Sexy Widget

Just Saw this on MySpace

On Disqus and the Challenge of Building Distributed Services

I recently added Disqus to Sexy Widget, and I think they have a real chance to make it.  To earn real estate on a blog, distributed service providers typically need to provide two levels of value; to the blog owner, and to the blog community.

Disqus checks those boxes.  As a blog owner, I find value in a service that will manage things like comment spam, and that over time can remove obstacles to posting – specifically, by providing a single login that will work on multiple blogs. 

From a blog community perspective, there is clearly value in a service that can preserve a commenter’s post history, login, and reputation across domains.

And it certainly doesn’t hurt that they have the Mac McCaughan factor working for them - a flagship client / investor who has big influence with the early adopter crowd.

However, Disqus does have challenges to overcome. 

First and foremost, if my experience on TypePad is any indication, it’s not a trivial install.  I had to upgrade my account from “plus” to “pro” ($8.95 per month to $14.95) to get enough control over my templates to install Disqus.  For me at least, this is not a free service.

My first install attempt didn’t work.  You have to hack a couple of different blog templates and replace bits of code, so it was conceivable that I might have screwed it up.  However, there was no way for me to tell if it was user error, a problem with Disqus, or a problem with TypePad.  Fortunately, I had made a copy of my original templates so I could go back to square one and try again.

My second install attempt didn’t work either.

While I am not a developer, I hack around enough with widget code and HTML that I feel pretty confident that if I’m unable to install a widget, there’s a good chance that others will struggle as well.

At this point, I reached out to the Disqus founders and explained the situation.  I got an immediate response from Disqus cofounder Jason Yan which was awesome, who offered to go in and check it out.  I gave Jason my login credentials.

Unfortunately, Jason couldn’t solve it either.  For whatever the reason, my TypePad templates weren’t updating properly, so Jason opened up a support ticket.

A couple of days later, TypePad got back to us saying:

"We made a configuration change to your account that will allow you to
see the changes immediately. You may need to republish your templates
in order to see the changes.

We are in the process of rolling out some new features to users,
including pagination and comment feeds. However, the features are not
fully supported for Advanced Templates and were affecting the display
of your blog. We are working on releasing these features to all users
very soon, and we appreciate your patience while we work to improve
TypePad."

So here we have a widget install that will cost me $6 per month, took several days, and required me to correspond with the cofounder of the service to make it work.  Despite the fact that none of this was the fault of Disqus, it doesn’t change the fact that this was not a painless install process.

By definition, distributed services are dependent on host environments to work properly.  When the Facebook platform acts up, your app looks bad.  When TypePad templates don’t work properly, it could cripple your adoption. 

In 1999, pursuing a distributed business model meant becoming an “ASP,” which required BizDev, sales cycles, and implementation conference calls.   APIs and embeddable widgets have brightened the picture considerably – but as the case of Disqus illustrates, there are still challenges for those who are entirely dependent on host sites. 

Post To

Post To is the feature that everyone developing for MySpace is waiting for.  From the MySpace Developer Blog:

PostTo - the ability to post to different MySpace areas from inside an application; messages, comments, profiles, bulletins, etc. This is a huge feature and defined as an extension.

It's not live yet, but it's coming.

Mixwit Mix Tape Widget

I saw Mixwit mentioned on A VC and took it for a whirl.  It's a neat little mix tape widget that allows you to search for songs on the Web, and include them in an embeddable mix tape.  Here's mine:

Widget experts will note Mixwit's use of Gigya as a distribution tool, as well as the fact that publishing to Facebook prompts you to add their app. Good stuff.

You can rate Mixwit here:

Music Widgets on RateItAll

Snap Summit - Dave Morin Keynote

Dave Morin - Facebook Platform Head

Within the last month, a woman built an app to 300K active users and sold it.

Over 67M active users.

More than half of users outside of US.

13% in UK.  20% of entire population of Canada.

100 apps added per day.  Licensing FBML and platform to other sites - like Bebo.

Launched FBFund - grants to get going on apps.

Next year: how to enable commerce.

Focus: creating long term value.
- Frictionless Platform
Leverage Social Graph
World Class applications.

Frictionless Platform

How can FB provide value to users.  Maintain trust. 
Valuable Information, what activity is meaningful.

Facebook Profile Upgrade is coming. 
New things coming to profile: focusing on Wall, and actions you are taking.
When you come to the new profile, you get a feel for the current version of a person.

