At the top of the results page, on the right of the experiment’s main information, there is the page’s general actions menu.
The options in the “Actions” menu allow you to act on the entire results page :
- Refresh data;
- Reset data;
- Stop collection of data;
- Custom view;
- Save display settings;
- Create alerts;
- Print;
- Export;
- Share;
- Enable bayesian;
- Add multiple testing correction;
- enable CUPED.
Refresh data
This action updates all the data on the page without reloading the page itself. You are thus sure to view the most recent data!
Reset data
Reset data if you want to delete all the experiment results. To reset the data of an experiment is to start the experiment a second time, all the results are reset.
Warning! This action cannot be undone! It is offered only to authorized users, and a confirmation pop-in secures your data in case of an unintended click.
Stop collection of data
Have you collected enough data on your campaign, and were you able to draw conclusions from your results? Then you can decide to stop collecting data on it. If you are used to leaving your campaigns online for long periods of time, this will allow you to have a better loading performance.
Visitors who enter this campaign will no longer be counted and data will not be collected or stored. You can resume the data collection later (in this case, your evolutionary graphs will show a gap between two curves).
This will not impact the status of the campaign, which will remain online.
Custom view
A pop-in opens, giving you the possibility to change the view of the results page.
View per goal/per variation
In the drop-down menu, choose a display of the results by goal or by variation. The presentation of the results in the graphs and tables will then be modified: the displayed curves will represent either the goals or the variations. In the same way, each table will correspond either to a goal or a variation.
Change the reference
By modifying the reference variation, the results (and especially the improvement rates) will no longer be calculated according to the original page of your website but according to the variation of your choice.
To set a reference variation different from the original, switch the button to “ON”. In the drop-down menu that appears, select the variation of your choice from the variations of your experiment.
Hide the results of one or several variations
If you only want to look at certain variations of your experiment, this option allows you to hide a selection of variations from the results page.
Click on the switch button to put it on “ON”; in the drop-down menu that appears, select the variation(s) you want to hide from the page.
Hide the results table of one or several goals
In the same way as with variations, you can hide the goals you do not want to see.
Click on the switch button to put it on “ON” then, in the drop-down menu that appears, select the goal(s) you want to hide from the results page. Associated targets tables will not be displayed anymore.
You can check or uncheck all the goals at the same time in the pop-in.
Save display settings
We offer a default view, but you can set your own display according to your needs.
You can then save it so you will find the same display on your next visit. It is also possible to load a predefined view to apply it to the page.
Save the current settings
To memorize the current display, just name it and, if you wish, set it as default view for this experiment. By checking this option, you will find this display with each new consultation of the results of this experiment.
Then click on “Validate”.
These elements are saved:
- Filters;
- Date ranges;
- Columns of the tables;
- View per goal/per variation;
- Reference variation;
- Display of the goal(s).
Load a predefined view
You can apply to your page previous display settings set up on this same page.
Select a view from the drop-down menu and, if desired, set it as the default view for this experiment.
Then click on “Validate”.
To the right of the selection field you will find a “trash” icon to delete a previously saved display.
To do so, select the display and click on the trash. Gone with the wind!
Create alerts
Configure alerts to be informed by e-mail of the evolution of your results.
First, enter the e-mail address to which you want to receive the alerts.
Note : You can’t enter more than 10 e-mail adresses.
Two predefined alerts are directly available, but you can also set custom alerts.
Predefined alerts
Choose to be informed when your experiment results are ready or when a variation is identified as a winner on the main goal.
Your experiment’s results are considered as ready when a variation has a reliability on the main goal superior to the minimum fixed. A reliable variation is a variation with a reliability higher to 95%. However, you can edit this rate from the setting page of your website.
Click on the switch button of the alert to put it on « ON » and activate it.
Custom Alerts
Custom alerts can be set on any of the data available in your results. Just click on “Custom alerts” to set up.
Setting up a custom alert includes three steps:
In the first drop-down menu, select the data for which you want to be informed of the evolution: improvement rate, conversion rate, reliability, number of visits, visitors or conversions.
- Conversion rate / Improvement rate / Conversions number: As soon as a variation reaches the value you set for each goal you have selected.
- Reliability: As soon as the main goal reaches the value you set for each variation you have selected.
- The number of visits / The number of visitors: As soon as the addition of variations reaches the value you set.
Note: Data is based on converted visits. In order to garanty an appropriate alert, we don’t send any e-mail before 100 visits and 20 conversions.
An alert will be sent to you as soon as the value reaches the threshold you set. Choose whether this threshold should be higher or lower than the value you want and fill in this value.
