Bubba Murarka

Updated 12/17: Corrected some typos. Also check out follow up post here.  

I attended the Finals presentation tonight as did ~500 of my closest friends. Seriously, it was amazing scene with gobs of people - mad props to everyone who helped pull the night off. Below are slightly cleaned up  notes from the presentations. I was hoping to live blog, but couldn’t get onto the wi-fi network. Not a bad thing as the info came fast and furious since each team had only 2.5 minutes to present…and they were strict about that. Luckily the presentations there was a bunch of mingling time to talk with the students and other attendees (vcs, entrepreneurs, and big company folk from google, msft, yahoo and of course facebook).

In the interest of sleep, I’ll post some takeways, links to the apps and links to some of the interesting research from PhD student Mike Weiksner on Friday. If you read the below and have questions please leave me a comment and I’ll try to clarify my frantic notes (or issue a mea culpa for miss understanding stuff).  Note, apologize in advance for poor English, formating, and general lossyness of the notes - comments fixing any of these issues would be much appreciated!

Introduction

BJ Fogg -> Persuasion since 1993 @ Stanford
“Facebook is #1 persuasive technology in 2007, ever”
Mass Interpersonal Persuasion
Global/ scalable, everyday people - interacting as friends, subtle dynamics with big powerful effects
Class site: http://captology.stanford.edu

Dave McClure
Metrics philosophy - AARRR!
-acquisition
-activation
-retention
-referral
-revenue

Use a Conversion Dashboard
Design apps by hypothesizing about customer life cycle through AARRR

Class App Stats
Total Installs Across all apps: 16M!
Daily Active Users: 925K!
Top 10 apps ~ $10M according to Adonomics
Conservative valuation $1M (discount for facebook hype)

Overall Class Metrics
Users who came back and session time (deep engagement)

First 5 teams -> engagement wins

PHOTOGRAPH
Photo Viewer
4.1B Photos on Facebook, 40% of click thru on FB
Facebook’s Photo App UI not so good
Wanted to bring collaborative filtering to photo browsing
Users interaction with app, but not viral
New hook is relevant invites (see app)

PICKMEUP
Lay a pick up line to garner a reaction
App is getting strong engagement, but no virility
Tried share your answers with friends -> good boost
Tried UI tweak to tell users to invite but don’t enforce -> big boost
Learnings: Include personal touch & tell how many invites to send

FUNNY IMAGES
Rate Pictures -> had high engagement, but no virility
Changed to send pictures to friends
Learnings: Forcing users to install worked and gave them a huge boast in adoption over last 7 days

GOOD EATS
Social Restaurants Recommendations
Focus on engagement, not viral (7.5 minutes of visit time, 6 ppv)
Deep plug into social graph (add more things to do, stickiness, new content)
Implemented a points system with different icons & leaderboard

LOVE CHILD
Competition to raise best virtual child
Aspirations:
Break into aysnc casual gaming
Custom branded digital goods
Mobile platform integration

Next 5 teams -> changing direction for success (persistence!)

DECLARE WAR to SHARE THE LOVE
Decided to focus on Love or War -> core idea was a gaming platform
Declare war -> simplified, invites only best friends (short attention span / time span, not many opportunities to generates news feeds)
Share the Love -> launched with little more than an invite box, 100K in 5 days, 750K to date
Learnings: keep it simple, appeal to emotion, use one action)

FUTURE MAKER to BLESS YOU
Future maker was like MASH -> too complicated
Bless you simple, with 1 action -> 15-25 visits a user over 30 days, 320K total users
Learnings: Simple, attractive, but not distracting

MATCHMAKER to SQUEEZE ME
Users can create matches between friends and get votes -> failed because feedback loop was broken
Tried to simplify -> 40K installs in first week
Learnings: App users receive and respond to invites in the same day

DODGEBALL to DODGEBALL
It’s nice to poke, but you want to throw blunt objects
Original app -> First class app to 10K users, two step process: had to pick ball than pick people on next screen
Revised app -> after get passed by other apps, dumbed down app because nobody reads rules, cleaner interface with all steps on one pages (long page design), page views from 2.1 to 4.5 (or something like that)
Learning: Tell users what to do increased invites from 4 to 11

WALL OF SHAME to YO MOMMA
Wall of Shame -> Socially linked hatred of objects / things, had pure invite screen without any actions
Yo Mamma is an app wrapped around invite process -> used single long page design
broad distribution in English countries
Learning: be positive, loving, flirtatious

Next 4 teams -> ???

