Hipset. It’s the stupid algorithm, stupid.
This weekend, the world’s gone gaga for Hipset.
It’s a cool service. Essentially, it delivers a stream of content that’s been socially shared by bands you’ve “liked” on Facebook.
The chances are, you would miss this content by looking at your newsfeed on Facebook.com. When a band (or anyone else) shares something on Facebook, less than 20% of their fans/followers see it. That’s because of EdgeRank - the algorithm Facebook uses to determine what content you see.
Facebook doesn’t show you all the content that’s aimed at you - from new videos posted by bands you like to status updates broadcast by friends. All that content would be way to noisy and impossible to consume. So instead, EdgeRank figures out “the best stuff” and shows you only that.
So it follows that lot of shared music-related content is missed. This sucks for the bands trying to communicate with their fans. And it sucks for music-mad users who would love to see this stuff. Enter Hipset. Now you’ll never miss a piece of content shared by a band you like.
In other words, the problem Hipset addresses is that “Facebook’s EdgeRank algorithm is wrong”. Or, more specifically, for the subset of users who are music-mad, and value content shared by bands over friends status updates… the EdgeRank algorithm is wrong. And Hipset makes it right.
This got me thinking about the work we did at OneRiot on so-called “realtime search”.
I remember the day Michael Jackson died. The social web - much smaller than it is today, but still a significant population in the hundreds of millions - went crazy. Huge numbers of people were sharing stories on Twitter, submitting news articles on Digg, posting videos on Facebook, etc etc. If you could take “the pulse of the web” at that point in time, there was an emotional explosion of MJ-related content being shared by millions of people. And even more people wanted to find that content: to engage with it, to react to it, to share it even more. But if you did a search on Google right then for “Michael Jackson” the top result was the official Sony Music artist page for Michael Jackson. Not the content you wanted to find.
In other words, “Google’s PageRank algorithm was wrong”. Or, more specifically, for the large number of users who at that point in time were MJ-mad, and valued content being shared by other emotionally charged people over well-SEO’d record label blandness… the PageRank algorithm was wrong. And OneRiot made it right.
We build a search engine that re-ranked search results based on what was hot on the social web at that point in time. We fixed the algorithm. Search for “MJ” on OneRiot that day, and the top result was the latest CNN video with new details on the emerging story. Search for “Hudson River” on Jan 15 2009, and the top result was a picture of a plane crash, not a Wikipedia Page earnestly explaining that the Hudson River is 315 miles long (which is what you got on Google then, and still get now). Search for “Tour de France” this summer and you would have seen eulogies to Team Sky and the winner Bradley Wiggins. But you couldn’t, because OneRiot realtime search didn’t exist any more. We killed it in July 2010.
We never got real traction as a stand-alone destination site. And neither did any of the competitive services who all had the similar ideas. Evidently, not enough users wanted realtime search enough of the time. To them, PageRank wasn’t broken. It was just a bit crap for certain use cases. And that wasn’t enough to make OneRiot a winner in search. (Note, subsequently we pivoted to a mobile content targeting network and sold to Walmart).
Which brings me back to Hipset.
For Google read Facebook. For the Search Results Page read the Social Stream. For PageRank read EdgeRank. For OneRiot read Hipset. For “the pulse of the web” read “content shared by bands you like”. Except this time i think it’s different. And this time i think that Hipset will win.
At OneRiot, we fell at three major hurdles… that i think Hipset will clear.
1 - User Acquisition.
We launched in an era pre-“growth hacking”. For example, the first incarnation of Facebook Connect launched after we did, and the Open Graph was years away. User Acquisition meant working two channels. A) marketing - expensive, and a complete waste of time when you’re competing for “search mindshare” with the likes of Bing and Google. B) Biz Dev - we ran a really successful distribution program to get our search results to appear on other search engines, including Yahoo’s (ie “powered by OneRiot”, with attribution links back to our destination site). But the number of users we acquired through this program was still peanuts in comparison to Google. Basically, we could never acquire enough users to be a success as a search engine.
For Hipset, user acquisition is very obviously built into the product’s and team’s DNA. A well executed PR exercise this weekend generated a ton of interest from early adopters - who all signed up via Facebook auth. Meanwhile, the product aggressively (that’s a complement) integrates with Facebook’s Open Graph. Every time i engage with content on Hipset, it gets posted to Facebook. So all those early adopters using the product are actually busy sending notifications to their impressionable friends on Facebook about this cool new service. And, of course, those friends click, join, start sending notifications to their impressionable friends and… the cycle repeats. Easy! (As long as the product rocks, which in this case it does).
In summary: for user acquisition, OneRiot faced a big marketing budget or a lot of lengthy BD hustle. Hipset have a smartly executed PR push and the Open Graph - and they are working it to perfection. (note: I’m ignoring any growth factors built into our respective revenue models - this post is long enough already ;)
2 - Most users generally want all the content, not just a specialized slice.
OneRiot’s top search result showed you “the pulse of the web”. In the Hudson River case mentioned above, on that particular day, OneRiot rocked. But what would you use OneRiot for the next day? Most of the time, you actually wanted everything that Google provided. Until the next disaster stuck or the next celebrity died. But by then you’d forgotten about us. Because, in the desktop browser paradigm - which is where we were - people never kept a tab open for OneRiot “just in case”. That was too much real estate wasted, or to much of a hassle. In the desktop browser paradigm, having all the desired content in one place is a winning strategy. Google, mostly, did that. We only provided a specialized slice. We didn’t win.
