Viral coefficient – Unveiling the

Holy Grail of online marketing

Joni Salminen & Aarni Hytönen

17.10.2012 1

Contents

1. What?

2. Why?

3. Results?

4. Implications?

5. Limitations?

17.10.2012 2

What?

• Modelling growth of viral marketing– diffusion of marketing messages in a social network

– The model can be used to determine the viral growth of visitors

in a website at a given time.

17.10.2012 3

”Viral marketing” is nothing new – it is like word of mouth or peer marketing, only the Internet eliminatesphysical distance and increases the number of potential connections inside the network.

Example

“Like, with ONElist, the grand total of all the advertising I ever did

for that company was I spammed some guy who had posted to

Usenet looking for a mailing list provider. And he was in Norway;

this was on a Saturday evening in January of ‘98, and I just said,

‘Hey, try my service.’ The next day, I wake up, and not only had

he created a list, ten of his friends had created lists. We had

hundreds of users, just within the span of a few hours and one

email. After 11 months we had a million users. Just from that.”

17.10.2012 4

Why?

• Limited base model used by practitioners

how to improve?

• Through better models better applications,

better theory.

17.10.2012 5

Base model (Tokuda 2008)

17.10.2012 6

If x * y > 1, then viralgrowth!

Limitations of the base model

• Lack of carrying capacity

• Circularity (static loop)

• Lack of time (saturation)

17.10.2012 7

Viral paths

17.10.2012 8

B

C

A

D

Results?

17.10.2012 9

where

a = acceptance rate (%)

b = average invites per person

c = initial target group

d = carrying capacity

t = time

Solutions to the base model

• Lack of carrying capacity added

• Circularity (static loop) coefficient

• Lack of time (saturation) added

• (Separating initial target group.)

17.10.2012 10

Practical implications?

a. increase number of patient zeros (initial group) through

advertising/promotion

b. convert key influencers to increase acceptance (influencers

as hubs, more connections and better quality)

c. run experiments on sub-sets of carrying capacity to find

proper marketing variables.

Combine a+b+c

17.10.2012 11

Fast wins or sustainability?

Consider also the cumulative growth of visitors:

if expiration is rapid, a long term strategy with a

low viral coefficient may bring better results than

short-term campaigns (topicality).

17.10.2012 12

Limitations?

• Focusing on visits (not customer relationship)

• Lack of empirical validation ( testing with data)

• The dynamic nature of d (boundaries of the

carrying capacity)

• Aggregates instead of algorithms

17.10.2012 13

Thanks!

17.10.2012 14