Installed Open Web Analytics for WordPress this afternoon. Just a few hours into it, it’s giving me data I haven’t seen since Urchin
Thursday, 31 May 2007
Thursday, 5 April 2007
There’s No Accounting for Taste
From my perspective, transparency is about being up front about biases. Objectivity is an unachievable. Covering all sides of anything equally and without a subjective adjective is not only futile, it makes for a boring read. The prerequisite for making anything interesting is a perspective, a slant, some reason to care. Without that reason, why bother?
I like pizza.
One of the most popular posts here is my two year-old review of Punch Pizza. As you can glean from the title, I didn’t have a great time. I said so. Nine people came by, some agreed, others disagreed, ‘CDSIII’ wrote a very lengthy piece arguing Punch was the best pizza in the nation. While I’m very grateful for the comments, they don’t change my less than superb experience that evening. Nor did that single experience keep me from giving my money to Punch in the 2 years since.
I don’t search out objectivity. Nor do I expect it from anyone I read. Whether I’m looking for the best French Toast in the Twin Cities, a good wine for dinner, or anything else. I only care about really good or really bad experiences, and I search out lots of them (objectivity through aggregation).
So Tim, rate and review what you’re interested in. I know you like fruit-forward wines. If a wine you’re involved in isn’t fruit-forward, I trust you won’t rate it very highly.
Tuesday, 27 February 2007
Ratings, Scores, Context and Sneezing
A while back, I watched Never Been Thawed a 90-minute, independent, very comprehensive, mockumentary. There was a lot going on in the movie. Lots of different story lines and highly-developed character relationships. More so than I’ve seen in many a documentary (the non-mocking kind). No, NBT isn’t a Spinal Tap or a Best In Show, but it is a good effort in that direction by a lesser known cast.
Now bounce all that off your taste and everything else you know about me. Do you think you’ll like the movie? Do you even have enough information to say?
The problem is single-attribute rating/ranking/tagging systems. They provide only the most basic value – and always with a caveat. Case in point – searching the web for any single keyword rarely provides anything useful.
Tim’s rethinking his wine rating system – maybe something more detailed albeit less glance-able.
Over the weekend, I found the ‘Origins’ wine label on store shelves – Temparnillo, Malbec, (2 of my favorites) and other varietals from around the world – all under the same label. On the back, 3 scales indicate where a specific varietal falls on the flavor scale. Three-times the information than a single score.
While even 3 scales can’t replace the wealth of information found in personal recommendations, it’s more than an arbitrary number or letter. But, what were the biases and preferences of the person that declared where a specific wine fell on each scale?
In nearly each episode of Winecast, Tim discloses his preference for “fruit-forward” wines. If Tim and I share the same definition of “fruit-forward”, I have some context for determining if a ranking of ’83’ is something I’ll like or dislike. If I don’t know if we share the same definition – I pick up that specific wine and see if my tasting matches his. And repeat.
All this metadata is why recommendations from someone you know really well is far more valuable and those from Amazon, Netflix, etc just aren’t.
Yes, this is word-of-mouth, and yes, we are all doing this today when we IM a YouTube link. There’s got to be a better place to sneeze.
Saturday, 23 December 2006
Aggregate Trust, Filter Relevance
“I want to subscribe to other people who I can trust, and also, who I can dial down a little” – Eric Rice
I like and trust Eric Rice…though I’m not as psyched about Second Life as he is. Same with Mark Cuban and basketball. Same with Doc Searls and photography. Same with Dave Slusher and SciFi. I subscribe to many others where there’s just the occasional interesting bit – that I gotta dig for.
Our current RSS aggregation tools don’t handle this problem very well. In fact – RSS is very different than email (just as radio is different than voicemail). All but 1 of the aggregators (share.opml.org ) I’ve played with treat them the same.
“We have always measured, instead, relevance, trust, usefulness, interest, attraction, action, value. Those are the measurements that matter, always have been, only now media must catch up to us.” – Jeff Jarvis
“I’m happy to let her sift through the left (and right) blogosphere for me and pick out the gems, that way I don’t have to read either.” – Dave Slusher
Tuesday, 14 November 2006
Measuring What You Can’t Automate
“[I] think how much better it would be if we could just measure how much people care.” – Dave Slusher
Like Dave, I don’t understand the fascination with measuring downloads. Well, I take that back – I understand it for producers trying to woo advertisers. I don’t understand why advertisers would want to base their ad buy on download stats. Downloads don’t equal listeners, fans, or impressions.