POLLS
Trying to be useful
Found out people are lazy
People polls worked out
(dang, I got distracted by a text and missed a bunch of stuff)

COMPLIMENT BOX
Everyone likes a compliment
High motivation to engage
Learning: enforce virility, sex sells, less is more

SUPER STATUS
Enhanced status to include html, images, and comments on status -> Sharing small bits of information (microblogging)
“is” sucks, facebook should change this, why can’t I view status in better / richer UIs?

SCRIBBLED PHOTOS
Do mashups on top of Facebook photos to amass huge user generated data store
Got very creative submissions
high user engagement (4 mins)
Learning: Short attention span

—————————————————————————————-

Break with Johnny Hwin from Love Child team, AKA the next host of the Tonight Show

—————————————————————————————-

Next 5 teams -> Apps with a purpose (social good)

COLLEGES (w/ Kaplan)
Let’s make something useful -> college admittances feedback
Challenge: get students to input sat, gpa, and interested school
Partnered with Kaplan to get credibility and distribution in exchange for giving them a presence on facebook
Deep engagement (10 mins +)

SOCIAL BUZZ
Explore social context / expertise
Try to build connections through classification
More complex than other apps in class
Learning: optimize first user experience, use profile data, must provide content for user immediately

COMMONALITIES
Find out what you have in common with others -> mine user profiles to find matches
5 APV, 4 mins a user, 8.4 visits per user

GIVING TREE (w/ Kiva)
Microlending -> Really complex idea and tried to explain it to facebook
Learning: Second guess complex ideas

ONE VOICE
Get tv / videos into facebook, users become VJs
Make playlists on youtube and add it to facebook app
Learned: interaction between users, simple as possible, iterate as much as possible, find niche

SAVE THE RAIN FORREST
Test socially conscious app
Word game that takes ad revenue and donates it to save rainforest
8:28 minutes of usage a day
Learning: Users have sense of ownership for profile

BJ Fogg: What makes Facebook Tick - Survey of 80 Students

Convenient & respectable way to:

feel connected to existing friends
stay updates on existing friends
get info about new people
start $ build relationship
express your identity and views

Mike Weiksner: 6 patterns of success for Facebook apps
Analysis of top 100 + classroom learnings

Native patterns:
provoke & retaliate - > 23 of 100
self expression -> 26 of 100
revel & compare -> 26 of 100
group exchange -> 14 of 100

Adapted Patterns:
competition -> 9 of 100
deception

Most apps focus on 1 pattern but incorporate other 5 to some level (neat hex diagram)

Next 4 Apps -> Light weight gaming

GUESS WHO
Like board game
Crazy number of page views, visits per month, time per day
Learned 3 things that mattered:
encourage reciprocity
create shared intimacy
act fast on surprising data

CAR OVERLOAD
Pageviews and users don’t mean nothing
biggest facebook page views are newsfeed, inbox, profile page
vertically segmented facebook applications

COCKTAIL LOUNGE
Apps they spent time on didn’t work, quick to build apps took off
things they did wrong:
don’t make things too complex
not enough focus on reciprocity
using .net with the platform didn’t work well -> re-wrote everything in PHP

TOURNAMENTS
Tournaments are good human behavior hook
Missed light weight interaction
Future:
cater to niches
match madness
tournament style polls

———————————————————————————————————

Amazing Fact: One of the sponsors for the night was the entire class itself (using ad rev from their apps)!

———————————————————————————————————

Last 4 Apps -> fast growing

SEND HOTNESS
5M users in 5 weeks
viral factor = X - conversion rate (percentage of invited user who install the app) * Y - engagement rate (install app + invite 1 friend) * Z - invitation rate (average number of invites sent per engaged user)
Maximizing Z -> create incentives for invites
Learning: Iteration is key, no real secret sauce

KISSME
Kiss: loving /provocative got lots of reactions
First app to cross 1M installs
100K active users a day with World wide distribution
Next: See if viral app can be converted to something useful?
looking for funding

PERFECT MATCH
easter egg: perfect_video
there are no two people who would both want to be with someone else (cs wedding problem)
1.  announce the best results you can on first visit
2. collect info on visits
3. calculate& then announce
Learning: Catchy mini-feed message with picture

HUGS
Built by TAs
3.2 bazillion dollar market
2M users, 15M Hugs sent
every country except north korea
Learning: giving people to connect in easy way…that is the big

Summary from Teaching Team (Yee Lee & Jia Shin):

1. go viral
2. then go deep

Metrics
K-factor: 1.4 - 2.1 is good
User engagement: 5mins a session, double digits visits per month per user are good
11 hours to 2 days of viral loop

Advice
Speed, Speed, Speed
Dev time should be 2-7 days, if not viral fix that first
Spend most time on rig (instrumentation & analytics) -> Cyclotron model for apps

BJ Fogg’s Summary

This stuff is learnable in 10 weeks…
Thanks to all of the sponsors!