1. 1
Facebook Financial Institutions
Norman Niemer
Used Facebook Graph API for effectiveness metrics on fans and posts
Top 10 US financials, 200 recent posts = 2000 data points. Another 1,200 for Citibank global
Stack: Preprocessing Python + Graph API, Storage SQL, Processing R, Excel
Ran a number of experiments:
#1: Potential for financial institutions to interact with customers. Only 3% penetration
#2: Potential for Facebook to gain share of $17bn buget. Only 3% penetration?
#3: A global opportunity. Citibank’s emerging market communities
#4: Who runs the most effective pages? MetLife and B of A. Laggards need to catch up
#5: When to post? Mondays are best, as are afternoon and evening hours
Additional experiments TBD:
Outliers analysis – what defines the viral post?
Weekly reports for clients – what’s on consumers mind?
Solution selling: build relationships with financial institutions by providing advice and solutions
Help them run an effective social media strategy
Norman Niemer: 8 years experience in quantitative data analysis at financial institutions
Entrepreneurial: drove improvements at both small and large companies
Tech affinity: founded tech company in high-school, running SaaS startup now
2. 2
Experiment #1: Potential for Financial
Institutions to Interact with Customers
Significant potential for financial institutions to interact with Facebook users
Chase, Blackrock do not even have a page
494.3
12.6
-
100.0
200.0
300.0
400.0
500.0
Top 10 Financial Institutions
# of Customersvs # of Facebook Likes
(Top10 US Financial Insitutions)
# Customers (MM) # Likes(MM)
Only 3%
penetration
Top 10 US Financial Institutions include: Amex, Capital One, MetLife, Wells Fargo, B of A,
Citibank US, Fidelity, U.S. Bank, Schwab, PNC Bank, Travelers Insurance
3. 3
Experiment #2: Potential for Facebook to Gain
Wallet Share
If user penetration = wallet share penetration, significant potential for Facebook to gain share
within large marketing budgets
Top 10 US Financial Institutions command $17bn marketing budget
16.8
0.4
-
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
Marketing Spend vs Facebook US FS Sales
(Top10 US Financial Insitutions)
Marketing Spend ($bn) Facebook US FS Sales ($bn)
Only 3%
penetration?
4. 4
Experiment #3: A Global Opportunity
Citibank has a very vibrant emerging market online community
Share rates in Singapore significantly higher than in the US (see appendix)
5. 5
Experiment #4: Who runs an effective Facebook
Page?
# shares the highest form of user engagement
Travelers off the chart due to agents
B of A and MetLife register high number of shares, likely shares related to sports events
Capital One to catch up with Amex, Charles Schwab with Fidelity
6. 6
Experiment #5: When to post?
# shares the highest on Mondays
Evening and lunch as well as afternoon hours see highest shares
Companies avoid posting in the morning already
7. 7
“A Tale of Two Cities”
Users post both positive and negative comments
Resources have to be set aside to handle feedback
8. 8
Solution Selling to Financial Institutions
Low Facebook penetration provides both challenges and opportunities
Challenges financial services customers are facing:
Regulatory concerns – some institutions seems to have solved them
Process for providing customer service via Facebook – see previous slide
New medium, still trying to figure out how to use it
Solutions:
Help them run an effective social media strategy
– Not just ads but integrate with their marketing campaign
Help them understand their customers
– Provide regular analytics e.g. sentiment analysis and what’s trending with customers
Provide regular analytics on Facebook campaigns and discuss with clients
Tailor to sub-industry, i.e. banks vs asset managers vs brokers
9. 9
Appendix: Significance Test for Sharing
Engagement Across Citibank Country Pages
Test if post sharing in emerging economies differs from US
Result: share rates in Singapore is significantly different but not in other countries
Tests conducted: F-test for combined means and t-test for pairwise means
Challenge: highly non-normal distribution with many values concentrated around zero
Comparison of % Shares in Different Countries
(Bonferroni Test)
India Malaysia Philippines Singapore
Malaysia (0.21)
1.00
Philippines (0.05) 0.16
1.00 1.00
Singapore 1.47 1.68 1.52
0.00 - 0.00
US (0.06) 0.15 (0.00) (1.53)
1.00 1.00 1.00 0.00
10. 9
Appendix: Significance Test for Sharing
Engagement Across Citibank Country Pages
Test if post sharing in emerging economies differs from US
Result: share rates in Singapore is significantly different but not in other countries
Tests conducted: F-test for combined means and t-test for pairwise means
Challenge: highly non-normal distribution with many values concentrated around zero
Comparison of % Shares in Different Countries
(Bonferroni Test)
India Malaysia Philippines Singapore
Malaysia (0.21)
1.00
Philippines (0.05) 0.16
1.00 1.00
Singapore 1.47 1.68 1.52
0.00 - 0.00
US (0.06) 0.15 (0.00) (1.53)
1.00 1.00 1.00 0.00
Notes de l'éditeur
Dont go through slide by slide – give them printout after presentation! (only show market size slides)