In performance marketing, there’s a risk of a partner joining your network with the intent to run fraudulent traffic and inflate their earnings. Though there’s no bulletproof solution to partner fraud, you can better protect yourself by catching a potentially fraudulent partner when they sign up. That’s where our partner profile fraud monitoring comes in.
TUNE provides partner fraud monitoring at the profile level to show any signs of potentially malicious behavior. By catching any suspicious information in the partner’s profile, you can be quickly notified to review that partner and even automatically block partners that hit a given threshold of suspicious activity.
Because networks vary widely in business practices and with business partners, profile fraud monitoring is disabled by default. This article explains enabling and adjusting your profile fraud monitoring settings. Also, see our article on partner activity fraud monitoring.
Enabling Profile Fraud Monitoring
Profile fraud monitoring is available for all account plans, and can be accessed from the Customize Application option in the Company menu. Click on Partner Profile Fraud from the Fraud Detection section. Enable this feature by setting the Partner Profile Fraud option in the Fraud Summary section to “Enabled”, then click Save.
Once enabled, you can tailor your fraud criteria to your needs and experiences. The following sections detail the criteria we evaluate and how to set them.
Setting Criteria & Weights
Once you activate the profile fraud system, you can set the weight of each of the optional criteria to determine how much each check will influence the overall score of the partner’s profile fraud. You set which metrics contribute to profile fraud scores.
Fraud scores are percentages ranging from 0% (no fraud criteria triggered) to 100% (all fraud criteria triggered), with values in between meaning some but not all fraud criteria triggered. This fraud score isn’t a guarantee that a partner will or won’t run fraudulent traffic, but a best guess based on your criteria.
Why You Set the Criteria
While TUNE has a recommendation for settings (see below), we force networks to set the criteria so you understand what you’re checking for. Depending on where your partners do business, some criteria are irrelevant and others are more important.
List of Criteria
Here are the criteria that you can set weights for. To modify them click on the Edit button in the Fraud Criteria section.
- P.O. box as an address
- No number in the address
- The address is fewer than 5 characters in length
- State/region and postal code do not match
- Invalid email address domain
- Free email provider
- The phone number does not match the country
- 1-800 phone number
- IP does not match state/region (when the IP address comes from a different region that the one specified in the sign-up)
- Checks are made payable to the individual’s name
- Duplicate sign-up
- Duplicate user (when a sign-up is similar to an existing user, based on the same IP, address, country, state/region, or phone number)
- Account information changed (when a user changes their account information on the same day they sign up)
- Country origin (when the country origin is entered incorrectly)
Weights
Each criterion’s weight can be set to a value from 0 to 10. Any weight set to 0 won’t factor into the fraud score at all. Setting a weight between 1 to 10 is a relative rating: one set to “5” is five times more important to your fraud detection needs than something set to “1”, and half as important as one set to “10”.
There’s also a setting for each criterion named “Immediate Flag”. If that criterion is triggered, the account is flagged regardless of its score with other criteria.
When an account is created, our fraud detection runs the account through those criteria. Each criterion that’s positive increases the fraud score based on its weight. See the Recommended Weights section below for a detailed example.
Our recommended practice is to set those settings important to you at “5” to start, and leave those not important to you at “0”, then adjust up and down based on which criteria are more likely in your experience to indicate a fraudulent account. Anything that immediately makes you critically worried about fraud should be set to “Immediate Flag”.
Recommended Weights
For those unsure of what weights to set, we provide these suggestions and the rationale behind them.
- P.O. box as address: 5
- No number in address: 5
- The address is fewer than 5 characters in length: 5
- State/region and postal code do not match: 5
- Invalid email address domain: 7
- Free email provider: 3
- The phone number does not match the country: 3
- 1-800 phone number: 3
- IP does not match state/region: 5
- Checks are made payable to the individual’s name: 5
- Duplicate sign-up: Immediate Flag
- Duplicate user: 5
- Account information changed: 7
- Country origin: 10 (select all countries but CA, US, and UK)
These settings consider criteria like P.O. box as address or the IP not matching the state or region as average for potential fraud. Since so many people use free email providers like Gmail or Yahoo, that’s likely to trigger a false positive on fraud, so we recommend that be lower than average. On the other hand, we consider someone changing their account information the same day after signing up as more suspicion, though some people notice typos in their account so it’s not worth immediately flagging. But if a user signs up twice, we will flag that second account immediately.
Regarding the country of origin, our recommended settings assume clients do business mostly in the English-speaking world, so this is our default. If you regularly do business with publishers in other regions, set this criterion accordingly.
Example of Fraud Checking
The fraud monitoring system takes the total number of points you have in your criteria and compares that total to the total from the criteria an account triggers, and uses that to make a percentage.
For example, if there’s a sign-up from a free email address that uses a P.O. box and a 1-800 number, that totals 11 points on that account, out of a potential 230 points (the total points in our settings). Based on that, our system estimates a 4.8% chance of the account is fraudulent.
For what that percentage means in your network, proceed to the next section.
Alert & Block Thresholds
The final piece to setting up profile fraud detection is setting thresholds: at which percentages you want an email notification of potential fraud or an account to be automatically flagged. To edit these settings, click the Edit button in the Threshold Settings section.
If an account’s percentage is higher than the Alert Threshold, you receive an alert in your network. If an account’s percentage is higher than the Automatic Block Threshold, that account is immediately blocked and you’re notified. The same is true if any “Immediate Flag” criteria are triggered.
If a threshold is set to 0%, that threshold never triggers. In the case of our recommended default threshold settings, we have Automatic Block Threshold set to 0% so no accounts are automatically blocked.
Managing Fraud with Individual Partners
You can adjust an individual partner’s fraud score by navigating to that partner’s account from the Manage Partners page, then clicking on the View button in the Fraud section.
In the Profile Fraud section of the Fraud page, you can enable or disable any profile fraud flags that were raised by our system. The Activity Fraud section covers activity fraud for that partner. You can also manage custom profile and activity fraud thresholds for that partner in the Custom Threshold Settings section.
Viewing Profile Fraud Report
See our support article on partner fraud report for information on this feature.