1. Urchin (General)
1.1. Database
1.1.1. MySQL
1.2. Profiles
1.2.1. Up to 1000
1.2.2. Can have up to 4 goals
1.2.3. Can apply filters to each profile
1.2.4. Filters
1.2.4.1. 4 per profile
1.2.4.2. RegEx to narrow things down
1.2.5. Used to segment userbase into smaller chunks
1.2.6. Examples
1.2.6.1. Users who have clicked on a given link
1.2.6.2. Users who have deposited money
1.2.6.3. Users from Paraguay
2. Site
2.1. Domains
2.1.1. AbsolutePoker.com
2.1.2. absoluterewardsstore.com
2.1.3. absolutebanking.com
2.1.4. ultimatebet.com
2.1.5. A whole lot more parked domains
2.1.5.1. Are we doing anything with these?
3. Current Omniture Setup
3.1. Affiliate ID
3.1.1. Parameter based
3.1.2. Cookie
3.1.2.1. Tied to an image
3.1.3. 302 Redirect to absolutebanking.com
3.1.3.1. Now a 301
3.1.4. Trying to move affiliates off of the banking system to reduce problems
3.2. Tagging The Client
3.2.1. Game server tags certain actions
3.2.1.1. Create new table
3.2.1.2. Change games
3.2.2. Omniture drops a cookie into IE in the client
3.2.2.1. Problems with FF users
3.2.3. None of this stuff works, currently
3.3. Pages Tagged For Omniture
3.3.1. Download Link
3.3.2. Registration(?)
3.3.3. UltimateBet.com homepage
4. Roadblocks
4.1. Browser
4.1.1. UB's client has a baked in IE client
4.1.1.1. Does it share cookies with primary browser?
4.2. Domains
4.2.1. Correlating users between multiple domains
4.2.2. Integrating reports across domains
4.2.2.1. MAC address
4.3. Client modifications
4.3.1. Hard to get resources to make changes
4.3.2. Multiple clients
4.3.2.1. Windows C++
4.3.2.2. Mac Java
4.4. Hosting Environment
4.4.1. Expensive
4.4.2. I/O is very expensive
5. Email Marketing
5.1. Same umbrella as Affiliate and SM Campaign
5.2. Simply need to add ID
5.3. Track specific campaign ROI
5.4. Integration between email and client is a 'nice to have'
6. Unmet Needs In Urchin
6.1. Realtime
6.1.1. Nobody has come up with a use-case for realtime data yet
6.2. Easy Custom Reports
6.2.1. Nobody wants to use Regex to update profiles
6.2.2. Could be solved by creating a bank of pre-made filters and profiles
6.2.3. Custom Front End
7. Desired Segmentations
7.1. Downloads
7.2. Deposits
7.2.1. With a dynamically changing amount
7.2.2. Repeat
7.2.3. Sum For Life
7.3. Registrations
7.3.1. Broken down by section of registration
7.4. Geography
7.5. Payment types
7.6. Defection per segment
7.6.1. Previous vs new depositors
7.6.2. Revenue for original campaign
7.6.3. Where in the cashier do people defect
7.7. How many users use live chat?
7.7.1. How many of those go on to deposit?
7.8. Language use
7.9. Aggregated Conversion Rates
7.10. Time from registration to deposit
7.11. Acquisition Channel
8. Proof Of Concept Requirements
8.1. Must
8.1.1. DL rate for keyword
8.1.2. Reg for keyword
8.1.3. Deposit for keyword
8.1.4. Deposit split by Geography
8.2. That'd Be Nice
8.2.1. $ per keyword
8.2.2. $ per player
8.3. Simulation
8.3.1. 1 Page
8.3.2. x number of variables
8.3.3. Split even and odd IP addresses
8.3.4. Through authentication
8.4. A/B Testing
8.4.1. In-site pages
8.4.2. Outside campaigns
8.5. Flash Cookie
8.5.1. Or some other workaround for the FF Windows users
8.6. ROI
8.6.1. Can we assign accurate lifetime values to segments?
8.6.2. Can we attach a username to a web viewer
8.6.3. Can we find problem points in our Golden Funnel
8.7. Retention
8.7.1. Fallout Reports
8.7.1.1. Segmented in a few ways
8.7.1.2. Page path
8.7.2. Tag download links
8.7.2.1. Track post-download behaviour
8.8. Acquisition
8.8.1. Back to the Golden Funnel
