Amadesa

NEW feature! “Pick-the-Winner” for Multivariate Testing
Amadesa is pleased to announce our latest feature release for multivariate testing that includes Pick-the-Winner functionality at the variation level, allowing for greater flexibility in testing content, images, designs and layouts. This new capability gives users the power to test significantly more variables in less time, without sacrificing statistical validity or reliability of the test.

See below a hypothetical Pick-the-Winner MVT self-optimizing process flow

Why the new feature?
Multivariate testing comes in two general options, full factorial or partial factorial. Amadesa uses full factorial, meaning we run every combination, or recipe, until we reach statistical significance and identify the winning combination. The alternative is partial factorial, wherein only a subset of combinations run, and the results are extrapolated across all combinations.
With partial factorial, the tests are generally constructed in a rigid format for consistency. The problem is that with both factorial tests is that it takes time to complete an optimization given the amount of traffic required to reach significance across the combinations. (By the way, full and partial factorial takes the same amount of time and traffic to run, contrary to our competitors’ claims.)

The solution—Pick-the-winner
Once an optimization is created, the user can set the multivariate test to run all combinations, or, by using Pick-the-Winner, narrow down the total combinations by identifying a single best alternative (variation) for each element. Customers already choose this option for A/B tests, now we’ve brought the functionality over for Multivariate tests.

How does it work?
Setup is simply an option on the ‘Activation’ step within the workflow. No other change to the workflow has been made. All Amadesa clients with multivariate testing product module enabled for the account will automatically have access to this feature. With Pick-the-Winner, set the KPI to conversion or revenue per visit, confidence, length of initial run and whether to keep the control group running. Here is a screen shot of the rules to set:

Getting started
For customers already utilizing the Multivariate testing product, use this option on your next optimization (or stop an existing optimization to reset but careful this will reset the test and therefore ‘start over’).
For new customers, start slow. Build a multivariate optimization with two elements, each having one alternative. Best practices for A/B use Pick-the-Winner, and now, with multivariate testing, the same best practice is recommended.

Benefits of Pick-the-Winner MVT:
- Automated and requires no maintenance by the user, as the feature self-optimizes based on initial business rules set.
- Provides rapid results, by allowing for more multivariate optimizations to run since each individual optimization will take less traffic to complete.

Additional reporting enhancements include:
- Reporting for Element Comparison analysis is available with revenue per visit information
- Improved “Combination Comparison” report for tracking MVT serving rates