The other day I wore my “I’m not Matt Cutts” T-shirt around Silicon Valley. That’s always fun cause it gets weird stares. Sometimes people are interested enough to ask “who is Matt Cutts?” Where I usually explain he’s my favorite Google blogger.
Anyway, I just caught up on reading Matt’s stuff now that he’s back from vacation. He has an interesting post on metrics and how Google both was harmed and maybe benefitted from a company’s rating of Google.
This is a real important point for marketers to understand. If you’re going to use metrics to make decisions you need to understand the biases those metrics have. Not to mention that an influential user (which is what weren’t being measured by the company Matt is talking about) is worth a LOT more than one who doesn’t tell anyone, or isn’t seen as an expert by his/her friends.
Speaking of which, I wish I had better metrics at PodTech.net. I wish I knew how many listeners we REALLY have. Or, whether the people who download a file actually listen to it. Or, whether they listen to the whole file, or just part of it.
Podcasting and video podcasting won’t be taken seriously as businesses until we figure this stuff out. Advertisers want proof that their money is spent well.
Scobleizer 
Robert, it sounds like you’re describing a lot of the challenges of radio station ratings. Of course, radio has even more difficulty than podcasters in determining who’s listening, where, and for how long. And so much of their livelihood from selling advertising is based on the metrics generated by one company: Arbitron Co. However, radio has been wrestling with this for decades and there are surely solutions that can be applied to podcasting as well. For example, I know tactics like “100th call in contests” can tell a radiostation a lot about their listeners. Could a message like this in a podcast work? Maybe. Certainly there’s fun and opportunity in experimentation within this area.
LikeLike
Robert, it sounds like you’re describing a lot of the challenges of radio station ratings. Of course, radio has even more difficulty than podcasters in determining who’s listening, where, and for how long. And so much of their livelihood from selling advertising is based on the metrics generated by one company: Arbitron Co. However, radio has been wrestling with this for decades and there are surely solutions that can be applied to podcasting as well. For example, I know tactics like “100th call in contests” can tell a radiostation a lot about their listeners. Could a message like this in a podcast work? Maybe. Certainly there’s fun and opportunity in experimentation within this area.
LikeLike
“Podcasting and video podcasting won’t be taken seriously as businesses until we figure this stuff out. Advertisers want proof that their money is spent well.”
That seems rather an extreme statement isn’t it? Surely advertising (and the metrics associated with it) has always been a very fuzzy art where no amount of statistics is going to give you absolute certainty about anything.
In fact, I’d have thought the metrics for podcasting are currently a damn sight more accurate than other advertising streams – whether traditional (TV/magazine) or more text-based stuff (e.g. how many people actually read from 1 RSS feed fetch? Dunno, but I bet a high percentage of completed podcast downloads get listened to)
Also, you’ve got to balance your desire for stats against your users desire for privacy – so in practise you’re always going to have to live with the imperfections that Matt points out …
Now excuse me while I go and download 7.23 different podcasts and half-listen to a couple of them 🙂
P.S. Am I right in thinking that you haven’t got a single link to podtech.net on your site? Is there a reason for that?
LikeLike
“Podcasting and video podcasting won’t be taken seriously as businesses until we figure this stuff out. Advertisers want proof that their money is spent well.”
That seems rather an extreme statement isn’t it? Surely advertising (and the metrics associated with it) has always been a very fuzzy art where no amount of statistics is going to give you absolute certainty about anything.
In fact, I’d have thought the metrics for podcasting are currently a damn sight more accurate than other advertising streams – whether traditional (TV/magazine) or more text-based stuff (e.g. how many people actually read from 1 RSS feed fetch? Dunno, but I bet a high percentage of completed podcast downloads get listened to)
Also, you’ve got to balance your desire for stats against your users desire for privacy – so in practise you’re always going to have to live with the imperfections that Matt points out …
Now excuse me while I go and download 7.23 different podcasts and half-listen to a couple of them 🙂
P.S. Am I right in thinking that you haven’t got a single link to podtech.net on your site? Is there a reason for that?
LikeLike
David: good point. The problem is that I don’t have control of my sidebar. Soon, I hear, soon.
Raffy: oh, that’s a good point. Buzz Bruggeman actually tried something like that at the end of a one-hour-long Podcast. He offered free versions of his software. Got hundreds of requests, if I remember right. So, that’s certainly one way to make sure that people watch all the way through.
LikeLike
David: good point. The problem is that I don’t have control of my sidebar. Soon, I hear, soon.
