What is it?
Top Performers allows you to pick a selection of candidates and only drive traffic through them.
Why would I use it?
- Maximize site conversions by easily driving traffic to your top performing candidates
- Stop exploration and focus traffic on your top performers
- Reduce risk and increase revenue during important sales periods (e.g. Black Friday)
- Maximize ROI on a campaign within a fixed time period
- Increase confidence in candidates
- Acts as a separate A/B/n test
- Reduce false positive winner selection errors
Where can I find it?
In the report view, it can be found in the top right - as soon as we have seen enough data to apply our Bayesian approach and declare the leading candidates it is available.
How does it work?
When you want to move your experiment into top performers mode - either to maximize conversions or to validate a candidate before implementation, click the 'Test Top Performing Candidates' button in the report view.
This brings up a screen to pick the top performers. By default, Ascend selects the top 3 candidates with > 80% probability to beat control. You can customize this selection by adding or removing candidates.
A confirmation of the selected candidates will then be shown - showing the performance information required to check you have picked the correct candidates.
The experiment will then move into top performers mode:
- Evolution will stop
- The selected candidates will be cloned – so they have a clean start with no conversion data
- Ascend will stop allocating users to all other candidates
The report view will now have a 'Top Performers' group available in the dropdown – this will be the default selected option. Reporting will look exactly as it does in evolution mode, but only showing information for the top performer candidates (including the best candidate data, timeseries and visitor counts).
The top performer candidates are clones of the original evolution experiment candidates. The info card for the top performer candidate has a link to the original candidate in the evo experiment
The original evolution experiment results are available from the group drop-down.
Individual candidates can now be selected from the 'CANDIDATES' dropdown