This is a presentation about the different kind of technology consumer products, where viral loops can fit in, and how to design an efficient viral loop.
2. Structure of Talk
• Value Loop
• Network Effects
• Viral Loops
• Designing viral loops
3. Types of products
• Inherent value or dependent on network effects?
• Primary value => Inherent
– even a single user can be a long-term user
• Primary Value => network effects
– No community => all users leave
4. Value loop
• More value + less cost = value loop
• Usually Word of Mouth
• Typically takes a while
• Concentrate on channels of distribution
• E.g. - mint.com, google.com, early days of facebook
5. How do we measure value?
• Net Promoter Score
– Users rate from 1-10
– 10 = best
– Promoters (9-10)
– Passives (7-8)
– Detractors (1-6)
• NPS = % Promoters - % Detractors
• High NPS (>40%) => No need for virality
8. Network Effects
• 100 users or 10,000 ?
• It depends!!
• Imagine 10,000 facebook users with 0 friends each
9. Network Effects
• 100 users or 10,000 ?
• It depends!!
• Imagine 10,000 facebook users with 0 friends each
10. Network Effects
• Metcalfe’s law:
Value of network = c(P^2)
P – no of connected users
• If n sets of connected users,
Value = c(P1^2 + P2^2 + … + Pn^2)
11. Network Effects
• Metcalfe’s law:
Value of network = c(P^2)
P – no of connected users
• If n sets of connected users,
Value = c(P1^2 + P2^2 + … + Pn^2)
• E.g. facebook, linkedIn, twitter
12. Network Effects
• Metcalfe’s law:
Value of network = c(P^2)
P – no of connected users
• If n sets of connected users,
Value = c(P1^2 + P2^2 + … + Pn^2)
• E.g. facebook, linkedIn, twitter
• Cold start problem
18. Psychology of sharing
• Facebook -> Socialite
• Twitter -> In the know (journalist)
• Yelp -> Critic
• Quora -> Expert
• Instagram -> Photographer
• Foursquare -> Person-about-town
• Your product -> ?????
19. Friendship graphs
• Symmetric or asymmetric?
• Friend or follower?
• Friend = more private
• Follower = public, but higher growth
– More specific to intent-based vertical networks
20. Building Network Effects
• Critical Mass
– community that sustains itself
– 150 users in each community
• Alpha Users
– High Social Networking Potential (SNP)
• Cold Start problem
– If I have only a few friends, where is the value for me?
22. Building Network Effects
• Social Learning
– Watch what others are doing
• Singling Out
– Messaging from others
23. Building Network Effects
• Social Learning
– Watch what others are doing
• Singling Out
– Messaging from others
• Feedback
– How does my network respond to my actions?
24. Building Network Effects
• Social Learning
– Watch what others are doing
• Singling Out
– Messaging from others
• Feedback
– How does my network respond to my actions?
• Distribution
– Extent of reach of my actions. Who all sees it?
25. Network Effect example
• Credit for user @a all along the chain
• More motivation to write funny/interesting tweets
26. Building Network Effects
• Network Effects
– Alpha users will help, but only for a while
– Slow growth => death
– Need viral growth
27. What is viral?
• Anything that grows exponentially, like a virus
28. Why do we want viral growth?
• It’s cheap (and fast)
29. What is Viral?
• Anything that grows exponentially, like a virus
• Word of Mouth propagation can be viral under
certain circumstances
30. What is Viral?
• Anything that grows exponentially, like a virus
• Word of Mouth propagation can be viral under
certain circumstances
– But typically slow
• How do we know when something is viral?
31. Viral Co-efficient
• Let’s say each user invites 5 friends
– on avg 22% of invitees register
– viral co-eff => 22% * 5 = 1.1
• > 1 = viral
32. Viral Co-efficient
• Let’s say each user invites 5 friends
– on avg 22% of invitees register
– viral co-eff => 22% * 5 = 1.1
• > 1 = viral
• If 1 user gets 1 more, can get the entire world
33. Viral Cycle Time
• If 1 user gets 1 more, can get the entire world
• But, how fast?
• Until market is exhausted, how fast does it grow?
– Viral Cycle Time
– Depends on each product
• Measurement is important!!
34. Measuring Viral Cycle Time
• Cohort Analysis
• Weekly (or daily) cohorts of users
• Track how users behave over period of time, based
on when they joined
• Idea: Don’t make decisions for new users based on
metrics from more experienced ones
35. What parts are viral?
• Invitations
• Engagement
• Trick is to invite other users as part of the core user
experience
• Not to be confused with network effects
36. What is the viral loop?
• User A takes an action
37. What is the viral loop?
• User A takes an action
• User B is notified as part of that action
38. What is the viral loop?
• User A takes an action
• User B is notified as part of that action
• As part of the response, user B takes SIMILAR action
39. What is the viral loop?
• User A takes an action
• User B is notified as part of that action
• As part of the response, user B takes SIMILAR action
• User B therefore invites User C
40. What is the viral loop?
• User A takes an action
• User B is notified as part of that action
• As part of the response, user B takes SIMILAR action
• User B therefore invites User C
• … and so on
41. What is the viral loop?
• User A takes an action
• User B is notified as part of that action
• As part of the response, user B takes SIMILAR action
• User B therefore invites User C
• … and so on
• Integral part of the product, not a layer
43. Designing viral loops
• Single core activity
• Single entity for user to interact with
• Can have multiple types of users, but with great
difficulty
• Channels for virality
44. Single activity
• What is the core activity?
• Can we add messaging to this activity?
• The new user who sees this responds to this activity
by taking the SAME action
• … and so on
45. Single activity
• What is the core activity?
• Can we add messaging to this activity?
• The new user who sees this responds to this activity
by taking the SAME action
• … and so on
• Viral growth
46. Single type of user
• Multiple sets of users – VERY HARD
• Motivations are very different for different groups
• E.g. Seller != Buyer
47. Single type of user
• Multiple sets of users – VERY HARD
• Motivations are very different for different groups
• E.g. Seller != Buyer
• Design loop for each set separately