Market Ruler HelpDifference between revisions of "A/B Split testing"
| (3 intermediate revisions by the same user not shown) | |||
| Line 5: | Line 5: | ||
== Mechanism == | == Mechanism == | ||
| − | When a visitor comes to a website, the site implements a means to divide visitors into one or more groups (The "A" group, the "B" group, etc.) ConversionRuler customers can download a script to do this (available in [[PHP}] or [[ASP]] - ask for other implementations). Best case scenario | + | When a visitor comes to a website, the site implements a means to divide visitors into one or more groups (The "A" group, the "B" group, etc.) ConversionRuler customers can download a script to do this (available in [[PHP}] or [[ASP]] - ask for other implementations). Best case scenario is to divide traffic equally between both sources. Typically the "division" can be handled by a [[redirect]] to another set of pages. |
When splitting traffic: | When splitting traffic: | ||
| − | * Do so equally, or consistently, to be able to compare leads to conversions | + | * Do so equally, or consistently, to be able to compare leads to conversions |
* Pages should have analogous "conversion points" - meaning you should not have one page have a "Buy Now" button and another lead to a dead end, for example | * Pages should have analogous "conversion points" - meaning you should not have one page have a "Buy Now" button and another lead to a dead end, for example | ||
* [[Conversion Tracking]] should be installed on each page, and pages should record differently | * [[Conversion Tracking]] should be installed on each page, and pages should record differently | ||
| Line 17: | Line 17: | ||
The larger your sample size for A/B Testing, the better results you will get. When your [[Conversion Tracking]] software runs a report, you should be able to get something like: | The larger your sample size for A/B Testing, the better results you will get. When your [[Conversion Tracking]] software runs a report, you should be able to get something like: | ||
| − | + | {| cellspacing=10px; cellpadding=10px | |
| − | ! | + | |- |
| − | ! | + | ! scope="col" |Test |
| − | ! | + | ! scope="col" | Landings |
| − | ! | + | ! scope="col" | Orders |
| + | ! scope="col" | Amount | ||
|- | |- | ||
|A | |A | ||
| Line 34: | Line 35: | ||
|} | |} | ||
| − | In this case, you can see that '''Test A''' had more conversions, but had a lower revenue | + | In this case, you can see that '''Test A''' had more conversions, but had a lower revenue total. Depending on the value of return customers for a campaign, marketers can make a decision based on the effectiveness of each message and the desired result to maximize profit. |
[[Category:Technical]] | [[Category:Technical]] | ||
[[Category:Glossary]] | [[Category:Glossary]] | ||
[[Category:Marketing Tools]] | [[Category:Marketing Tools]] | ||
Latest revision as of 20:21, 10 July 2020
A/B Split testing (or A/B Testing) is a marketing term which describes testing two variations of a marketing message on customers to determine which message is more effective in producing Conversion Actions by customers.
The "A" and "B" refer to the two marketing messages or designs. New Page Here
Mechanism
When a visitor comes to a website, the site implements a means to divide visitors into one or more groups (The "A" group, the "B" group, etc.) ConversionRuler customers can download a script to do this (available in [[PHP}] or ASP - ask for other implementations). Best case scenario is to divide traffic equally between both sources. Typically the "division" can be handled by a redirect to another set of pages.
When splitting traffic:
- Do so equally, or consistently, to be able to compare leads to conversions
- Pages should have analogous "conversion points" - meaning you should not have one page have a "Buy Now" button and another lead to a dead end, for example
- Conversion Tracking should be installed on each page, and pages should record differently
Analyzing Results
The larger your sample size for A/B Testing, the better results you will get. When your Conversion Tracking software runs a report, you should be able to get something like:
| Test | Landings | Orders | Amount |
|---|---|---|---|
| A | 1023 | 56 | $394.32 |
| B | 1021 | 32 | $561.44 |
In this case, you can see that Test A had more conversions, but had a lower revenue total. Depending on the value of return customers for a campaign, marketers can make a decision based on the effectiveness of each message and the desired result to maximize profit.