Adding tabs - available to devs.  Users can choose to add your app in a tab.

Apps can still add profile boxes on left (narrow module only!)

Facebook believes that only one social graph exists in the world - their goal is to and map it. 
Strengthening relationships, building empathy.

Facebook likes apps that drive users to interact with each other.

4B photos in the system, but if I look at the home page, i just see my friends'.

What are world class applications:

Not many apps are taking on sports or medicine.

The next phase is apps that help people work better.

Religion is underutilized in terms of apps also.

More commerce functionality is coming - allowing credit card payments.

Maximize the user experience
- respect the ecosystem
- blocking types of behaviors (force invites)
- ensure long term value
- every time they see a news feed, they need to know how it relates to their friends.

What's next?

- share tech and tools
- provide core infrastructure and resources
- enable viral distribution tools

FBFund

Backed by Founder's FUnd and Accel, $10M fund.

Infrastructure partners; Joyent, AMazon Web Services, MS Popfly

User Driven Internationalization
- volunteer translations
- FBML tags that can help you get your apps translated by FB users

Future of Platform:

Over 30K apps on FB.

Question: we have a viral app that brings in a lot of new FB users.  We don't have metrics.
Answer: Ok, more data needs to be made available.

Question: Preferred Application Developer Program?
Answer: we've thought about, but it's no something we're working on necessarily. 
We're doing roundtables, no formal program.

Question: Will gifs app credit card database be available to developers?
Answer: still working, and we don't know.

Question: FBFund, how many grants given out?
Answer: don't have specifics, but we have given out grants.  $25-$250K is the range of grants.

Question: Things are so unpredictable.  Allocations change so ofen.
Answer: We're trying to engage the community to balance user needs with dev business needs. 
Roundtables, develper blog, etc.

Question: CBS had 100 invites... are we going to see more of this?
Answer: Yes, we've heard the feedback.  We have a history with CBS.  Can't say it won't happen again in the future.

Question: Any plan to eliminate duplicate invites?  Why can someone who has an app receive another invite.
Answer: thanks for the feedback.

Quesion: Status of instant messaging and timeline of interoperability?
Answer: It should be available next week.  Open as possible.  Evaluating Jabber.  Scale problems wih chat.

Question: Is OpenSocial in your plans?
Answer: We haven't chosen to participate.  We have 300K devs on our platform.  It's one of those things. 
We're not ruling it out.

Snap Summit - BJ Fogg

BJ Fogg (entrepreneur, teacher of Facebook class at Stanford)

Psychology of Facebook - psychology of persuasion within social networks.

FB is changing people's behaviors - may be the most persuasive technology of all time.

Was iLike the fastest growing company in history (on the back of Facebook?)

Combines interpersonal influence + mass media.

six components:

persuasive experience (install, use, invite, comment, customize, join) (leverage social influence, copy what works)
(not so obvious: understanding psych, simplest touch works best over time)

Automated Structure (familiar environment, how the system works) (e.g. it's easy to create a group on FB, not so easy in real life.
(don't leave persuasion to chance, design for lazy people) (not so obvious: behavior change happens) (motivation, ability, trigger)

Social Distribution in a controlled environment - what's different about FB is that it's a high trust environmen. 
Implications: apps must have a social angle.  Not so obvious implications: don't rely on hey, this is cool.  Establish your credibility.

Rapid Cycle.  Implication: design for a fast cycle.  Not so obvious: show your momentum.  Momentum growth.

Huge Social Graph.  You need big, and you need adoptive.

Measured Growth.  You can see what's growing.  Instrument everything you can.   Never argue, just test.  Demand better metrics tools.

Paper: Mass Interpersonal Persuasion: bjfogg@stanford.edu

Snap Summit - Benedetto Keynote

Jim Benedetto Keynote - MySpace

**Clarification** I just got clarification from Jim that it's not a cap of one invite per user - it's a one invite per click.  So each user can invite as many people as they want, but they just have to go one by one.  That's a whole lot better than a cap of one total invite, no? **

The bombshell here is that, for now, users can only message / invite one other user.  Crowd is starting to get surly during Q&A.  People are pissed that MySpace is telling us, the developers, that they are assuming US to be spammy.

MySpace has a history of apps.  Photobucket, YouTube, Slide, RockYou.

These existed because of MS values of minimum creative restrictions, encouraging self expression.

Beneficial to users, devs, and MySpace as a whole.

