With a split campaign you can determine the most promising version for two or more eMailings. A part of the target group will receive one eMail from each test. The most successful version determined by winning criteria is sent to the remaining profiles of the target group.
Create eMailing (Split)
Split campaigns are created in the folder eMailings with a click on + New, eMailings (Split).
In a split campaign at least two existing eMailings are sent to a certain target group.
Configure eMailings (Split)
These two eMailings are sent to a part of the recipients when they are sent, the size of which you can define.
For each eMailing, you can optionally enter a separate text in the Configuration tab in the split campaign that briefly describes the corresponding eMailing or what distinguishes the selected variant.
A split is carried out in such a way that two parts of the recipients in the target group are first sent one eMail each. After the "winner" eMailing has been determined (automatic determination of the winner eMailing), the rest of the target group is sent the winner eMailing after a certain period of time. However, you also have the option of sending the eMailing in equal parts (classic split campaign 50:50, without winners) to the entire target group.
If at least one email is set that is part of an acceptance process, the split campaign is also subject to the acceptance process. For details on the acceptance process, see the information for regular eMails.
Split campaigns with acceptance/release necessary for shipping
Icon-eMailing-Spilt_wait.jpg Split campaigns waiting for approval
Icon-eMailing-spilt_sharing.jpg split campaigns with sharing issued in the dispatches folder
The recipient filters for the dispatch - such as blacklist - are set directly in the split campaign, as these relate directly to the target group to be set for the dispatch. Settings that are made directly in the eMailings are not relevant for sending the split campaign.
As with "normal" eMailing, a target group, the time of sending and the speed are set.
When splitting the test sendings, you define how many recipients the first part of the sending should go to. A percentage of 5% to 100% of the recipients is set in the selection. This proportion is then distributed over the number of existing mailings. For example, with a share of 30%, if there are two eMailings, each eMailing is sent to 15% of the recipients. If 100% is set, a classic A-B split is made for two set eMailings - i.e. all recipients are sent. With the A/B split, the profiles are distributed randomly to the individual eMailings.
Division for the A/B test
Under the tab Transmission you can set the percentage of the target group that will participate in the split test.
After you have defined the split, you define the criteria for how an eMail is determined as a "winner". Statistical parameters such as the opening rate or the click rate are available here. If a winner is to be determined, at least one criterion must be set. If several criteria are selected, they will be summed up. An exception here is the unsubscribe rate. Of course, it is not the eMail with the most but the one with the fewest unsubscriptions that wins. Finally, you define after how much time the system should determine the winning variant. The default value here is one hour. However, the decision can also be delayed up to one day. If several criteria have been chosen, the criteria are merged internally and a ranked list and therefore the winner is determined. For testing purposes you can set 15 minutes here. However, this is only recommended for internal use. When the specified period of time has expired, the eMail that scored the best overall for all criteria is automatically sent to the remaining recipients. If no winner can be determined - for example, because the target group is too small - any eMail will be removed from the list. Please note, however, that a tie is rather unlikely in a real mailing.
To achieve optimal results with the split campaigns, the winning criteria should correspond to the test variants of the eMailing. Therefore, if you set the opening rate as the winning criterion, you should define different subject variants in the eMailings. Different header images are not meaningful for the opening rate, since the eMailing must first be opened.
With regard to the click rate, the prominence of the image/button, the design and wording of the link text and the arrangement in the eMail are decisive for success.
The winning emailing will be sent to the target group that was originally set to receive the complete mailing. This means that profiles that have been sent out in the meantime will still be sent.
If all settings for sending have been made, sending can be started after a preview with all settings and a final control dialog (comparable to e-mailing). After the start of sending, a copy of the split campaign and all contained e-mailings is created in the folder Mail transfers.
Split campaigns wait until the start of dispatch (see dispatch time) in the dispatches folder.
When starting the dispatch, the split campaign can be attached to a report immediately - just like an eMail. Make sure that you make the appropriate settings for correct display in the report itself. If you have started a split campaign by mistake, delete it from the Dispatches folder. The system now generates the internal information required for the start of shipping. Only in this step are the set filters such as blacklists, duplicate check, etc. checked. Only then is the actual number of addressees determined. After this step, the dispatch to the set recipients of the respective eMailings begins. If there are still recipients for a winning eMail, they will be sent after the set time period has expired. Due to the usually very small number of recipients during testing, it can happen that no profiles remain for sending the winner eMailing due to hardbounces, duplicates, etc.
While waiting for the winner, deleting the split campaign will prevent the winner's email from being sent. However, this will delete all information - even those already sent! If you do not want a winning mailing to be sent, please alert support immediately.