Requests for downloads are not full downloads.
Full downloads are not plays.
Plays are not listens.
Listens are not engaged.
Engaged are not customers.
And as Dave points out, download requests can be automated.
Kris Smith’s CastLock application provides unique feed urls and could be spun out to deliver a custom, complimentary ad (or other) message to individual subscribers – based on some measure of engagement (i.e. some bastardized quantification of caring).
As early-stage as it is, it still provides more useful metrics than download stats. Mapping individual listeners to customer purchases still needs some work, but the gap would be shorter.
The real question is – what’s the Effort/Engagement ratio of a publication like a podcast or weblog. I’m glad you’re reading this, and I’m glad you know who I am. That’s return enough for me.
“Any website that attempts to improve time spent on every page (or pageviews for that matter) is just wasting time. What matters is intent. Permission. Action. Retention. Likelihood that ideas get spread. Clickthroughs.” – Seth Godin
Tuesday, 17 October 2006
Lowering the Barriers of Connecting Podcasters and their Listeners
I wrote this article earlier this year, and with the original site transformed into something completely different, I thought I’d put it here for us to enjoy for years to come.
Podcasting has the same problem as television, radio, and that legendary tree falling in the forest. There’s no accurate way to measure the number of listeners.
That’s not to say there aren’t metrics. There’s are plenty of things to count – each with it’s own issues. Counting pageviews excludes listeners that only pay attention to the RSS feed. Counting hits to the RSS feed also includes directories and aggregators. Counting downloads of a specific mp3 file then shaking out duplicate requests from the same IP address and failed attempts just might provide the most accurate number of downloads – for that specific podcast.
Despite the effect this number has on your monthly bandwidth bill, it still says nothing about the number of people actually listening.
Emails, weblog comments, audio comments, trackbacks, a mention in another podcast are all great methods for measuring the hardcore fans – the people deeply affected by a podcast to act – post-listen. Arguably, these are the only listeners worth caring about. These are the listeners more apt to spread the word about you – to yet to be fans.
Personally, I’m a big – perhaps not hardcore – fan to a number of podcasts and I don’t regularly invest that level of communication. Partially because not every podcast compels me to comment and partially because it’s not always easy or convenient. Eric Larson, from the Ericast mentions the comment line phone number. I’ve called and left a message more than once. Other times, not a all. Does that mean I wasn’t there to hear it?
As a listener and podcaster, I don’t think so. Neither does AttentionTrust.org.
Why does the number of listeners matter?
Traditionally, this is what advertisers mean by ‘reach’ – the number of people exposed to a specific message. Even in the comparatively mature fields of television and radio, counting the audience any more specifically than a millions is a shot in the dark. In the internet world, browsers knows where they’ve been (browser history) and the sites visited have a similar record (server logs). AttentionTrust.org believes if more than those 2 parties aren’t involved today, they will be shortly. Though this attention data is valuable to advertisers, competitors, researchers, in addition to the the site owner and the visitor, the visitor needs to have control over how it’s shared and what level of identifiable. All in a well-formed, easy-to-share XML document.
So, advertisers care. Not everyone cares about advertisers.
A podcaster’s audience is made up of 2 groups of listeners – those driven to invest the effort of crafting and send a response, and those not. Any given listener alternates between action and apathy continually. Within the apathetic group, some will never signal their existence, others just need a method of communication with a lower barrier. This is where someone like the AttentionTrust comes in – providing tools to reduce the barriers of involvement.
Flip on the AttentionTrust AttentionRecorder and your browser clickstream history can be sent to RootVaults service and the default ACME Attention Service, or stored locally (redundant to your browser history). As of this writing, the AttentionRecorder captures HTTP traffic from the Firefox browser. In the audio realm, Last.FM’s iTunes plugin will send everything you play up to their database, where you can share it with others and find new and similar things. Like the Podcasts.Yahoo.com directory, the iTunes podcast directory, PodcastAlley, and so many others, there’s someone missing from the equation. The podcaster themselves. Each one of those services offers ratings and commenting – each distinct, redundant, and independent from the podcaster’s own site that may very well have both. As Dave Winer said about more than one RSS-like format, “two is more than twice as bad.” Each additional service shoe-horning themselves between saving listener responses increases the burden on the podcaster and the confusion for the listener.
The First Crack Podcast already has commenting & trackbacks – standard in the WordPress system and a 5-star rating system (using Owen Winklers Votio plugin
So, let’s build a little something to reduce the level of investment required to connect listeners and podcasters. Something that reuses existing information.
