Preventing Attribution Fraud: Overview

Singular’s Fraud Prevention Suite actively protects your marketing efforts using a mixture of out-of-the-box detection methods and custom user-defined rules. Rejected installs won’t be attributed or reported to the partner preventing billing for CPI/CPA campaigns.

The fraud prevention process consists of four main steps:

First, Singular collects all the touchpoints relevant to the install. Touchpoints include every end-user event that might have led to the install, such as:

  • Clicks
  • Impressions (view-through)
  • Self-Attribution Network touchpoints (from Facebook, AdWords, Twitter, and Snapchat)
  • Google Play Install Referrer and additional special touchpoints.

When the app is first opened, an install event is triggered and sent to Singular’s servers. Singular logs the install and collects the matching touchpoints.

The Fraud Engine then receives the install event information alongside each of the touchpoints. The Engine decides for each touchpoint, based on the activated rules/methods and their respective actions, if the touchpoint is:

  • Valid - the touchpoint is valid and can be attributed to this install as usual.
  • Suspicious - the touchpoint is suspicious, which means it can still be used for attribution, but if used the attribution will be marked as suspicious and will appear in the suspicious report.
  • Reject - the touchpoint is rejected and it will not be used for attribution. Additionally, it will be reported in the rejected report.

Finally, based on the Fraud Engine’s decisions, Singular makes a final attribution by choosing the highest-priority touchpoints that were not rejected. If all touchpoints have been rejected, the install is marked as either Organic or Untrusted (see Untrusted Installs).

Every rejected touchpoint is counted and presented later on in the report. Note that one install can trigger multiple rejections.

Examples

Use Case #1

# Touchpoint Type Decision Effect
1 - Highest Priority TP Click Rejected Click #1 is being rejected and counted in the rejection reports.
2 Click Valid The Install is attributed to Click #2, postback will be sent to the relevant partner.

Use Case #2

Special Cases

One Install Triggering Multiple Rejections - Once the Fraud Engine finishes evaluating all the touchpoints, every touchpoint that has been rejected is added to the rejection count. This means that a single install with multiple rejected touchpoints is counted more than once in the rejection reports.

Untrusted Installs - If the install itself is rejected (for example: if it comes from a blacklisted IP and the system is set to reject on blacklisted IPs), the install is attributed to Untrusted and marked as suspicious. We exclude these installs from Organics because a rejected install is characteristic of a bot and we want to enable tracking of bots without polluting the Organics report.

# Touchpoint Type Decision Effect
1 - Highest Priority TP Click Rejected Click #1 is being rejected and counted in the rejection reports.
2 Click Suspicious The Install is attributed to Click #2 but marked as Suspicious, postback will be sent to the relevant partner.
# Type Decision Effect
1 - Highest Click Rejected Click #1 is being rejected and counted in the rejection reports.
2 Click Rejected

Click #2 is being rejected and counted in the rejection reports.

3 Click Valid The Install is attributed to Click #3, postback will be sent to the relevant partner.

 

Was this article helpful?
0 out of 0 found this helpful