, content of the show, and what that ... Let's say it's a gaming company advertising on a gaming show, you see a high, really high lift there. it's interesting what we're learning because there are some shows out there, some categories of shows, where they just do not care about a media code. They'll just buy the product, okay? And then you see you know, not ... you s- you see a fairly low amount of media code redemption, but a really high transaction level when we use the pixel-based attribution tool. So exciting growth there.
Andrea Schwarzbach:
So let's talk about some challenges that we have faced thus far in pixel-based attribution.
Marshall Williams:
Right. Good question because it is challenging. First of all, it's very new. I mean, the, um ... we ... So here it is the end of February or middle of February, 2020. The first conversations we had about this were first quarter of last year. So it's an incredibly new space. So it's, it's, it's morphing and changing all the time. one of the things that we would ... There was a little bit of concern early on that the broadcasters, the networks, the shows were concerned that this tracking would track downloads.
Marshall Williams:
And there was a little bit of concern about whether it could count downloads, and then would the clients say, "Oh, we have a problem with the download you're ... we're paying against versus what's really being done." Well, it doesn't work that way. Okay? It doesn't work that way at this point. It may in the future, but right now it's a ... It's, it's kind of a ... The download number versus the pixel-based tracking are two separate things.
Marshall Williams:
Pixel-based tracking is the pixel is embedded in the beginning of the RSS feed to ... for the great ... to the greatest extent. And that doesn't indicate actual listens. It just indicates downloads. Now, obviously you can draw the conclusion that somebody downloaded and listened if that ... the, the pixel that corresponds to that listener hi- hits the brand's webpage and they can execute a transaction. I mean, the assumption is there. Okay? But it's still an assumption.
Marshall Williams:
Okay, there's a certain amount of math that then ... and um ... What's the word I'm looking for here? Understanding, or quantifying that listener against the pixel that fires on the brand's webpage, because there's all kinds of static out there. If somebody downloads a podcast at their house via a Wi-Fi connection, that's a very specific IP address. And IP is one of the ways that these companies use to quantify that listener, that fingerprint, I think, is a metaphor I used earlier.
Marshall Williams:
If you're in a business or that ... a building that has a thousand people in it, that may have five independent IP addresses. There's a lot of static there. Okay? So you have to have a way to ... a data set or a company that handles a lot of data aggregation help you filter it out. So there's not a one-to-one connection between a download and a response. There's a certain amount of math that works in there.
Andrea Schwarzbach:
Right.
Marshall Williams:
That's the way it is in the, the data business for a lot of different media. but that's one of the challenges that they have. There's a lot ... There's some assumptions that have to take place in there. My response is the clients get the information and they look at it, and they, they correspond that back to what that person, where the pixel fires on that particular webpage, what that person does. And they're like, "Yeah, that is a connection. That's a sale." There's no ... I mean, it's pretty clear to them. So it's much more one-to-one from an attribution standpoint from the client's end than it is the, the front end or the download side.
Andrea Schwarzbach:
Right.
Marshall Williams:
Make sense?
Andrea Schwarzbach:
Yes.
Marshall Williams:
Okay. Anyway, so that's one of the things we'll get to over time. I'm sure these companies will get further refined and better. So we'll ... And we'll hi- help them, we'll guide them because I know what I'd like to see. I think our organization would like to see a little more clarity when it comes to that and we'll get there.
Andrea Schwarzbach:
And what are the reactions that we're getting from our clients about attribution in general? I know we've had some, some of our clients have been testing this for several months. Some of them are just now and starting to test. What's the general reaction that we're seeing?
Marshall Williams:
Most are ecstatic. Most are very ... They're seeing real promise in terms of ... You know, that was my supposition all along, because of course, you'd expect me to say this "The media ... Your media is working better than you think it is," okay? And I would explain the whole thing about how, "Ah, it's consumed over here and the response is over there, so it's difficult to get these media codes to work. It's really working better than you think it does." And our clients were like, "Sure, sure, sure, you know, "You're supposed to say that. It's your job," right?
Andrea Schwarzbach:
Yeah.
Marshall Williams:
Well, as we started going through this and it indicated that it was really working better than we thought it was, they're just like, "Okay, great."
Andrea Schwarzbach:
(laughs)
Marshall Williams:
"Let's do more." And again-
Andrea Schwarzbach:
Awesome.
Marshall Williams:
... uh you know, it's a little bit, self-serving, and it's self-serving from the context of this is our business and we want it to grow and we want the space to grow, but I love audio. I've always been a huge fan of spoken word and listening, and I get mad and frustrated and laugh and all the things you're supposed to do with audio. And I've always known that it was an incredibly viable medium. And now we have a tool that helps us show performance better. Really excited about that.
Andrea Schwarzbach:
Awesome.
Marshall Williams:
I know-
Andrea Schwarzbach:
Okay. So we've talked a lot about what pixel-based attribution is, what are some of the challenges. Who have been some of the major players in the space? Who is kind of leading the charge as far as attribution goes, and who are some of the networks that have kind of taken and run with this?
