Random Sampling allows you to target an interaction to a percentage of eligible customers. For example, you might choose to display a survey to only 5% of users who meet your targeting criteria.
This helps you:
- Reduce survey fatigue
- Control response volume
- Collect a representative sample of your audience
How to use random sampling
Random Sampling is configured within the Who targeting section of your interaction.
To apply random sampling:
- Navigate to your interaction’s Who targeting.
- Add the Random Sampling rule.
- Enter the percentage of eligible users who should see the interaction.
You can:
- Apply different sampling percentages to different segments
- Use the same percentage across all segments to control overall audience size
How random sampling works
Alchemer Digital assigns a random number between 1 and 100 to each user interaction based on the conversation ID.
This number is then compared to your defined sampling percentage:
- If the assigned number is less than or equal to your sampling percentage, the user is included
- If the number is greater than your sampling percentage, the user is excluded
Because this process is random, results will vary in smaller samples. As your sample size increases, the distribution becomes more accurate.
For example, a 50% sample may not result in exactly 50 out of 100 users, but it will trend closer to 50% over a larger population.
Changing random sampling on a live interaction
When you update the sampling percentage, previously assigned users are re-evaluated using their original random number.
Example
Sampling ID:
id1- Initial sampling percentage: 20%
- Assigned random number: 15
Since 15 < 20, the user is included and sees the interaction.
If the sampling percentage is updated to 10%:
- The assigned number (15) does not change
- Since 15 > 10, the user is no longer eligible
- Previously collected responses are retained
If the sampling percentage is later increased to 25%:
- The same number (15) is evaluated again
- Since 15 < 25, the user becomes eligible again
Sampling across segments
Random sampling is applied at the interaction level, not per segment.
If multiple segments use random sampling:
- The same randomly assigned number is used across all segments
- A user will either qualify or not qualify consistently across those segments
Testing random sampling (debug builds)
To make testing more predictable, apps built by your mobile developers for debugging purposes (debug builds) use a fixed sampling behavior.
- Debug builds always assign a 50% random sampling threshold for all interactions
- Production builds (such as those released to the App Store or Google Play) use true randomization for each user and interaction
Testing recommendations
- To confirm users are included in the sample:
Set your sampling percentage above 50% - To confirm users are excluded from the sample:
Set your sampling percentage below 50%
This approach allows you to validate other targeting criteria without relying on unpredictable randomization.
Key considerations
- Random sampling helps ensure a balanced and manageable response set
- Sampling behavior is consistent per user based on their assigned random number
- Debug builds behave differently from production apps for testing purposes
