Web Analytics: A new Statistical Domain
-
Upload
paul-askew -
Category
Business
-
view
204 -
download
0
description
Transcript of Web Analytics: A new Statistical Domain
Web Analytics
A New Statistical Domain?
Paul Askew
Royal Statistical Society
2010 International Conference 13-17 September 2010
Brighton, UK
Introduction
Web analytics:
Measurement, collection, analysis and reporting of internet data for understanding and optimising web usage. (WAA). …. reporting of metrics
1. Domain
2. Data
3. Measures
4. Tools
5. Opportunities
1. Domain
• 8,722,474 UK web sites (Nominet Aug 2010)
• 81,632,634 Site management transactions (Aug 2010)
Large number of sites, and increasing
1. Domain
No. 10 (Aug 2010)
• 498,871 unique visitors
• 721,767 visits
• 2,135,427 page views
Direct.gov.uk (Aug 2010)
• 15,107,447 visits for 52,687,308 page views
BBC Radio (July 2010)
• 3,477,571 visits for 21,071,588 listening hours
Large (public) sites with lots of activity
2. Data
home
Log File
html java
Service
A B AB C
track every click
2. Data
Phase 1: Visits
Phase 2: Characteristics
• Browser
• Source (incl. search engine, ‘spiders’)
• Date/time, entry/exit pages, duration
Phase 3: Engagement
• Dynamic content
• Blogs and forums – sentiment
• Social Media – networks
The scope of the data is increasing and complicating
2. Data
Overall
• Large volumes in real time
• Precise and accurate
• Consistent (ABCe)
Issues
• Exit time
• Cookies and Java
• Hotel problem
Data has some defining characteristics
1. Measure of activity
• Unique visitors
2. Measures of effectiveness
• Bounce rate, conversion rate
3. Measures of relationship
• More process than event based (sequence detection)
• Frequency....loyalty
• Propensity….days and visits to action
Measures evolve from marketeers and web designers
3. Measures
4. Tools
a. DIY
b. Sector Specific
Google (+74)
Omniture
Technorati
etc…
c. Generic
SPSS,
SAS
etc…
…are maturing
4. Tools
5. Opportunities
1. Public good
• Online services and user interface/experience
• Age of austerity – tougher decisions, tighter evidence
2. Focus on messages from the data
• 90/10 Rule, insufficient expert capacity
• Value of narrative commentary (eg UKSA)
3. Real time vs Strategic analysis
• News vs trends
4. Statistical Opportunity
• Data volumes, issues, new techniques?
5. Opportunities
5. Experimentation and geolocation
6. Visualisation
7. Multiple data sources
8. Free data and free tools
9. Role for Meta-Meta- Data (‘sweater’ data?)
10. Interesting and challenging…
5. Opportunities
“A new era is dawning for what you might call the datarati….The sexy job in the next 10 years will be statisticians” (Google, Jan 2009)
“A society in which our lives and choices are enriched by and understanding of statistics” or “Understand the society and world we live in, and get the most out of our lives.” (Getstats, Sept 2010)
Do let me know how you get on with web analytics [email protected]