¶ Low Session Time Anomaly
Low Session Time Anomaly filter protects against low-quality or suspicious conversions by evaluating session duration patterns. It prevents conversions from being processed when a high percentage of them occur within an unusually short time after the initial click, suggesting potential fraud.
This filter is often used to detect incentivized traffic as it checks for patterns where the session time (click to conversion) are generally low. The thresholds can be adjusted.
This filter can also be considerd to detect incentivized traffic and it operates based on flexible thresholds and historical trends.
It considers overall traffic behavior, ensures there is enough data to draw meaningful conclusions, and can optionally evaluate traffic quality at a more granular level (e.g., per Sub ID).
¶ Key factors influencing this analysis include:
- The proportion of conversions with unusually short session durations.
- The volume of data required before analysis becomes active.
- Historical context used to benchmark traffic behavior.
- The ability to isolate and respond to anomalies at a sub-id level as well.
These mechanisms work together to prevent the acceptance of low-trust conversions without requiring manual intervention or client-side configuration.
¶ Case Study
This filter compares the Average CTIT/Session Time values and detects outliers with lower session time averages. These reasons for rejection are often flagged on Rewarded traffic as the users show a lower session time than normal traffic.
The low session time also indicates a lower engagement with the game.
Lower ARPU (Average Revenue Per User) values can be expected.
High flaggings on non-rewarded traffic are problematic as
- Rewarded traffic may run on non-rewarded campaigns
- Bots that install the App significantly faster than regular humans.
This is especially problematic if the percentages of the flaggings are high.
For rewarded Traffic Partners instead, this should not be eventually rejected.
Argument with Partner
The CITT shows significantly lower session time averages than other non-rewarded traffic sources. On this campaign we do not allow Rewarded Traffic or other forms of incentivization for the users.