Raffy: oh, that’s a good point. Buzz Bruggeman actually tried something like that at the end of a one-hour-long Podcast. He offered free versions of his software. Got hundreds of requests, if I remember right. So, that’s certainly one way to make sure that people watch all the way through.
LikeLike
Robert: And you could extend a method like that and place contest announcements halfway through the podcast..or right in the beginning. How much would that increase the number of entries? Repeating this over a long enough amount of time and you’d be able to make some informative statements about the data.
LikeLike
Robert: And you could extend a method like that and place contest announcements halfway through the podcast..or right in the beginning. How much would that increase the number of entries? Repeating this over a long enough amount of time and you’d be able to make some informative statements about the data.
LikeLike
I think the problem with applying the “100th caller” type deal to podcasts is the environment they exist in (blogs, other podcasts, lots of social software). Some of the people who listen and hear the offer are going to blog it, tell their friends, etc. That may skew the data a bit. Radio stations don’t really have this problem because the timeframe from 100th caller contests is a lot shorter. By the time you tell someone, it would be over.
That said, there are some interesting things you could do here. The simplest is just getting some basic demographic info from people when they claim their prize (giving you information about your audience). For a more complex example, include a unique alphanumeric code in each download (this would take some special work on the distribution end) and ask users to enter the code to get their free stuff. You could compare when they entered the code vs. when that particular file was donloaded (establishing an outer bounds on how long it takes a listener to listen to a program once they downloaded it). You could also see how many duplicates you got (though you’d have to exclude any numbers that got posted on websites) telling you how many people are sharing downloaded files.
LikeLike
I think the problem with applying the “100th caller” type deal to podcasts is the environment they exist in (blogs, other podcasts, lots of social software). Some of the people who listen and hear the offer are going to blog it, tell their friends, etc. That may skew the data a bit. Radio stations don’t really have this problem because the timeframe from 100th caller contests is a lot shorter. By the time you tell someone, it would be over.
That said, there are some interesting things you could do here. The simplest is just getting some basic demographic info from people when they claim their prize (giving you information about your audience). For a more complex example, include a unique alphanumeric code in each download (this would take some special work on the distribution end) and ask users to enter the code to get their free stuff. You could compare when they entered the code vs. when that particular file was donloaded (establishing an outer bounds on how long it takes a listener to listen to a program once they downloaded it). You could also see how many duplicates you got (though you’d have to exclude any numbers that got posted on websites) telling you how many people are sharing downloaded files.
LikeLike
Radio is not alone in this. The way advertising works on TV is hogwash as well. Nielsen is hardly accurate. I had a bit more faith in it until I was part of the Nielsen system for a few weeks. A little booklet where I indicate when the TV was on and what I watched? Fancy stuff. And “sweeps week” is an idiotic idea as well. Why should an advertiser have to pay $100,000 for an ad on the 8pm Monday time slot just because the premiere of Lost was during sweeps week?
It gets even worse in magazines. The idea there is just to make the prettiest picture and hope it gets stuck in the readers’ minds. It doesn’t necessarily matter if they read the ad as long as it makes an impression.
Anyway, IMO, website metrics and podcast metrics are already far more accurate and detailed than TV, print or radio. Robert, I’ve had the same thoughts you had in this blog post when dealing with my sites/podcast, but once you really sit down and think about it, you realize how much more you know doing internet content. For instance, I can tell you what percent of my viewers use iTunes to get the content. How many TV stations can tell you how many people get their content in HD (without guessing) or how many use a Sony TV?
I think advertisers like to pretend they know a lot. After all, they can make a lot of pretty charts to back up their knowledge. But in reality, it’s very fuzzy and very unscientific. That’s why many ad departments measure share of voice. The theory there is that as long as the customer is more likely to see their ads than the competition, it doesn’t matter about how many people actually see your ads.
LikeLike
Radio is not alone in this. The way advertising works on TV is hogwash as well. Nielsen is hardly accurate. I had a bit more faith in it until I was part of the Nielsen system for a few weeks. A little booklet where I indicate when the TV was on and what I watched? Fancy stuff. And “sweeps week” is an idiotic idea as well. Why should an advertiser have to pay $100,000 for an ad on the 8pm Monday time slot just because the premiere of Lost was during sweeps week?
It gets even worse in magazines. The idea there is just to make the prettiest picture and hope it gets stuck in the readers’ minds. It doesn’t necessarily matter if they read the ad as long as it makes an impression.
Anyway, IMO, website metrics and podcast metrics are already far more accurate and detailed than TV, print or radio. Robert, I’ve had the same thoughts you had in this blog post when dealing with my sites/podcast, but once you really sit down and think about it, you realize how much more you know doing internet content. For instance, I can tell you what percent of my viewers use iTunes to get the content. How many TV stations can tell you how many people get their content in HD (without guessing) or how many use a Sony TV?