Developer APIs:

OpenSocial APIs

Rest APIs - enables server to server communication. Outh authentication

AcionScript APIs - Flash Support

Why OpenSocial?

Commitment to open standards

Portability

Leverages existing web technologies

Open SOcial Support

Support for 0.6, support for 0.7 this week.

MySpace specific extensions:
- Bulletins
- Attributes for bands

Viral Surfaces:

Profile Surface
Canvas Surface (devs keep 100% of revenue)
User Home Page (can be customized one to one!)
Application Gallery
Application Profile

Safety and Privacy spiel

Balancing Virality and User Experience

- Long term approach to growth and distribution
- Ensure a clean user experience (lawrence: didn't know this was a priority for MS)

Artificial / Spammy Growth is not necessary

Measured Approach to application growth -
initially applications will be able to initialte the workflow for sending a message on a one to one basis.
You can send one message to one friend.  (lawrence: gulp)

Start throttled, slowly increase communication channels (lawrence: opposite philosophy of Facebook). 
WIll increase comm channels IF NECESSARY over he coming months.
(lawrence: GULP)

Monetization:

MySpace using "Hyper Targeting" to deliver more targeted ads to users.

Hypertargeting sounds like a proprietary ad network to be offered to developers in the future.

Breaking: MySpace to allow only one invite / message per user

*False Alarm.  Jim clarified his, and it's one invite per click, not one invite per user... it just means there will be no batch invites - each MySpace user will have to click individually on each friend that they want to invite / message.  He estimates that this functionality will be turned on in a few weeks.

Jim Benedetto, MySpace VP of Technology, just said that they will only allow one invite per user sent initially, but they may open up more communication channels in 2-3 months.

This is devastating news for folks developing for MySpace.  If you look at the viral coefficient, much depends on getting multiple invites sent from each user, with the knowledge that only some will convert.

Profile module is looking much more important now.

Snap Summit: RockYou Metrics Presentation

Eric Yieh, Director of Product Dvlpt
RockYou Metrics

RockYou approach to product design:

Apply advertising principles to user facing web design.  Think of everything as a page view.
Grow by increasing the number of touch points and maximizing conversion on each touch point.

Build fast and launch asap
- Super Wall coded in a weekend

Iterate over and over with A/B

Data guides product decisions, no user experts. 

Tenets of Social App dvelopment:
- Channels are better than features
- Channels bring in new users
- Don't let concept drive viral channels
- Have a plan to maximize use of every viral channel
- Tune the viral loop - user adds, engages, invites, new user adds...
- Build new features to keep them engaged and happy

How do you launch a winning app:
- Half art, half science
- Intersection of psychology and web analytics
Develop concepts around high converting calls to action
- SImple
- Universal
- Social Persuasion (reciprocity)
- Novelty (get it out first)

Potential for sustainable long term engagement
Nail down viral loop.
Timing with launch is key - launching with maximum virality (as opposed to features) is key.

Metrics for Viral Loop

Install Flow, Engagement Flow (request), Engagement Flow (notifications)

Number of requests sent and CTR
Notifications sent and CTR
Feed Events and Feed CTR
Adds / Rmoves
And everything else

How do we track?  Started with Google, but it didn't meet their needs.

When you're more serious, you need to build your own. 

A/B Test Flows AND invividual calls to action.

Making the most of viral channels:

Graph and compare activity metrics over time
Focus resources on tuning the largest traffic drivers
All viral channels eventually decay.
Keep tuning and stay ahead of the curve

Other Channels to drive installs:

Paid
Link Exchange
Cross promotion of apps

Snap Summit: User Engagement Panel

Please forgive typos, I'm blasting this out as it comes in.

Moderator: Dave McClure

Panelists: Slide BizDev, Bunchball CEO, Bebo Platform Head, and one more (I didn't catch the company name)....

What is user engagement?  People coming back to use your service.   Example?  Scrabulous.

Bebo - engagment is about capturing the attention of their users.

CEO Bunchball - How invested is a user in using your application.  Example: The Sixty One - music discovery site.

Jared, Slide - Engagement has to do with getting good content to the user, about communication.  Interaction between user and network. 
Flixster is an example of a deep, engaged vertical.

Measuring User Engagment.... what are the specific metrics, any tools?

Slide: Daily active users is important, but it's more than that. 
Session time, what they're doing, depth of experience, combination of what they're doing and how often and how long they are doing it for.

Is there a specific number?

PVs per visit, and they track uninstalls carefully.