You can also be alerted when the value of the measured data increases or decreases in proportions that you set.
Finally, in the last drop-down menu, define the variations or goals to which the alert applies.
If you have chosen to be alerted about the improvement rate, the conversion rate or the number of conversions then you will define here the concerned goals. In other cases, you will have to choose among the variations of your experiment.
The “Actions” menu allows you to print the results of your experiment.
Two options are available: the detailed printing of a part of your results or the printing of a global summary.
Print the detailed results
In the first drop-down menu, select the graphs you want to print. All the graphs available in your results page are also available here. In the second drop-down menu, select the goals or variations for which you want to print the tables.
If your results display is set on “Goal view”, the drop-down menu will let you choose the goals you want to print. If the display is set to “Variation view”, it will allow you to choose which variations you want to print.
Then click on “Validate”.
Print the recap
The impression of a summary of the results of the experiment includes:
- The parameters of the page;
- The main results and the winning variation;
- The graph for the main goal/variation;
- The table for the goal/variation.
Then click on “Validate”.
Export
The results of your experiments can be exported under two different formats: Excel format and PDF.
Excel
To export your results in an Excel file, select “Excel” from the drop-down menu.
You can choose which type of data to export and your preferred display of Excel files: tabs per variation, tabs per goal or all data in the same tab.
Then click on “Export” to save the Excel file containing the result tables.
When opening the table, the first tab, “Info”, reminds you of the main information about the experiment: title, date range, and targeting. The following ones depend on the view you have chosen.
For example, the Excel file below corresponds to:
- All data
- View per variation
All data is displayed and each tab is a variation.
To export your results in PDF format, select “PDF” from the drop-down menu.
Then select the information you want to export in the same way as for printing.
The results of an experiment can be shared by e-mail or by sending the URL of the page directly. The page thus shared only allows the consultation and does not give access to the editing options, the setting of the page or the various actions.
To share the results page by e-mail, select “Send report by e-mail” from the drop-down menu.
In the text box, enter up to 10 e-mail addresses separated by commas.
To delete an e-mail address, click on the cross next to it.
Then click on “Share” to send a report containing the results to the e-mail addresses.
To get a link to share your experiment results, select “Share URL of this page” from the drop-down menu.
The link appears directly in the pop-in and two options are available:
- Limit the validity of the link to 24 hours. 24 hours after the creation of the link, it will no longer be valid and will no longer allow access to the results;
- Secure access to this page with a password. Access to the results page from the link will only be possible with the password of your choice. Set this password by entering it in the text block.
Click on “Copy and Close” to confirm the creation of the link, copy it to the clipboard and close the pop-in.
Bayesian
By default, the results of your experiment are presented to you according to a Frequentist analysis of data. Kameleoon also offers you another statistical method: the Bayesian one, which is based on the the deduction of the results’ evolution according to trends seen previously on this experiment and/or on similar experiments.
You can read more about the correction implementation in our statistical paper.
For the details of the Bayesian statistical method, and which one is best for you, check out our dedicated articles: Frequentist or Bayesian Approach? and Results according to the Bayesian method.
To switch the Frequentist results display to Bayesian, click the switch button to the right of the display settings to set it to “ON”.
Add multiple testing correction
This action is only available for experiments with more than one variation. When enabled, our multiple testing correction method will make sure your likelihood of getting a false-positive result stays under your configured level when determining the winning variation. (You can configure the required reliability on a Project-basis under Admin > Projects > [your site or app] > Experiment.)
You can turn on multiple testing correction for individual experiments. Once activated, Kameleoon will continue to display the “corrected” performance metrics on your experiment’s results page.
Note that when you apply this method to your experiment, its performance will likely look different from the results you are used to seeing. Namely, your experiment’s reliability will be lower when multiple testing correction is enabled. This is an expected outcome of enabling our method for multiple testing correction because it corrects for statistical significance to account for issues arising when a test has multiple variations. Furthermore, if you have previously used other correction methods, these numbers will likely differ from those as well.
It’s up to you whether you want to enable multiple testing correction, although we highly recommend using it. (It’s better to use a foolproof correction method rather than an approximate one!) If you choose not to activate it and you want to feel fully confident in making decisions based on the results of an experiment that has multiple variations, we recommend that you wait to reach a higher reliability (e.g., 97% vs. 95%) before calling a winner.
You can read more about the correction implementation in our article in our statistical paper.
Enable CUPED
Controlled-experiment Using Pre-Experiment Data (CUPED) is based on existing data on your former visitors, to speed up the time needed to obtain a reliable result.
You can read more about CUPED in our dedicated article and in our statistical paper.