I think advertisers like to pretend they know a lot. After all, they can make a lot of pretty charts to back up their knowledge. But in reality, it’s very fuzzy and very unscientific. That’s why many ad departments measure share of voice. The theory there is that as long as the customer is more likely to see their ads than the competition, it doesn’t matter about how many people actually see your ads.
LikeLike
Addendum to that share of voice thing. Take godaddy.com. They have 100% share of voice for their products on podcasts. You’ll only ever see their ads on podcasts right now, so it’s a 100% successful ad campaign in that regard.
LikeLike
Addendum to that share of voice thing. Take godaddy.com. They have 100% share of voice for their products on podcasts. You’ll only ever see their ads on podcasts right now, so it’s a 100% successful ad campaign in that regard.
LikeLike
THere is no way Podcasting companies can extract the data their advertisers want. It’s just not possible.
The technology doesn’t need to change here, advertisers do. An advertiser needs to change how they measure the “effectiveness” of a Podcast. For example, they can see how much it is blogged. They can see how much it is downloaded. They can monitor the pulse of the Internet and get a good sense of “nobody is watching this” or, “People are paying attention”. This is a tough shift for MSM to make. There is NOTHING tangible in this new economy. Advertisers need to just figure out how to measure that – I don’t think there is much Podcast distributers can do to change things (besides creating a closed Podcasting Network with a proprietary Media Player that “phones home” and tell the Podcasting company that I listened to 37 minutes of a 68 minute podcast on X, Y, or Z). I don’t think users will but into that. Not even novice ones. I sure wouldn’t.
The marketing/revenue model problem that companies like PodTech.net have will NOT be solved by companies like PoTech – unless they stop thinking about things like, “How can I give these advertisers the data they want” and start thinking (and acting) on things like, “How can I educate these companies to think differently, and use a new set of metrics”.
Until that happens, I am listening to my Podcasts – blistfully (mostly) ad-free. I don’t think that will last.
Rob
LikeLike
THere is no way Podcasting companies can extract the data their advertisers want. It’s just not possible.
The technology doesn’t need to change here, advertisers do. An advertiser needs to change how they measure the “effectiveness” of a Podcast. For example, they can see how much it is blogged. They can see how much it is downloaded. They can monitor the pulse of the Internet and get a good sense of “nobody is watching this” or, “People are paying attention”. This is a tough shift for MSM to make. There is NOTHING tangible in this new economy. Advertisers need to just figure out how to measure that – I don’t think there is much Podcast distributers can do to change things (besides creating a closed Podcasting Network with a proprietary Media Player that “phones home” and tell the Podcasting company that I listened to 37 minutes of a 68 minute podcast on X, Y, or Z). I don’t think users will but into that. Not even novice ones. I sure wouldn’t.
The marketing/revenue model problem that companies like PodTech.net have will NOT be solved by companies like PoTech – unless they stop thinking about things like, “How can I give these advertisers the data they want” and start thinking (and acting) on things like, “How can I educate these companies to think differently, and use a new set of metrics”.
Until that happens, I am listening to my Podcasts – blistfully (mostly) ad-free. I don’t think that will last.
Rob
LikeLike
Wondering if iTunes keeps track of what people actually listen to & for how long, once they download it (when you connect your iPod back up the next time you want to download). I believe they can & do (thinking of the “most played” function)… who knows if they would provide those types of metrics to the content providers…maybe at a cost?
LikeLike
Wondering if iTunes keeps track of what people actually listen to & for how long, once they download it (when you connect your iPod back up the next time you want to download). I believe they can & do (thinking of the “most played” function)… who knows if they would provide those types of metrics to the content providers…maybe at a cost?
LikeLike
You should check out Audible’s Wordcast. It solves your problem of measuring your audience – wordcast.audible.com
LikeLike
You should check out Audible’s Wordcast. It solves your problem of measuring your audience – wordcast.audible.com
LikeLike
Sounds like a cool t-shirt.
Did you make it yourself?
Could you post a photo of it?
I want one 😉 but maybe with Scoble instead?
André Hedetoft
Movie-geek
Blogging geek porn over at
http://www.andrehedetoft.com/geekporn
LikeLike
Sounds like a cool t-shirt.
Did you make it yourself?
Could you post a photo of it?
I want one 😉 but maybe with Scoble instead?
André Hedetoft
Movie-geek
Blogging geek porn over at
http://www.andrehedetoft.com/geekporn
LikeLike