Hi @mchu - thanks for being a part of our Community and for asking a great question on segmentation. 

You mentioned that this is a segment and not a list of 13k subscribers - I’d love to know what your definition is for this particular segment? I might recommend that you segment this group based on action, vs “email starts with”, as that would allow you to be more detailed and provide a more customized/tailored experience for the flows/campaigns you create. Here are some more details on Segment Conditions that you might want to consider for the most engagement. 

You can see best our collection of Segment Best Practice guides here. 

For email segmentation, these are the 7 key segments you can use to boost your email performance: 

1. Cross Sell Segment
2. Location-based Segment
3. Item and Brand-specific Segment
4. Engagement Tier Segment
5. Churn Risk Segment
6. Average Order Value (AOV) Segment
7. Holiday Shoppers Segment

You can learn more about Segmentation for Emails here. Lastly, you may already be doing this, but another recommended segment is VIP Customers. This segment is helpful to run rewards program for your VIPs, or if you want to know who your most valuable customers are. 

Not sure if anyone else has additional Segment best practice to share but I’m sure there are lots of ways to slice and dice it! 
-Cass. 

Hi Cassy, great observation. I am using a “starts with” to basically create a list from a larger list, but using the segment rules.

The definition is something like:

• is in large list A
• is not in small list B
• is not in small list C
• has properties rating = 5
• has properties email starts with aa
• has properties email starts with ab
• …
• has properties email starts with ad

So that we can send to a small group, a certain message to test out open rates.

Does that information make it more clear?

Hi @mchu! 

If you are looking to maintain an existing segment definition and then adjust it, you may find it helpful to start by cloning the original segment. This will generate a new segment with the original definition, which you can then edit.

You could also try taking a random sample of your larger segment (the linked doc says “lists,” but this works for segments as well!). Just note that if you pull multiple random samples, they will not always be mutually exclusive (ie, two random samples could have overlapping members). 

Hope that helps get you closer to your goal! 

THANK YOU SO MUCH