Yammer Groups and Business Value - Does size matter?
-
Upload
pekadad -
Category
Data & Analytics
-
view
501 -
download
0
description
Transcript of Yammer Groups and Business Value - Does size matter?
Yammer Groups and Business
Value
Lee Romero
October 2014
Does size matter
Preface 3
Background 4
Hypothesis 6
Data source and methodology 7
Findings
bull Activity levels 11
bull Responsiveness 16
bull Knowledge impact 18
bull Private vs Public Groups 19
Summary 21
Contents
2
This material has been assembled as part of an informal review of communities at
Deloitte
It presents some insights relating the size of a community to its value
It does not represent an official Deloitte position
Preface
Disclaimer
3
As part of Deloittersquos Community program we have long used a health report to
help drive desired changes
That health report considers a number of factors ndash the goal is to identify specific
factors which if we can effect change will improve the value of our communities
One specific factor used is the size of the community
The rationale has been that in light of the ldquo90-9-1rdquo principle there is a certain
minimum size of a community need to reach a critical mass of activity
Background
Deloittersquos Community program
4
See httpenwikipediaorgwiki125_rule_(Internet_culture)
As part of our Community program Deloitte has adopted the use of Yammer as a
primary tool of discussion within our communities
Yammer provides a completely user-driven means to create ldquogroupsrdquo ndash which
roughly speaking can map to communities
Following the same logic as our communities through some moderation of new
groups we have pushed for fewer larger Yammer groups to increase their value
Recently the importance of size of a community Yammer group has been
challenged with the assertion that people are more comfortable and therefore
more likely to communicate in a smaller group of people
Background
Deloittersquos use of Yammer
5
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Preface 3
Background 4
Hypothesis 6
Data source and methodology 7
Findings
bull Activity levels 11
bull Responsiveness 16
bull Knowledge impact 18
bull Private vs Public Groups 19
Summary 21
Contents
2
This material has been assembled as part of an informal review of communities at
Deloitte
It presents some insights relating the size of a community to its value
It does not represent an official Deloitte position
Preface
Disclaimer
3
As part of Deloittersquos Community program we have long used a health report to
help drive desired changes
That health report considers a number of factors ndash the goal is to identify specific
factors which if we can effect change will improve the value of our communities
One specific factor used is the size of the community
The rationale has been that in light of the ldquo90-9-1rdquo principle there is a certain
minimum size of a community need to reach a critical mass of activity
Background
Deloittersquos Community program
4
See httpenwikipediaorgwiki125_rule_(Internet_culture)
As part of our Community program Deloitte has adopted the use of Yammer as a
primary tool of discussion within our communities
Yammer provides a completely user-driven means to create ldquogroupsrdquo ndash which
roughly speaking can map to communities
Following the same logic as our communities through some moderation of new
groups we have pushed for fewer larger Yammer groups to increase their value
Recently the importance of size of a community Yammer group has been
challenged with the assertion that people are more comfortable and therefore
more likely to communicate in a smaller group of people
Background
Deloittersquos use of Yammer
5
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
This material has been assembled as part of an informal review of communities at
Deloitte
It presents some insights relating the size of a community to its value
It does not represent an official Deloitte position
Preface
Disclaimer
3
As part of Deloittersquos Community program we have long used a health report to
help drive desired changes
That health report considers a number of factors ndash the goal is to identify specific
factors which if we can effect change will improve the value of our communities
One specific factor used is the size of the community
The rationale has been that in light of the ldquo90-9-1rdquo principle there is a certain
minimum size of a community need to reach a critical mass of activity
Background
Deloittersquos Community program
4
See httpenwikipediaorgwiki125_rule_(Internet_culture)
As part of our Community program Deloitte has adopted the use of Yammer as a
primary tool of discussion within our communities
Yammer provides a completely user-driven means to create ldquogroupsrdquo ndash which
roughly speaking can map to communities
Following the same logic as our communities through some moderation of new
groups we have pushed for fewer larger Yammer groups to increase their value
Recently the importance of size of a community Yammer group has been
challenged with the assertion that people are more comfortable and therefore
more likely to communicate in a smaller group of people
Background
Deloittersquos use of Yammer
5
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
As part of Deloittersquos Community program we have long used a health report to
help drive desired changes
That health report considers a number of factors ndash the goal is to identify specific
factors which if we can effect change will improve the value of our communities
One specific factor used is the size of the community
The rationale has been that in light of the ldquo90-9-1rdquo principle there is a certain
minimum size of a community need to reach a critical mass of activity
Background
Deloittersquos Community program
4
See httpenwikipediaorgwiki125_rule_(Internet_culture)
As part of our Community program Deloitte has adopted the use of Yammer as a
primary tool of discussion within our communities
Yammer provides a completely user-driven means to create ldquogroupsrdquo ndash which
roughly speaking can map to communities
Following the same logic as our communities through some moderation of new
groups we have pushed for fewer larger Yammer groups to increase their value
Recently the importance of size of a community Yammer group has been
challenged with the assertion that people are more comfortable and therefore
more likely to communicate in a smaller group of people
Background
Deloittersquos use of Yammer
5
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
As part of our Community program Deloitte has adopted the use of Yammer as a
primary tool of discussion within our communities
Yammer provides a completely user-driven means to create ldquogroupsrdquo ndash which
roughly speaking can map to communities
Following the same logic as our communities through some moderation of new
groups we have pushed for fewer larger Yammer groups to increase their value
Recently the importance of size of a community Yammer group has been
challenged with the assertion that people are more comfortable and therefore
more likely to communicate in a smaller group of people
Background
Deloittersquos use of Yammer
5
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
I decided to try to address this assertion with data
My hypothesis
bull Larger communities are more likely to be active and vibrant ndash and therefore
more likely to deliver business value to Deloitte
To test this hypothesis we need to define ldquovaluerdquo ndash which is a hard problem
So I simplify the problem and assume that the value of a community group is
proportional to the activity within that community group
bull Certainly it is true that a community group with no activity offers no value
My Hypothesis
Size does matter
6
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
In order to test the hypothesis I collected data from Deloittersquos global network
The data included
bull All public messages from July 2012 ndash April 2013 including
minus Group
minus Sender
minus Time stamp
minus Threading
bull A listing of all groups as of early June 2013 including
minus Current group size
minus Create date
minus Summary counts of messages and total posters
Data source and Methodology
How to address the hypothesis
7
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
I did not have detailed data about Private groups in our network
I also did not have access to detailed data on ldquoLikesrdquo or ldquoSharesrdquo of conversations
For the purposes of the analysis here I aggregated the data as follows
bull First for each group I averaged measures across months to represent that
group
bull In order to aggregate across the groups I used the size of the group (as of June
2013) and grouped together groups by size
minus Groups are represented by the largest multiple of a power of 10 smaller than
its size though any group smaller than 100 members was grouped into a ldquo0rdquo
minus For example a group of 250 members is grouped into a 200 range a group
of 1180 members is grouped into a 1000 range
Data source and Methodology
Part 2
8
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
In order to provide some context on the remaining slides this chart shows you the
number of groups (both private and public groups) for each size grouping
The Landscape
Overall view of the landscape
9
Group
Size Public
Groups Private
Groups
0 1112 1178
100 223 67
200 81 22
300 42 8
400 33 5
500 18 2
600 17 3
700 7 1
800 2 2
900 4 2
1000 24 4
2000 7
3000 4
4000 2
6000 4
10000 1
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Findings
10
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Activity Levels by Group Size
11
Number of messages
does positively correlate
with size of group
As does the number of
distinct posters
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
You might notice a relatively low correlation in the data on the last slide ndash however
if we remove just 4 groups (the four largest in our network) we get this ndash
significantly higher correlation
Activity Levels by Group Size
Another view
12
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Besides raw size another factor in our health report is a minimum amount of
activity (messages) ndash one message per week
This chart shows the percent of groups for each size that achieve that minimum
The impact of size flattens but from this in order for a group to have a good
chance of meeting this minimum they need to achieve at least 200 members
Achieving ldquogreen communityrdquo status
Minimum activity
13
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
So overall activity does correlate with size ndash is an individual member more or less
likely to contribute based on size
So in larger groups it can be harder to get people to post but once someone does
post they will be as active as in any group
Messages per Member
14
Engagement with the community
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Similar to messages member on a monthly basis the percentage of members
who will post at least once does negatively correlate with size
Similar to activity in the groups if you ignore the largest 2 groupings (the four
largest groups) once a group is larger than 100 members this becomes nearly flat
at 25-3
Percent of members that are active
How likely is someone to say anything
15
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Beside pure levels of activity another important factor is how quickly someone
gets a reply ndash how does that correlate with size
Answer You will in general get an answer faster in a larger group Faster to get an
answer and faster to get value for clients
Time for reply
How long does it take to get an answer
16
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Besides how quickly you might get a reply what is the likelihood of getting any
reply
Not only will you get a reply more quickly you are more likely to get any reply at all
in a larger group
Likelihood of a reply
Are you ever going to get a reply
17
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Another effect of size of a group to consider is the impact on the lurkers in the
groups ndash people who join and read but donrsquot post
One way to approximate this impact is with a calculation I call ldquoknowledge flowrdquo
Knowledge flow estimates the overall impact on the members
This shows a very strong correlation with size (note the Y axis is logarithmic)
Impact on Lurkers
What about the silent majority
18 See httpblogleeromeroorg20081120measuring-knowledge-flow-within-a-community-of-practice
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
The data I have for private groups is very limited ndash just total posts and posters and
size
Based on this though we can compare the percent of members that are active at
each size for public and private groups
Because large private groups
are much fewer in number
the average varies a lot
However overall the likelihood
of a member in a private group
posting does not appear any
different than in a public group
Private vs Public groups
Are people more likely to share in private
19
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Summary
20
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21
Based on the insights here I believe there are a lot of compelling reasons why
fewer larger communities with broader topics provide more business value
Based on the data a community manager may want to encourage members to
post but once they post they seem more comfortable in doing so again
I would encourage you to guide your communities to start broad and go narrow if
the community becomes so active as to be overwhelming to newcomers
What are your thoughts and experiences with regard to this
Summary
Size does Matter
21