FW Omniture Summit 2008 Wendsday 2-3 BreakOut | Evolving from manual to Automated Optimization

Break out Session Testing and targeting
Brent Hieggelke Omniture VP Strategic Marketing
Worked at touch Clarity before it was acquired by Omniture
Why Automated Optimization in mandatory
Off site resources Email Marketing, affiliate programs behavioral marketing Paid search, banner, seo, ofline marketing to web instore web promotion
Onsite resources are limited and represent a large investment gap.
Close to 10billion to convert less than 4%
We are dumping lots of traffic but we are not getting conversion, we need to optimize the customer experience? Web sites are monologs, Customers demand dialog, to be serviced how they want to be serviced.
Targeting enable dialog in several ways
Sample visitors dictate best content and layout
Deliver different content based on the customer
Deliver content based on behavior
Testing: Landing page optimization Creative optimization, flow and form optimization, optimized for general population, optimize for segments(search engine, brands vs category) AB testing has been done since 2000 but it was manual and rudimentary.
Targeting: Segmentation targeting, 1to1 predictive takes it to the individual customer, used for less clear segments and increases the relevancy of key pages. Leveraging data to drive content, Real time listening and response, Target vs control group.
When do you test and when do you target? Every site should do both. Merchandising display, layout, and presentation. Sales and service forks who listen to customers and learn about preferences and opinions
AB testing. Which creative works best. It works on pages and versions that work the best.
What does multivariate Testing (MTV) accomplish measured which combination of elements work best. Basically it is AB testing on steroids., move individual items around on the page.
Targeting example musiciansfriend.com You can segment based upon who goes to which tab at the top and then display banners on the home page on all future and following events.
Automated 1 to 1 predictive Targeting What is most likely and engaging for this visitor at this time. Info done in real time. Real time data mining, predictive modeling, continuous learning and it is reactive to market changes.
Case example in Bank of America detects activity and serves content
Site behavior variables, temporal variables, environmental variables,
Different types of Optimization: Testing and targeting
How Auto Snything.com uses automated optimization to increase revenues
Wrap up questions
Trevor Klein owner Auto Anything:
User experience will be the area that differentiates your site.
History of using tools, primarily to maximize margin dollars on promotions Chose offermatica for 1. Available ad DIY vs Managed Solution 2. Simplicity in setting campaigns, 3. Price
Implementation ius much lighter than thought, Prior of Omniture make things easier
Tested: Offers, colors, promotions, images, search function, category affinity
Offer testing Putting a minim on free shipping. BEFORE YOU DO A TEST HAVE YOUR METRICS AND KPI”S CLEARLY DEFINED. One my win by use one may win by revenue
-Every site can segment in one way or another.
We used offermatica to serve up two containers, goal was to serve up relevant content and decrease abandonment
Moving on to behavioral targeting solution Implamented via existing Omniture tags.
Serving the right offer to the right person at the right time
Took learning’s from test& target category affinity
These kind of predictive analysis
Tips
Be clear on exactly what you are testing
Always have a control in order to evaluate potential lift
Once the test has been statistically proven do not serve the control offer to a large percentage of users or the benefit will be minimized
Create tests/user experiences that are unique to achieve significant results.
To create tests/user experiences that are unique in order to achieve significant results
We preach that you should put 15% of your add spend for optimization. That will make that other 85% that much more effective.

Wow that is a whole lot of info. Please ask me if you have any questions with my translation.