8.8.2. Good sense of sources
8.9. Geo
8.9.1. Geo-Map
8.9.2. Two-segment reports
8.9.2.1. Keywords vs Geo
8.9.3. Trends
8.10. Testing
8.10.1. Can we accurately track A/B Testing?
8.10.2. Can we accurately track multivariate testing?
9. Client Tagging
9.1. Urchin
9.1.1. Data could be pulled from server logs
9.2. Omniture
9.2.1. Nobody has seen it implemented in a poker client either
9.3. GA
9.3.1. Doesn't have the ability to reprocess log files to access player data
9.4. AP/UB Desires
9.4.1. Browsing the game client is similar to browsing a website
9.4.1.1. UI changes should be done in a similar way
9.4.2. How are users interacting with the client?
9.4.2.1. Click-path analysis
9.4.2.2. What about survey based approaches?
9.4.2.2.1. 100 person sample?
9.4.2.3. Segmenting information by gameplay attributes
9.4.2.3.1. Betting patterns?
9.4.3. Segment players navigation through the client based on these tests
9.4.4. Trending capabilities
9.4.5. Connect pre-registration to client data
9.4.5.1. Track all of the information cradle-to-grave
9.4.5.2. Segmenting player-base
9.4.5.2.1. Loyalty
9.4.5.2.2. Betting
9.4.5.2.3. Winner/Loser
9.4.5.2.4. Entry channel
9.4.5.3. Optimize keywords through client information
9.4.6. Heatmap analysis is important
9.4.7. Specific Tags
9.4.7.1. Pages of lobby
9.4.7.2. Buy-in Window
9.4.7.3. Table
9.4.7.3.1. ????
9.4.7.4. How many people fail to complete tasks?
9.4.7.4.1. Registering for a tournament
9.4.7.5. How much - in chips - are people bringing to the table?
9.5. Things the main DB cannot do
9.5.1. Skin
9.5.2. Heatmap
9.6. Multiple Clients
9.6.1. 3 Per Brand
9.6.2. Back-end connections are --we think-- the same
9.6.2.1. Common database
9.6.3. Mac client represents 5% of current users
9.7. Website conversion is still the most important part of this
10. Testing
10.1. Within Urchin
10.1.1. A/B Testing For Inbound Links
10.1.1.1. Tag two versions of an email or other marketing message
10.1.1.2. Track results
10.1.1.3. Twitter, Facebook, Etc
10.1.1.3.1. There is only one version of a tweet
10.1.1.3.2. Can be overcome by running your own URL shortener to split traffic
10.1.2. Hack to A/B test pages
10.1.2.1. Two versions of a page
10.1.2.1.1. Two funnels starting from these pages and leading to a common goal
10.1.2.2. Javascript redirect 1/2 of visitors of page 1 (original) to page 2
10.1.2.2.1. Include a marker to exclude page 1 traffic from stats
10.1.2.3. Compare results of the two funnels
10.1.2.4. Advantages
10.1.2.4.1. Can segment the test
10.1.2.4.2. All within same service, with same login credentials
10.1.2.4.3. Can be combined with current reporting system
10.1.2.5. Disadvantages
10.1.2.5.1. A hack
10.1.2.5.2. Requires you to keep a 301 for page 2 traffic after the test is over
10.2. Google Website Optimizer
10.2.1. A/B
10.2.1.1. Redirect 1/2 of traffic to a separate page
10.2.1.2. Google will track results
10.2.1.3. Define a goal page within GWO
10.2.2. Multivariate
10.2.2.1. Google will dynamically replace portions of HTML to test multiple changes on a page
10.2.2.2. Used to cause problems with complex layouts
10.2.2.2.1. I haven't heard of this happening recently.
10.2.2.3. Does not work with CMSes which cannot support an unbalanced <noscript>
10.2.3. Disadvantages
10.2.3.1. Requires second set of tracking codes
10.2.3.1.1. Slight performance hit
10.2.3.2. Two interfaces
10.2.3.3. No access to Urchin profiles, filters, or reports
11. The Golden Funnel
11.1. Also known as the Critical Funnel...I just like Golden better. Sorry
11.2. Download
11.2.1. Registration
11.2.1.1. Deposit