What is Repeat Chance?
Repeat chance is applying a chance that any content will appear. If you set repeat chance to 0%, you would thus get no content. This is why 1% is the minimum possible repeat chance %. As you increase your repeat chance, you are opening up your pool of content to repeat in the future so you get more and more content added to the feed based on the other criteria. The lower the chance, then the less likely it is that a piece of content will be included in successive feeds. The higher the repeat chance, the more potential content from the pool of available content will repeat the next day.
For example, after taking all the data feed setting requirements into consideration, you have a pool of 1,300 available content to be used in the feed.
- 1,300 eligible items * 5% chance of getting into the feed on any given day = ~69 pieces of content
- 1,300 eligible items * 30% chance of getting into the feed on any given day = ~390 pieces of content
With a repeat chance of 5%, the feed only has ~69 pieces of potential content to initially display.
Unlike Fresh, Evergreen is also not pulling content based on recency. So you are getting a random selection of the entire content collection, after it has taken all requirements into consideration. As repeat chance increases (eg, 30%) you will get more content pulled into the feed, and that content will contain more unique items.
Why can't Evergreen feeds have a number to control the minimum content that goes into the feed and the repeat chance?
Having that kind of option would actually cancel out the functionality of at least one of those numbers. If there was a number to control minimum number in the feed and repeat chance, the two may contradict each other.
Taking the following example into consideration:
- The total content pool is 1,300
- A minimum of 50% content is set -- 50% of 1300 is 650
- The repeat chance is set to 5%
With 1300 content available, requiring 650 pieces of content from one send to the next will cause content to be repeated more than 5% of the time. Unless enough new content is published day-to-day, the content will likely repeat about 50% of the time from one send to the next. Therefore, maintaining a 5% repeat chance would not be possible.
Instead, with the current functionality, 65 pieces of content will be pulled in (5% of 1,300). This will ensure that in the next pull of 65 pieces of content, the content from the last send will only repeat 5% of the time.
Whenever the feed is accessed, we pull in the total number of articles based on the repeat percentage to use in an email. The higher the repeat chance, the more content we pull in up to 1k items, and the more likely content will repeat in the future. The lower the repeat chance, the less content will be pulled in up to 1k content items, and thus the less likely that content will be repeated in the future.
For more on data feeds, see: