Ross Mayfield's Weblog

Promoted Tweets and the Next Ad Recovery

A lifetime ago I worked on an Eastern European telecom group's IPO and one of my favorite slides in my deck was borrowed from Goldman Sachs to paint the picture of a Multimedia Cycle. Yes, I said multimedia, but right now there is something shifting in the ad market that may or may not mean a recovery to yet another boom for the internet.

Picture this in a cycle:

Ads -> Content -> Applications -> Telecom Services -> Telecom Infrastructure -> Repeat

In 2002 I described the multimedia cycle like this,

"Just as potential customer spending drives advertising expenditures, Ad spending drives the development of content to create ad space.  Development of content drives spending telecommunications services to delivery it.  Development of telecommunications services requires investment in applications and infrastructure."

I also pointed out how GDP is correlated with ad spend and noted that ad spending drives internet recovery.

In 2004 there were signs the ad market recovered and the Web 2.0 boom was getting started. Google drove advertising innovation. To the point where way too many bloggers thought they could live off it. This ad recovery fueled a venture boom and falsely led many startups to leverage someone else's business model instead of finding their own. Many of course didn't make it through the crash, and those that did focused on creating their own leverage.

Ad revenue isn't what it used to be. And it won't be. Which may lead you to have concerns about the current recovery:

• The last boom didn't deliver innovation in new ad formats and metrics, and there still isn't a solid ad model for social networks like Facebook. Put simply, advertising still works for the old web. More on this later...
• Ad commoditization is moving far beyond text ads into other formats. The display ad market has moved to Real Time Ad Exchanges (disclosure: I'm an advisor to Triggit, which helps helps buyers optimize real time spend). Conversion rates always decline as the audience becomes sensitized, with the recent exception to newer formats like video ads (won't last long.
• The B2B lead gen business has been cutthroat as well. 
• CPMs, CPCs, CPEs & CPLs are falling through the floor. In part this may reflect the broader economy, but every last efficiency is being wrung out of the markets.
• What this means is what we've known. That traffic & engagement are less easy to monetize than during the last recovery. Or at the very least less than what was believed at the time.

This may lead you to believe the multimedia cycle will not be fueled, and internet recovery will stall the creation of the next wave of startups. But things have changed.

In Apple's Garden of Eden, you can skip advertising and directly monetize your content and apps. Subscription and ad models are ready for the taking as well. But this forbidden fruit drives your app down to $.99 and you only get 2/3rds. You find yourself in the hit business, and usually it ends up being Hollywood's studio model instead of an indie.

Startups already have to innovate their own ad models, and rise above them. When shit lobsters (a complement) like ^Dave McClure are calling for coupons, that's what he means.

The big case in point is Twitter's launch of Promoted Tweets, the launch of Bill Gross' TweetUp and other new ad formats for the real time web like MyLike. It's not just about Twitter finally finding a business model, built upon the problem they created for brands. They are testing a new ad format and ranking algorithm that hopes to strike a balance between advertiser and user. And the real test is if they gain distribution through their search and other ecosystem partners starting tomorrow @chirp.

I'm going to go out like a bird on a limb here to suggest that it may be another tipping point for Twitter. One that involves actual tips. And supports a good portion of the ecosystem. This will serve as a litmus test for ad formats on social networks.

But the key for Twitter is if they can gain a distribution advantage for how they monetize. The qualms of them buying partners shouldn't concern partners, a lack of acquisitions should. "Filling holes" is a ridiculous analogy, btw. It reminds me about that story of a Dutch kid taking his finger to a dike. But not literally.

Others ad formats are arising, like with SlideShare's AdShare for professional sharing (advisor disclosure).The question I have is if new ad formats and other ad innovation will fuel another modest boom, because ads will always be a portion of new rising internet business models that are widely adopted. What form will they take? And will social gain a creative form that is valued as much as the creativity we have despite it.

The More Than RSS Market

Richard McManus of RWW notes the continuing decline of RSS Readers, suggesting the market is largely dominated by Google and in disarray. Five years ago there was a perception that this was a hot category. An underlying standard was freeing up new atomized content and conversations that could be pulled and curated. Bloglines was acquired, and new clients were popping up weekly.

I believe an opportunity to cooperate beyond the syndication format, on standards for the basic usability of subscribing (the Coffee Cup problem) was missed at the time when it mattered. Innovation continued with the rise of Netvibes with the widget model, and Google Reader as a Bloglines 2.0. But then it stopped.

Last night I found myself looking at RWW's list of the top 10 RSS Readers, and Tweeted: Why didn't a single top RSS newsreader adapt into a Twitter client?

We all know that Twitter cannibalized RSS Reader habits with something simpler and social. And innovation happened elsewhere for aggregation with simple focused things like Techmeme. And that enterprise RSS innovation moved away from clients. But iGoogle and Netvibes widgets as Twitter clients were developed by third parties. Perhaps it was innovator's dilemma on a compressed scale, but the Readers didn't expand what could be read.

There is a ton of innovation in the Twitter client space. From Seesmic to Tweetie (I paid for it) to Brizzly (Jason Shellen created Google Reader). 107 clients on oneforty.com. There are some that bring baseline newsreading and conversations together, like Threadsy. With blogs adopting the Twitter API, the future will be reverse engineered.

Jeff Nolan and I usually spend our time disagreeing about politics and agreeing about tech, further demonstrating that smart people can be idiots. I can't disagree with him that there is no RSS Reader market today. But I can surmise that if the original players in the market stopped thinking of it as software, and kept evolving it into something closer to FriendFeed for people who want less information -- the network would be the market. And an opportunity remains.

The Politics of Attention & Microblogging Governance

Marshall Kirkpatrick surveyed the use of Twitter by the people who work at Twitter and found that they use it like regular people use Twitter. They post 2-3 times a day and follow a reasonable level of less than 500 people. However, being people who work at Twitter they have a lot of followers. The man bites dog part of the story, however, is that Twitter gets its power from the people who use and develop upon Twitter, so shouldn't they be paying inordinate attention to those people?

There is only one Robert Scoble. He's the prisoner who takes the experimental drugs to see part of the future for us, and we get the FDA approved drug years later. He follows, well, everybody (almost 100k people).  Did the same thing when RSS came onto the scene by subscribing to 10k feeds.  Yeah, sure this is in part a publicity stunt, and drives up his own follower count, but that's not his point. By emodying the power user he sees things we don't see. Not what he gets from his followers, but from coping with attention overload. Guess what Robert wants Twitter to develop next? Filters.

But guess what?  In Twitter people are the filters. When you follow them you not only get what they say, but what they share, often from people you don't follow. Its not just a reply-optional medium, but subscribe-optional, and social serendipity lets it stay that way.

Despite the recent @replies kerfuckle, social serendipity works on Twitter because of asymmetric follow. On the other end of the spectrum, Facebook is trying to break out of how serendipity happens despite symmetric follow, often in ways where privacy has to be broken to reveal new people (photo tags, comments, etc.). Social software is stuff that gets spammed, especially on Twitter. Note how trending topics is gamed by spammers and watch what will happen as new microsyntax conventions get popular. With @, RT, # and more Twitter employees discover plenty.

80% of Twitter accounts have <10 followers. I follow less than 500 people and have a lot more followers.  This lets me spend a reasonable amount of focus on some people I know and curate good things to share for many people I don't. I make a point to follow new people, but also drop others, so my filter is constantly shifting. I'd be a lot more concerned if the Twitter employee follow list was static than it being small and naturally focused.

Now I don't want to take away from Marshall's core point that Twitter employees need to listen to their community. But I don't think their base follow list is enough to say they aren't listening. At #140tc, Alex Payne noted how they use CoTweet for organizational tracking and internal assignment, for example. Similar to how Signals pulls Tweets in for organizational awareness. Listening, learning and responding as an organization takes a different approach than how a power user could and should. And you can bet Anamitra and other team members pay a lot more attention to things that trend than we do.

Stowe Boyd rightly notes the key issue is governance. While we shouldn't expect a representative democracy from the venture that is Twitter, a broad check and balance is in place.  As Stowe quips, "But it is our dancing that makes the house rock, not the planks and pipes. It is us that makes Twitter alive, and not the code."

Google is an advertising company, not a search company, so too Twitter is a media company not a communication utility company. Form follows funding, and so too governance.  Twitter the venture has lately taken a position similar to an old fashioned media company, relatively neutral and unbiased in enabling commercial relationships to form on its platform.  A good approach for long term value. But as a new media company, these opporunities arise asymmetrically from what emerges from communication. I find this a useful analogy to help explain Twitter's business, and if the neutrality principle is true, we can help keep it in check.

I wouldn't want Twitter employees to drive up their following count in response to this. At best it could be a false guesture and at worst it would distract them from getting shit done. Similarly, I wouldn't want the structure of a Signals network to map to the org chart of a company.  The politics of attention may want it to, but as a culture of use emerges it is more likely to represent flows than silos.

Twitter is Reply-Optional

Replying on Twitter is optional, and is part of what makes it scale, even socially, quite well. It is a small part of how letting go helps you cope with overload. Following a few posts on Asymmetric Follow, JP blogged on continuous partial asymmetry. Its Sunday, I've quit smoking for a month, blogged twice now, and the present isn't evenly distributed, but I'll comment.  JP plucked three things from the thread:

Three in particular are worth emphasizing:

• People in a Web 2.0 network are not uniformly connected; some have more connections than others
• Connections have directions; the number of inbound connections may far exceed the number of outbound connections, creating an asymmetric environment
• This is particularly true of “default-public” networks such as Twitter; Flickr is also likely to evince similar behavior.

I spell checked the British out of the blockquote just to fuck with him.

He quoted Stu Berwick that "In IM, it's polite to be silent." True that. Jerry Michalski taught me the power of silence online, Steve Gillmor did online. Then Steve chided me for apostrophe usage, which I still see as blurred a line as post-enlightenment serfdom. That and how my English teacher aunt derided how email destroyed English.

But, really.

What makes Asymmetic Follow scalable technically is that its easier. Scalable businessmodelly, if that it is a word, is that its good for adoption. Scalable socially in that it doesn't create non-casual attention burdens.  Twitter is reply-optional in the 'verse.

JP:

What is this thing that works? Asymmetric follow. Why? Because I am no longer expected to reply to everything that comes in. People who receive a lot of snail mail or e-mail don’t reply to everything that comes in either, so what’s the difference? The difference is in the perception of polite behaviour.

The politeness issue alone is not enough either. This whole thing is exacerbated, beautifully exacerbated, by the 140 character limit of Twitter. Because we can now have “continuous partial asymmetry”. Someone who has 4000 followers can choose to reply to the @s of 400 of the followers, because of two critical things. One, the cost of replying to the @ is low. And two, you can vary the particular 400 you’re replying to. Yes you’re constrained, ostensibly by personal bandwidth, from replying to everyone all the time. But because you manage to reply to some of the people some of the time, nobody feels left out, the weak ties remain in place and everything works.

There is another thing that works. DMs. I don't subscribe to Scoble's linkbait raint on how Direct Messages suck. DMs blow. DMs are push attention modelled messages within confirmed ties.  What that actually means is they are like having IM between two people in that both parties confirmed the relationship (tie), but the modality is still new enough that you don't have to reply soon. 

Hell, Stowe is right that Scoble would like it all to turn into email. So familliar and comfort. Ben Gross might pipe in with media displacement theory (you can't completely displace a modality of communication, it stays with us, those habits and tools last waaay to long) here, but let's try to appreciate what is new.

Maybe the tools, and burden from tools, have created actually productive practices, and tools.  Where some channels have different burdens of reply.  Twitter has a near zero burden of reply.  DMs have a slightly less burden of reply, but still a cost to the sender in the form of relationship, initial and retention.  Some of the best and busiest people say to me today that DMs are the best way to get ahold of them, when 'elst fails.

But what is interesting implementing a Twitter for the Enterprise is the practices to encourage use of Socialtext Signals as reply-optional. As a toolmaker, you cast a tool into the world, but people can use it differently. You try to set the right defaults, but don't want to assume intelligent design. There will be a manager that will demand reply in the new modality, but the conversation that makes things really work is what modality works best for what kind of attention.

Identi.ca Launches, an Open Source Twitter

Evan Prodromou, one of the better citizens of the wiki community and founder of WikiTravel, launched Identi.ca today.  Its a Twitter clone that is also distributed as Open Source licensed software.  I've been playing with it in semi-private beta.  From the FAQ:

The goal here is autonomy -- you deserve the right to manage your own on-line presence. If you don't like how Identi.ca works, you can take your data and the source code and set up your own server (or move your account to another one).

Its made Dave Winer's day and Jevon is Canadian. Marshall Kirkpatrick has a longer writeup and sees promise in federation:

Ultimately, this means federation.  I put a customized version of the foundation software (called Laconi.ca) on my server, you put one to your liking on yours, we both get friends on our local copy and any other versions around the web - and everyone can communicate with each other just like we were using the same service from the same provider. Whoever comes up with the best alternative to the garbled name Identi.ca wins!

That's something that many people have wanted to do for a long time. 

Can it work, work it scale? At least it's open source so the development community doesn't have to play armchair quarterback for a black box like they are with Twitter. Maybe these little puppies can get tied into Gnip, the social media switchboard service we wrote about yesterday.  Especially once Twitter integration happens, we expect to see Identi.ca become an important part of our work day here at RWW.

I wonder what role this could play if Twitter goes down and a community emerges to help scale this.  Or how open source contributions bend it in different directions.  Should be fun to watch.

Twitter's Business Model: A Gift Economy

Brian Alvey is joking by suggesting that Twitter's business model should be advertising supported only when its not working.  But life actually should imitate such art and I've been saying for some time that Twitter can monetize the goodwill it creates.

Specifically, Twitter should ask for donations with a Radiohead choose your own price model.  Wikipedia's fundraising increases substantially during service outages.  If Twitter is to become a global and dependable utility it will need additional resources over time.  If governed well, I believe an adequate subset of users will gift an adequate subset of monetary resources to not only sustain itself, but become a thriving business.

How much would you pay to Twitter?

P.S. Twitter! (apparently, It's Made of People)

Further, 17% of replies to this question would pay $2 per month, or $7M annual, for what its worth

Google Acquires Jaiku

Google acquired Jaiku, a Helsinki-based continuous partial presence startup.  I'm really happy for Jyri and the whole Jaiku team.

My best post about Jaiku is here.  Google just got one of the better mobility and social software design teams.  They have a far richer understanding of presence than anyone I've met.  It seemed when they got started that they would be bought by Nokia, who is trying to turn into an internet company.  Jaiku is more than a Twitter alternative.  With Google's portfolio, you could easily see it augmenting Gtalk and Orkut on the web and mobile clients. 

But perhaps the greatest direction they can go with this is lifestreaming.  Follow me on this.  Google has said they will compete with Facebook through openness.  Facebook's Social News Feed is the new Inbox, the focus of attention when it is the economy.  To compete against it, the challenge will be overcoming the semantic advantages Facebook has by keeping everything on their platform for constructing the Social News Feed.  With Google's savvy around structuring the unstructured, picture lifestreaming evolving into something that infers permalinks for social activity.  One day your Google homepage may be a stream of your friends and what they are doing, sharing, and adopting.

Then again, Anil might be right that Jaiku is the new Dodgeball.

Pownce: Collaboration from and at the hip

At first glance, Kevin Rose of Digg's new startup Pownce is Yet Another Status Message Service (YASMS) like Twitter, Jaiku or Plazes.  But really, its a collaboration app made for the most modern web.  It's bound for adoption because the founders can drive word of mouth and its inherent virality.  And perhaps what it does is less important than the three trends it represents.

Like others, the primary activity is messaging to your social network.  You message to all your friends or public like others, or directly like typing "D Username" in Twitter at the beginning of a message, but also lets you select a subset of friends.  Beyond messages, you can share links, files and events.  Beyond doing this on the web, there is a Windows or Mac rich client.

The digerati and diggerati will probably rant away about how it doesn't have SMS or IM integration like Twitter, how the content is mundane (same thing with blogging five years ago), how it needs APIs and microformats (which it does, and hooks into Twitter, del.icio.us, Flickr, Upcoming and Facebook are inevitable), or just complain about adding friends again (Adding friends is the new zen).  The design is slick on both the web and client and they will polish up key details like last names, comment threading like Jaiku, permalinks and need a more public space to explore. 

What does it matter how one comes by the truth so long as one pounces upon it and lives by it? -- Henry Miller

But here's the three trends:

YASMS Gets and Ad Format -- I admire startups that launch with an actual business model.  They have introduced a new Ad Format, a message broadcast into the stream with the Pownce icon (the green P in the above screenshot is an ad from PBwiki, I love wikis) that doesn't seem to persist.  I didn't mind the ad from LaughingSquid (first I saw) in my peripheral attention.  And if I did, I could pay to make it go away, but subscribing to the Pro version for $20/year and also be able to send files over 100MB.

AIR Gets a Viral App -- The client is built on the Adobe Integrated Runtime (AIR), formerly Apollo and still in Beta.  Pownce's virality will give AIR the airtime it needs for base of users to install it, making it easier for the next AIR client to come along.  This should be AIR's showcase.  That said, Pownce's model is what I call contained virality, where limits are part of the draw and when you are in you feel in (at least to share music, a hidden driver).

Consumer Collaboration Get Hip -- Anyone who follows the enterprise collaboration space will immediately see parallels with P2P collaboration apps like Groove or Shinkuro.  Or IM, Skype and more directly enterprise IM like MindAlign.  The key difference is group forming by social network and default modes of sharing more publicly.  Pownce will appeal to a very different demographic, that's already collaborating on blogs, wikis and IM, and potentially full a space in between.

There are several vectors in which Pownce could go, or others could go towards including presence, location, public IM, security, indexing and integration.  Pownce will have to open up invites soon (I'm out, please don't ask) to build its network effect before others encroach.  It isn't unique enough to gain the continuous or at least partial attention of users for yet another client.  Infrastructure costs will be greater than P2P.  At the risk of breaking the design and making it too complex, Pownce should give serious thought to the role of standards and how they could be a client for Facebook.

Please, I don't have any invites.

Twitteroll

What do you get when you cross a Blogroll with Twitter?  A Twitteroll.

Status Contests and Attention Aggregators

I find myself updating my status, or answering the question "what are you doing?" across Twitter, Jaiku, Plazes and Facebook.  This is made easier through clients like Twitterific, Juhu, Plazer and some Facebook hacks that are less attractive.  I'm using these clients for more than updating status easily, however.  They are a new kind of attention aggregator -- bringing the status or lifestreams of my social networks to me in real time.

They are pretty cool tools, if you haven't tried them.  For Twitter, having a client fits my laptop-oriented daily use, so I can mostly turn the mobile client off (text "off" to 40404, and "on" when you are roaming around) and subscribe to a larger social network.  For Jaiku, having a client brings continuous partial presence to my laptop, that is far richer because people are lifestreaming (adding feeds from other tools like blogs to enrich their presence).  For Plazes, the Plazer has always let me share my location, but now it is richer with status sharing and a reverse-chronological view of your network.  Facebook will surely have a better client soon as part of their quest to be the social operating system, and hopefully incorporate your Mini-feed.

But all this client development seems like one-offs, status service providers have stayed out of development perhaps wisely, and there is ample room for innovation.  As is usually the case, we've been here before -- a lack of standardization and cooperation yields less and leaves room for a large service provider to monopolize.

I don't think we want to wait for Google, Yahoo or Microsoft to provide a status and lifestream integrated user experience that flows through clients, let alone browsers or operating systems.  I do think this will happen as status services fill a valuable niche in our demand for social interaction, let alone being a new command line for web services.

Compare where we are to how RSS and Atom provided common standards for developers to innovate.  There was a time where almost every graduate CS student would write a news aggregator for fun and new startups proliferated.  Adriaan Tijsseling leveraged Atom from the earliest versions to create client blog editor, Ecto, and went beyond the Flickr Uploader to also offer aggregation with 1001.  Some, like Newsgator, plunged deep into clients across operating systems, email clients and mobile clients to let you work across them.  But the result isn't the case where one vendor, could own publishing, syndication and reading including clients.  We have a healthy marketecture that continues to innovate.

There is a new kind of aggregator, for more real time attention, that needs to be build to work across status services.  I'm not sure if it will be built into existing news aggregators, if existing status clients will evolve into them, or it will be something new.  I just know it is coming.  It will leverage status service providers and Lifestreaming you find in services like Dandelife and Jaiku.  You will be able to edit your status and perhaps more, like location to Plazes or a blog entry.  Maybe it will be built on Apollo or Google Gears, maybe a Firefox extension or a mobile version on WidSets.  But it won't happen too soon.

The problem is, while the REST APIs are easy to work with, they aren't standardizing.  Maybe they will converge on using the Atom Publishing Protocol.  Maybe they can work out a way to let you write your status once, publish everywhere, and remove dupes when aggregating.