Overview
The Random Evaluation optimization rule allows you to randomly fail postback delivery for a defined percentage of conversions. This rule is part of the Optimization Tool and is evaluated as a postback decision rule, meaning it decides whether the postback is sent to the publisher or not.
For a general overview of optimization rules, scopes, and priority, see the Optimization Tool Overview article.
How Random Evaluation Works
When the Random Evaluation rule is active:
each conversion is evaluated independently,
a percentage defined in the rule determines the % of conversions will fail postback delivery,
for conversions that fail the evaluation will postback will not be sent to the publisher.
The conversion itself is still tracked and reported in Swaarm.
Where to Create the Rule
Random Evaluation Optimization Rules are created in Automation → Optimization with
Rule Type: Postback Decision
Rule Name: Random Evaluation
The rule can be applied globally (if fields are left empty), or to specific Advertisers, Offers, or Publishers only.
Configuring the Random Evaluation Rule
Below is a description of each field available when configuring a Random Evaluation Optimization Rule.
Chance - defines the percentage of Approved conversions that will have their postback decision marked as Failed due to random evaluation and will not be sent to the publisher.
For example, if set as 20% - 20% of conversions with status Approved will not be sent to the Publisher.
Event Types - allows you to apply the rule to specific types of events.
If an offer is selected, you can choose one or more specific events, which conversions will be affected by the rule
If no event type is selected, the rule applies to all events
Threshold - defines the minimum number of conversions that must be achieved before the rule starts applying.
If the threshold is not met, the Random Evaluation rule will not be triggered. For example, if Threshold is set as 5, the rule will start acting only after 5 conversions have been received:
💡 Note:
If the rule contains multiple advertiser, offer, or publisher combinations, the threshold applies to their combined traffic, not individually. Deleting and re-creating the rule will reset the threshold.
Threshold Reset - defines how the threshold is evaluated over time.
Daily - the threshold must be reached each day before the rule is applied
Not selected - the threshold must be reached only once; after that, the rule applies continuously
Batch Size - defines how postbacks are evaluated when Randomize is disabled.
Postbacks are processed in batches of the specified size. Example logic:
If Chance = 50% and Batch Size = 2, one postback will be passed and the next postback will be failed.
To edit the batch size, Randomize must be disabled.
Randomize - controls whether the Chance is applied randomly or in a fixed sequence.
Enabled - postbacks are randomly selected to fail.
Disabled - postbacks are failed in a predictable order based on the batch size.
Trigger Only on Paid Postbacks - defines whether the rule applies:
only to paid postbacks
to all postbacks, regardless of payout
Rules Examples
Example 1: Hold 30% of Postbacks for a Specific Advertiser
Goal - randomly prevent 30% of postbacks from being sent to Publishers for Random Advertiser (id 1819).
Configuration:
Rule Type: Postback Decision
Rule: Random Evaluation
Advertiser: 1819 Random Advertiser
Chance: 30%
Randomize: Enabled
Threshold: 0
Threshold Reset: Not selected
Event Types: Not selected (applies to all events)
Trigger only on paid postbacks: Optional, based on use case
Result - approximately 30% of conversions attributed to Advertiser X will have their postback decision marked as Failed and will not be sent to the publisher.
Example 2: Hold 20% of Postbacks for a Specific Offer, Publisher, and Event
Goal - randomly hold 20% of postbacks for Publisher 9 on offer 29322 and event 'Collect Token'
Configuration:
Rule Type: Postback Decision
Rule: Random Evaluation
Advertiser: not mandatory to add it, when the offer is selected in the rule
Offer: Offer 29322
Publisher: Publisher 9
Event Types: Collect Token
Chance: 20%
Randomize: Enabled
Threshold: 0
Threshold Reset: not selected
Result:
Only conversions matching the selected offer, publisher, and event are evaluated. Around 20% of those conversions will not be sent to Publisher 9.
Example 3: Hold 10% of Postbacks Only After Publisher delivers 10 Conversions per Day for all Publishers on a specific offer
Goal - apply Random Evaluation for offer 29321, across all publishers, only after 10 conversions are reached, with the threshold resetting daily.
Configuration:
Rule Type: Postback Decision
Rule: Random Evaluation
Offer: 29321
Publisher: Not selected (applies to all publishers)
Chance: 10%
Threshold: 10
Threshold Reset: Daily
Randomize: Enabled
Event Types: Not selected (applies to all events)
Result: Each day
the first 10 conversions for the offer are not affected,
once the threshold is reached, Random Evaluation starts applying,
the threshold resets the next day and the process repeats.