Hipset need to be careful here. Right now, they have a destination site in the desktop paradigm that provides a slice. They show me all the content bands I like have shared, but not my friends status updates. Facebook might only show 20% of the content aimed at me, but it’s “the best stuff” and so generally feels ok. I going to go to Facebook.com more than I will to Hipset.com. And I’m probably ok with only seeing 20% of my liked bands’ shared content. And, eventually, i’d forget that Hipset.com was there.
Will Hipset suffer in the same way we did - i.e. by showing the specialized slice rather than the generalized whole? Being to Facebook what OneRiot was to Google. I don’t think so… if they can figure out mobile fast.
Albert Wenger has an excellent post on how mobile is “unbundeling” the content we used to find on one site in the desktop browser paradigm. As he says: “On my phone another app is just a button push away and there is relatively little that fits on each screen. So it is just as much effort to go to another part of the Facebook app as it is to go to a different app altogether. So ‘Facebook for mobile’ may not be Facebook at all but rather a combination of say Instagram, Kik, Twitter, Foursquare and others”. Others like Hipset for music content.
Hipset doesn’t seem to have much mobile DNA. There’s no mobile app. Not even a mobile-optimized website that i would be willing to “Add to Home Screen” as a web app. Of course, they have just launched. Mobile may be around the corner. But i hope it is. With a mobile app, they would be on my phone’s home screen. It would be easy for me to dip in and out of their content stream - it would be just one button-push away. Push notifications for new content would pull me back even more regularly. But if Hipset stays in the desktop browser paradigm, i will probably forget it.
In summary: it’s hard to become a successful destination site in the desktop browser paradigm if you’re “just” presenting specialized content slices of the general whole. However, that approach works really well on mobile. Hipset dont need to clear the same hurdle OneRiot faced. Instead, they should sidestep it completely with a mobile product.
[note: What Hipset does have in its favor is email outreach to drive return usage. According to this coverage in Forbes, a band i’ve liked can now email me if i’ve signed up to Hipset. Assuming this content is valuable (and i dont get spammed with crap, leading to an unsub) i can see this working. At OneRiot, we tried something similar with email alerts and RSS feeds for search queries. This meant users could subscribe to a realtime stream of top search results for specific queries as they changed with the ebb and flow of the social web. But user adoption was very small. I hope Hipset’s execution is better than our was - let’s see.]
3 - The big boys saw what we did, then built their own version.
In Oct 2009, both Google and Microsoft announced deals with Twitter. That gave them access to the data that could power realtime search - to help them re-rank their algorithms to reflect the pulse of the web. And they both went ahead and built what we’d built. (For the curious, i’m not going to divulge anything about conversations we might have had with one or both at the time, but suffice to say they decided to build vs buy ;).
Although we faced a user acquisition problem, OneRiot had enough users to prove the point that there was pent up demand for realtime search results - when delivered at the right time. So Google and Bing got busy building “slot 1 inserts” - which would automagically inject realtime results at the top (“slot 1”) of a “usual” search results page when the licensed social data suggested something unusual was happening right now (a plane crash, a celebrity death, etc). Although what OneRiot did was technically difficult, for Google it was relatively easy to replicate - after all, they had hundreds of search engineers when we had 10. Once Google replicated what we were offering - and could provide that “slice” at the right time to millions of people already glued to their destination site every day - we were toast. Said another way, for users wanting to know the pulse of the web on certain occasions, Google’s PageRank algorithm was now magically right. Why go anywhere else? (Twitter is the exception that proves the rule - millions of users were already glued there too).
So if Hipset gets traction, won’t Facebook simply replicate what they have built? I don’t think so.
There are two reasons for this. First is that Facebook has a pretty poor record of replicating “slices” at the heart of other single-purpose social services. Think about Facebook checkins? Did you stop using Foursquare when that was released? But perhaps more important is the second reason - which is that Hipset contributes to the Facebook experience. Using Hipset means i am posting more (music-related) content in to stream, giving more of my friends more to engage with… which, of course, generates more opportunity for Facebook to monetize us.
I’m not even going to fantasize that OneRiot was any threat to Google. But no matter how small our size, we were taking searches away from Google - searches they couldn’t monetize any more. Far from adding value to Google, and giving them more opportunity to monetize, we were taking away. They had to swat that gnat. Conversely, Facebook should be encouraging Hipset - for contributing content to the Facebook stream, and for showcasing the power of Facebook’s platform tools and Open Graph which can only encourage more developers to do the same (and, in turn, contribute even more to the Facebook ecosystem).
In summary: OneRiot worked against Google and, theoretically, could have reduced monetization opportunities. So they killed us. Hipset works with Facebook, contributing to Facebook’s experience and increasing opportunities to monetize. That behavior should only be encouraged.
So, what happens next?
I like Hipset. The consumer-facing product rocks (again, i’m not contemplating the business model here). They are leveraging smart PR and Open Graph for a winning user acquisition strategy. And I don’t see them being chopped at the knees by Facebook, because they are adding value to that ecosystem rather than competing and taking away. I just hope they hurry up and deliver a mobile experience, otherwise they risk being that tab that’s never opened in a desktop browser paradigm.
It appears that the team behind Hipset (Tracks.by) is exclusively focused the music industry. So i expect others to fill the void and create Hipset-like services for additional content verticals - “Hipset for Sports”, “Hipset for Celebs”, etc. (Again, to borrow some thinking from Albert, Facebook have already moved in this direction with their Camera app - the consumption experience is “Hipset for Photos”.)
And I’m fascinated to see how the business models for these new services pan out. I’m assuming they’ll go for a native ads model - paid content to appear on my Hipset page, or in an email.
But let’s not get ahead ourselves. For now, let’s just celebrate a cool new service for music-content consumption. For music-heads everywhere, we can’t get enough of this!