Podcast Networks Leveraging Pixel-based Attribution
Marshall Williams:
Really excited that the networks have said, "We're, we're good with this." Wondery has been a great partner at this. Again, um the gentleman who runs that, Hernan Lopez, is, is a good friend and a very thoughtful guy, and I was so excited when he was like, "Yes, I want to try this," because he wanted to prove that it did work better than we had earlier interpreted, even with all the tools that we had worked on.
Marshall Williams:
And Barstool, another one completely different from Wondery from a content standpoint, but still they were like, "Yeah, we think that our audio works better than much of the attribution tools we had a year ago," and they've embraced it and it's worked, it's worked well. It's re- really indicating that the audio channel is working significantly better than we thought it did with our previous tools.
Marshall Williams:
The tools, the ... There are two companies that we work with primarily, a company called Pod Sites out of New Hampshire at another company called Chartable out of New York that have really become allies with us in terms of this pixel-based attribution. They're a little ... nuanced a little differently, they, they interpret the data a little bit differently because there is an interpretation factor there.
Marshall Williams:
The Chartable guys one of them came from the mobile echo systems, so I think both of them could do that, but the Chartable guys were the first ones to be able to have mobile attribution. and we're working through that with some of our clients now. I'm excited about that because I'm curious to know if we run an ad in a podcast space and somebody goes to their phone and immediately downloads an app, I'm excited about what that future holds.
Andrea Schwarzbach:
Yeah, absolutely.
Marshall Williams:
And so those are the ... You know, the Pod Sites guys are, are ... have been good friends and helped us learn in this space, 'cause as mentioned, it's really early still, and we will continue to learn more going along. the early reads are exceptional, though. I'm very happy with, with what we're seeing so far from a response standpoint for our clients.
Andrea Schwarzbach:
Final question for you for today. Where do you see podcast attribution in the future? Say in the next five years?
Predictions on Podcast Attribution
Marshall Williams:
Great question. Further refinement of what we're doing now. Okay? So more digital ask attribution style. When I say that I mean a little bit more one-to-one where we can quantify if somebody downloaded a specific podcast, listened to the ad, and then responded, 'cause as mentioned most of this exists right now in the RSS feed at the beginning of it, so we can't show whether it actually c- the ... consumed the ad or not.
Marshall Williams:
Again, there's a, a logical leap you make that they did because they responded. So, but something a little more finite. Okay? I'd also love to see, and again, let me quantify all this by saying, we're going to w- w- work towards maintaining the privacy elements that we need to in this. Okay? Anytime you talk about IP address GDPA CC ... no GDPR, CCPA, d- which is the California privacy rules, and it's being written right now, I mean, literally you know this ... in this timeframe that we're in right now.
Marshall Williams:
As they quantify what they consider to be “PII”, or personally identifiable information, we have to stay between the lines there. And right now we're doing it. But if the legislation moves, we have to be able to move with it. Our brands may have to disclose that they're using these tools and, you know, allow somebody to opt out or ensure that they don't resell that information, 'cause if that's what the rules say, that's what they are. So we need to stay between the lines there. but I think that's very doable.
Marshall Williams:
So that's an advancement. I'd also like to see if we can take this tool and do brand-lift studies or the like for large, branded enterprise clients so that they can see what their brand lift and brand awareness might be if they're advertising in a particular podcast or a group of podcasts when they don't use a lot of these direct-to-consumer tools, okay? They don't ... Miller Beer, which is a client of ours, doesn't have a tool to quantify if somebody purchased via their website. They certainly don't need it.
Andrea Schwarzbach:
Right.
Marshall Williams:
But they do need and would like to have something that says, "Hey, I heard you on this podcast and I love the podcast and I'm going to really gravitate towards your product." So that's exciting. I think that's something we'll work towards in the near future, and I'm talking about this year. podcasting is still so new as an industry that everything moves like at light speed.
Andrea Schwarzbach:
Yep.
Marshall Williams:
And so I'm excited about going down those roads in the, in the near future.
Andrea Schwarzbach:
Awesome. Well, it sounds like podcast attribution in general is really exciting for the industry right now, so hopefully this was super informative for some of our listeners, and thank you so much for answering our questions today.
Marshall Williams:
Thank you. I hope I helped.
Andrea Schwarzbach:
Yeah.
Marshall Williams:
And we oughta do this again, because ... at some point in time, because it'll probably change in six months.
Andrea Schwarzbach:
(laughs) It probably will change-
Marshall Williams:
So-
Andrea Schwarzbach:
... in three months.
Marshall Williams:
Yeah, exactly. And so, we'll provide, here, we'll do the update version of the podcast.
Andrea Schwarzbach:
Yeah.
Marshall Williams:
Okay.
Andrea Schwarzbach:
Sounds good.
Marshall Williams:
I'm totally good with that.
Andrea Schwarzbach:
If you enjoyed this episode, be sure to subscribe for updates on future episodes and leave us a comment with your feedback, questions, or ideas for future segments. If you would like more info on Ad Results Media and what we do, please visit our website at adresultsmedia.com. This podcast is an Ad Results Media production.
