Gunfight at the Podcast Corral
Bryan Noguchi, Senior Vice President, Media Director
I’m an advertiser that likes podcasts. Low clutter. Captive, engaged audiences. Relatively low out-of-pocket costs for the most part. Great potential for contextualized messaging. Digital – or is it? “Digital” implies so much with regard to performance intelligence, and I think most advertisers would be disappointed to know how little information is actually returned. And what we’re finding is that a lot of seemingly basic questions go unanswered. Today, I’m thinking that if enough advertisers habitually ask the same questions, we might start to get some answers. The truth of the matter is that regardless of how measurable a podcast actually is, most ad products in this space have more in common with traditional radio on the accountability and measurability front (which is to say that the offering here is a little anemic.) This is significant, because despite whatever assumptions you have about podcasts, it’s an issue that has direct bearing on the definition of what an advertiser has purchased. So let’s start there.
What have you purchased?
On a basic level, you’ve purchased a :15 to :60 audio endorsement in one of three positions (pre, mid or post-roll). Maybe it’s like a radio spot, maybe it’s more of a live read. On some big podcast properties, you’ve probably found out that they’d rather your ad make no sense, than say something that sounds like marketing. You can imagine then, how important it is for us to understand the actions of the listener beyond the download – how many actually play the podcast? Where do they drop off? Half way through? Never? These are fair questions – it’s like knowing your digital display ad is above or below the fold. So you need to evaluate and estimate value on assumed behaviors: for example, you might over-value the pre-roll positions because fewer people would be likely to fast forward through this, while the post roll might be expected to under-deliver because maybe no-one listens to a given podcast all the way to the end.
Do you know if your ad is in a first-run episode or in an older podcast? Is it possible to buy one versus the other? And then, if there are older podcasts (that maybe you had a mid-roll in when it first ran) is your ad still there? If it is, is anyone quantifying these long-tail exposures? Does the vendor break out downloads of old versus new podcasts? Does the vendor distinguish how your ad gets placed in this case? So, say for example that they estimate 125,000 downloads of a given podcast. You’ve purchased 100,000 downloads, but the new episode only got like 75,000 – but older episodes combine for another 50,000. Are they going to make up the difference from the older episode downloads – or worse, split out evenly between the old and new? Should we treat the older episodes as de-valued relative to the new? I know that the jury’s largely still out, and that the answers vary from vendor to vendor, but you can bet I’m asking these questions now.
What is the basis of your purchase?
Odds are you’ve purchased your podcast on a CPM basis. But what are those thousands? Listeners (people who actually listen to the whole podcast end-to-end)? Or downloads? Or streams? We’re finding a big grey area on this front, where mostly what you’re purchasing are downloads based on an average number of downloads a show tends to deliver. There are varying degrees of opacity with regard to delivery – few vendors offer up much in the way of verification beyond the number of downloads, and almost none can tell you how many people actually heard your ad. This is where it becomes more like traditional (analog) media – you’ve purchased an “opportunity to hear” – not an actual exposure. But this is a digital medium, right? It seems like someone out there should know if your ad at least got played. Is there not a tag that I can associate with my creative that fires when played? Maybe there should be.
How are you defining success?
We have to be honest with ourselves, too. What constitutes success for a podcast advertiser? If it’s reach, you can see the shortcomings of the existing model/marketplace from the questions above. If it’s awareness, it seems like you’d need a robust way to canvas exposed audiences. For a lot of people (my clients included) it’s response. You hear a lot of “unique” URLs and URL extensions in podcasts. Some advertisers successfully deploy offer codes. Both are acceptable workarounds to capture response, but we all know that audiences are sophisticated – they routinely bypass these mechanics, so we’re probably not seeing the full response picture. Again – this is a digital medium. Shouldn’t I be able to see when people hear my ad? Shouldn’t I be able to know what device they were on? Shouldn’t I be able to tie back an exposure to a later action? I’m not saying that this should be easy, and I’m not even certain that the ability to do these things doesn’t exist. But I do want this to be the norm, and as near as I can tell this is not currently the case.
We used to characterize online advertising as the “wild west” but the border of the frontier appears to have shifted, and the lessons we’ve learned from that part of our history should be brought to bear here. Podcasts are the epitome of an on-demand media world where audiences dictate the terms of engagement and I am not interested in diluting that engagement and experience: as a consumer podcasts are awesome! That’s precisely why I want podcasting to survive and thrive, and that’s only going to be possible if those of us underwriting this channel understand its strengths, benefits and utility. And we only get that by challenging and improving it.