Multivariate Testing for Greater Lift and Conversion #SESSJ
This Search Engine Strategies session was geared toward those who are passionate about testing and improving their website’s effectiveness.
If your tests doesn’t yield any winner or losers, which happens, then test something else. When Google worked on the Website Optimizer sign up page they tried Video on the landing page and it was a big loser.
Advantage of multivariate over A/B because is if you have 4 variables and 4 values you would have 256 possible combinations. The winning elements of this landing page resulted in a 27% increase conversion rate. Match.com uses Optimost as their multivariate testing tool. They changed the button from “Join Now” to “Continue” and got a 10% lift. Test buttons! Adding a yellow background added 2.2% conversion lift. Changing styles of links and making button background green yielded a 23.5% increase in conversion.
It’s a game- have fun!
Everyone agreed that the most valuable target audience is the visitor that is already on your website! In other words, don’t push your customers away.
Factors that impact an experiment include internal and external factors. Challenges may include dynamic pages and Ecommerce sites that require implementation – it will vary based on platform. Set-up of test and length of time are real challenges. When a test runs over 4 weeks there may be external factors that are impacting your results. Reduce the duration of test by reducing the # of test combinations. The presenter recommended working from micro to macro conversion model.
Assurance centers on the cart pages generally helped in test where there were trust issues- but in some cases it can negatively impact. Again- test for your unique situation.
PGiFax2Mail case study yielded a 235% increase in conversion rate
"Blended" or "weighted" metrics allow you to measure incremental success using multiple criteria
Advantage of multivariate over A/B because is if you have 4 variables and 4 values you would have 256 possible combinations. The winning elements of this landing page resulted in a 27% increase conversion rate. Match.com uses Optimost as their multivariate testing tool. They changed the button from “Join Now” to “Continue” and got a 10% lift. Test buttons! Adding a yellow background added 2.2% conversion lift. Changing styles of links and making button background green yielded a 23.5% increase in conversion.
It’s a game- have fun!
Everyone agreed that the most valuable target audience is the visitor that is already on your website! In other words, don’t push your customers away.
Factors that impact an experiment include internal and external factors. Challenges may include dynamic pages and Ecommerce sites that require implementation – it will vary based on platform. Set-up of test and length of time are real challenges. When a test runs over 4 weeks there may be external factors that are impacting your results. Reduce the duration of test by reducing the # of test combinations. The presenter recommended working from micro to macro conversion model.
Assurance centers on the cart pages generally helped in test where there were trust issues- but in some cases it can negatively impact. Again- test for your unique situation.
PGiFax2Mail case study yielded a 235% increase in conversion rate
"Blended" or "weighted" metrics allow you to measure incremental success using multiple criteria
- Considerered Purchases
- Lengthy Applications
- Offline Conversions
- Consistent User Behavior
- Offline Conversion Rate
- Average Order Value
- Choice of Success Metrics
Comments