Bunchball: track changes to applications.  Persistence is a big thing for increasing engagement. 
How often are people doing the things that are investing themselves in the app.

Bebo: Looking to reward apps with high engagement / high quality.  Take emphasis off of viral tricks, and put it on engagement. 
Look at uninstall rates, and block rates for apps.  Benign reasons for blocking?  More granular in terms of user feedback is better.

Same metrics as Facebook?

Bebo is not reporting daily active users.  Trying to come up with the right metrics to expose engagement.

Three main metrics: Daily Actives, Bounce Rates, PVs per visit.

How long does it take for someone to get tired of the app.  And why do they get tired?  (internal tools, internal metrics)

Monetization: advertising and CPA. 

What type of apps and features are most successful for user engagement?

Bunchball: user centered design is key.  Model seems to be throwing stuff out there, seeing what sticks, and abandoning the others. 

to Bebo: what features are working?  What do you consider good behavior for an app?

Apps that understand the Bebo audience do better.  What resonates with one might not resonate with another. 
Self expression is huge on Bebo - while FB is more about utility.
Apps that are more about ME do well on Bebo.  Example: a gift app that you give the gift, but you can also tell people what YOU want on your profile. 
Flipping gifts on its head to focus on self expression.
Competition also works.  Virtual currency is interesting.

Great user engagement features?  Any interaction between users trigger a notification for reciprocity.  Someone pokes you? 
Poke them back.  They also push meeting strangers - you can't do that on FB, you can only do it through the application. 
Creating scarcity with your applications, capturing new data not available on FB.

Slide: thoughts on user engagement.  Gaming engages users immediately.  But the challenge is keeping them coming back. 
Slide doesn't focus on the shorter term stuff.  Scrabulous is an exception - it's a communication form to tell somebody you're smarter then them. 
Slide tries to expose content - best case, when a friend sends another some content.

Dave: Bebo does not expose app activity to users who don't have the app.

Dave: why is this?  Isn't engagement the sign of a good app?  Feeds are not a good means for app discovery now. 
Therefore devs are forced to focus on viral tricks, notification spam.

Bebo's Response: It's a fair argument.  Feeds could be about discovery.  But, we've given plenty of channels for discovery.
(dave: but those channels are spammy).  Jessica: but activity feeds could be spammed too. 
But, we are thinking about greater feed reach as a tool to reward highly engaged (well behaved?) apps.

(lawrence's note: this is the interesting stuff, hope this is coming through)

Bebo: you shouldn't dread checking notifications.  If a channel burns out, it doesn't help anybody.  We're taking our time and doing it carefully.

Dave: this could be a real separator in the competition betwee OpenSOcial and FB Platform.  Whoever opens the feed will attract more developer.

Slide: I'm not conviced that users respond to invites to add an application.  As a user, you don't know the value. 
But you do understand your relationship with your friend.  A feed item demonstration could be more effective than just a bare invitation.
The problem is: when a channel is exposed, app developers are going to go after that. 

Dave: are you guys creating your own feeds where you control the rules?  Gaming feeds, Top Friends Feed...

Slide: yes.

To Bunchball: Feeds helpful or not helpful?

It's a hard problem.  How do you make apps discoverable without being spammy.  People will optimize for whatever is measured.

Dave: How do you make decisions about distribution, versus engagement, versus monetization.

Slide: here's the order: Growth, Depth, Money.  First, how do you get them to add it.  Then, how do you get people to go deeper, and come back.

Bunchball: totally focused on engagement.  Growth is organic - we've never focused on it.

To Bebo: is there any thought to enabling inter app currency?

Jessica: there's thoughts about everything.

Dave: what does AOL think?  (lawrence: ouch)

Jessica: I have no idea what they think, or what they have been thinking for the last 10 years (lawrence: ouch again)

Jessica: that was a joke.

Jessica: I can't comment on whether we're doing cross app currency, but it's interesting.

Virtual currency is based on fundamental human behavior.  Points drive behavior.  People are predictably irrational.

Dave One Tip from each panelist:

Slide: Algoritmic way to recommend things.  Three Super Poke options are chosen by algorithm.

Bunchball: Give people missions.  More trophies, more challenges, more things to finish.

Dave: people are simple, give them stuff to click on.

Bebo: Keep things within two clicks.  Make it easy.

Last Panelist: single button request form - one button to send an invite to a user. 
Big screen with lots of little checkboxes is not the only way to send invites / notifications.