First off, like your web browser, iTunes (what i use for my podcast listening – though not for receiving) records what’s played and when. That’s how it populates the ‘Recently Played’ and ‘My Top Rated’ playlists. On the Mac OS X side – all the information iTunes stores is accessible through Applescript.
To access all the information about the podcasts you’ve recently listened to, open up your Script Editor and enter in:
tell application "iTunes"
repeat with theTrack in (get every track of playlist "Recently Played")
if (genre of theTrack) is "podcast" then
set theAlbum to (album of theTrack)
set theArtist to (artist of theTrack)
set theTitle to (name of theTrack)
set theRating to (rating of theTrack) as string
set theDate to (played date of theTrack) as string
set theCount to (played count of theTrack) as string
display dialog "Podcast: " & theAlbum & "
" & "Podcaster: " & theArtist & "
" & "Episode: " & theTitle & "
" & "Your Rating: " & theRating & "
" & "Last Played: " & theDate & "
" & "Times Played: " & theCount
Now, we can use Applescript to connect the iTunes ratings to the Votio plugin in WordPress. To do that, we need some common bit of information between the mp3 file and the podcast’s website. What I’ve done is put the Votio rating URL in the ID3 tags ‘grouping’ field.
"http://firstcrackpodcast.com/wp-content/plugins/votio.php?id=297&index=" for example.
tell application "iTunes"
repeat with theTrack in (get every track of playlist "Recently Played")
if (genre of theTrack) is "podcast" and (grouping of theTrack) contains "votio" then
set theRating to ((rating of theTrack) / 20) as integer
set theDate to (played date of theTrack) as string
my sendRating((grouping of theTrack), theRating)
on sendRating(theURL, theRating)
open location theURL & theRating
Running this script incrementally updates the average rating and total votes (listens) on FirstCrackPodcast.com for the individual episode.
In the end, it provides a very low barrier, low commitment way for listeners to tell podcasters what they’ve heard, and what they like on a per episode basis – on the podcasters own site (where this information belongs).
There are a number of ways to refine and extend this script – like integrating it into Jon Link’s iRate Dashboard widget or Last.FM’s iTunes plugin. Things are moving so fast, by the time you read this – those extensions might exist. Or, an altogether more sophisticated method.
Thursday, 31 August 2006
Pageviews Dead: Killed by RSS, AJAX, Widgets
I’ve stopped tracking my web stats. I’ve only got one site left on the useless Google Analytics and haven’t visited the reporting page in forever.
Incoming links and comments are the useful measures to me. Those tools are nowhere as mature as they need to be. Then again, I’m not looking for investors, advertisers, or the press. I’m looking for a way to share my thoughts with you on your terms.
Ev has the same complaints about our current metrics as I mentioned in the above rant:
“But Ajax is only part of the reason pageviews are obsolete. Another one is RSS.”
Also, this beautiful dig on MySpace.
“…part of the reason MySpace drives such an amazing number of pageviews is because their site design is so terrible.”
A little more eloquent than this:
Wednesday, 10 May 2006
Share You OPML, Exposing 1% of Your Audience
According to the Feedburner widget 53,657 people are subscribed to TechCrunch.com. 746 of those people are also sharing their opml.
That’s 1.4% overlap.
If you’re running a tech startup, thanks to Share Your OPML, you now know which 746 people to talk with first.
Coincidently, I’ve talked to a bunch of ecommerce, direct-to-customer organizations over the years and 1% is the most frequent conversion rate quoted. On the one hand, it’s pretty cool that successful businesses can be built on 1%. On the other, it feels highly inefficient (99% waste).
If you’re interested, I’ve shared my feeds as well.
Monday, 8 May 2006
Down and Dirty Listener-Centric Podcasts Metrics Thoughts
The article I whipped up for ID3 Magazine on a listener-to-podcaster metrics method is up; Trees Falling in the Forest: Lowering the Barriers of Connecting Podcasters and their Listeners.
In it, I use Applescript to connect iTunes to WordPress as a way for listeners to rank and send rankings of individual podcasts to the podcasters.
Monday, 17 April 2006
Now You Know I’m Listening To Your Podcast
I’m in the early stages of a couple podcast listener measurement methods. Methods that put the control in the hands of the listener – not the podcaster or a third-party – and the metrics in the hand of the podcaster, not a third-party.
I’m trying one of the methods out with a small handful of my favorite podcasts. So, if you’re one of those podcasters – you now know for sure that I’m subscribed.