Segmentation best practices

Best practices for campaigns and segmentation performance

Every customer has different use cases, so we don't limit what is possible with our segmentation engine. However, each query will add additional overhead and take longer when calculating the segment. Overly complex segments might result in slower sends or might cause the campaign to not send at all. Complicated segments are more difficult to understand when reviewing audiences for a campaign as well, and can result in unintentional targeting of users if there are multiple levels of nested segments. The best segments are the simplest ones.

To make sure you get the best performance on the platform and to optimise your segments, follow the suggestions below.

General best practices - Scheduled single-stage campaigns

On scheduled campaigns to send in the future the prepare upfront feature is helpful as it pre-calculates segments and counts ahead of time so that they send quickly at scheduled time.

Rolling segments allow you to create a pre-prepared audience that is updated regularly and is available to use in breaking news type send now campaigns, which ensures quick sending.

For send now campaigns, it is important to optimise your segments so that the campaign is sent as quickly as possible without taking time to calculate the segments and counts.

Create efficient segments

  • Using CSV files in segments can slow things down. Instead, use built-in user lists or individual user details captured on custom profile attributes (such as favourite sports or VIP status) to create your audience.
  • Keep it simple: The more complex your audience definition gets (lots of OR and AND conditions), the longer it takes to find the right recipients. Focus on clear and easy-to-understand criteria. For example, set up and use profile attributes for your segments.
  • Less is more: Avoid including or excluding large groups of people at once.
  • Focus on user details: When defining your audience, do not mix up profile conditions and device conditions. Profile conditions are based on information stored within a user's profile and related to profile-based channels (email, SMS, etc). Device conditions are based on attributes directly associated with a user's device. This will include information from the Data > Devices section of the platform and device info from segmentation such as OS name, app version, etc. Also, everything related to mobile-based channels, such as app last opened, web-push opt-in, etc.
  • Look back in moderation: When targeting people based on past actions (events, website visits, etc.), limit your timeframe to the last week or two. Looking at years of data can be slow or counter-productive.

When to use workflows

Multi-stage journeys (also known as workflows) are automated journeys for your customers. They are great for sending a series of messages triggered by specific events such as when a new user signs up. However, if you have a very large number of people entering a workflow at once, it might take some time for them all to get through. Because of this, workflows are not ideal for sending a single message to a very large group.

Segmentation vs. Workflows:

Segmentation helps you define the right audience for your message, while workflows help you deliver a sequence of messages based on user actions. Use them for different purposes. For example, consider a new user signup workflow. Users sign up, receive a welcome email, and then receive a push notification the next day, and so on. Since actions are processed one by one this means a campaign sending emails and push notifications to 150,000 new signups could take several hours to complete as each user goes through the workflow individually.