1. Optimising the Facebook Stack
Mat Morrison
Thursday October 6, 2011
Draft for Chinwag Insight: Facebook Marketing
2. How people use Facebook
Ignore,
Login
to
View
Leave
Like
or
Facebook
Newsfeed
Facebook
Comment
1
3. How people use a Facebook Page (Client 1)
∑uwd e e e
edges e
EdgeRank
See
Brand
See/Don’t
Ignore,
Post
in
See
Like
Page
Like
or
News
Future
Comment
Feed
Posts
Like source First exposure Responders Ongoing Exposure
0.6% 11%
36%
99%
0.6% attributed to Estimate Comments + Likes DAU
“on Page” Likes Total Fans
DAU 2
4. How people use a Facebook Page (Client 2)
∑uwd e e e
edges e
EdgeRank
See
Brand
See/Don’t
Ignore,
Post
in
See
Like
Page
Like
or
News
Future
Comment
Feed
Posts
Like source First exposure Responders Ongoing Exposure
1.4%
21%
60%
99%
60% in-Ad unit Estimate Comments + Likes DAU
0.5% attributed to Total Fans
DAU 3
“on Page” Likes
5. Most people don’t visit the Page (Client 1)
All Fans
100%
470K
MAU
53%
250K
DAU
11%
52K
Daily Page Visits (Unique)
0.3%
1.4K
4
6. Most people don’t visit the Page (Client 2)
All Fans
100%
56.7K
MAU
99.7%
56.4K
DAU
21.3%
12K
Daily Page Visits (Unique)
2.1%
1.2K
5
7. Response Windows (Client 1)
50%
• 80% of responses within 3 hours.
91.4%
• 90% within 6 hours
40%
30%
20%
10%
0%
0
6
12
18
24
30
36
42
48
Elapsed
Hours
%age
responses
cumulaNve
6
8. Response Windows (Client 2)
35% • 70% of response within 3 hours
• 85% within 6 hours
30%
25% 84%
20% 69%
15%
10%
5%
0%
0 6 12 18 24 30 36 42 48
% response cumulative
7
9. Activity by hour and day (Client 2)
Posts by hour Posts by day
50 60
45
40 50
35 40
30
25 30
20
15 20
10 10
5
- 0
0 6 12 18 Mon Thu Sun
8
10. How fan growth Affects Daily Active Users (Client 1)
90 600 11.60
Thousands
Thousands
80 11.40
500
11.20
70
y = 1.6541x - 10.598
11.00 R² = 0.59878
60 400
10.80
ln(DAU)
50
DAU
Fans
300 10.60
40
10.40
30 200
10.20
20
100 10.00
10
9.80
12.50 12.60 12.70 12.80 12.90 13.00 13.10 13.20
0 0 ln(Fans)
Feb Mar Apr May Jun Jul Aug
1% increase in fans leads to 1.65% increase in DAU
9
(0.35% increase in MAU)
12. What’s the impact of Post Frequency?
Count of Post Frequency Weekly Post Frequency Trend
25
90
40%
80
20
70
60 27%
15
50 23%
40
10
30
20 7%
5
10 3%
0
0 1 2 3 4 0
Posts Per Day Feb
Mar
Apr
May
Jun
Jul
Aug
13. Impressions grow strongly inline with post frequency
3500 15.5
Thousands
y = 1.108x + 11.63
21 15
3000 R² = 0.94383
14.5
2500 17 17
14
ln(7-day rolling imps)
2000
13.5
12
1500 13
12.5
1000
4 12
500
11.5
0
11
Jan Feb Mar Apr May Jun Jul
0 0.5 1 1.5 2 2.5 3 3.5
7-day rolling imps 7-day rolling posts ln(7-day rolling posts)
12
14. Reach grows with post frequency
25%
-1.4
-1.6
20%
y = 0.4005x - 2.3461
-1.8 R² = 0.5301
15%
-2
ln(reach)
10%
-2.2
-2.4
5%
-2.6
0%
-2.8
Posts
Reach
0 0.5 ln(posts) 1 1.5
13
17. Active Users increase with fan growth: Daily Reach
around 20% (Client 2)
DAU
vs
Fan
Growth
Fan
Reach
vs
Fan
Growth
20000
60%
70,899 70,899
18000
50%
16000
14000
51,508 51,508
40%
12000
Reach
DAU
10000
30%
8000
20%
6000
19,857 19,857
4000
10%
2000
2,051 2,051
0
0%
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
DAU
Fans
reach
Fans
16
18. Unsubscribes grow strongly in line with active users
(Client 2)
700
7
y = 1.0593x - 5.1934
600
6 R² = 0.93978
500
5
400
Ln(Unsubs)
4
300
3
200
2
100
1
0
0
Jan
Feb
Mar
Apr
May
5 6 7 8 9 10 11 12
7-‐day
Unsubs
WAU
ln(WAU)
17
19. So unsubscribes grow strongly inline with fan growth
(Client 1)
250
5.5
y = 2.2197x - 24.016
R² = 0.50316
467,512 5
200
4.5
ln(unsubscribes)
150
288,631
4
100
3.5
50
3
0
2.5
Feb
Mar
Apr
May
Jun
Jul
Aug
12.5 12.6 12.7 12.8 12.9 13 13.1 13.2
Daily
Unlikes
Fans
ln(fans)
18