Mobile Network Fraud & Brand Mitigation for RMG Advertisers

The following document outlines Flatiron’s approach to mitigating mobile network fraud and brand risk.

Mobile Network Fraud & Brand Mitigation for RMG Advertisers


The following document outlines Flatiron’s approach to mitigating mobile network fraud and brand risk.

Fraud mitigation: Implement a process to ensure any bot or fraudulent traffic is promptly identified and refuted/refunded (richer data helps to drive early detection)

Brand risk management: Implement a process to ensure ads do not fall alongside any offshore or other unsavory entities

Here are the 4 key phases:

  1. CONTRACTING PHASE

The initial contracting/IO phase provides the groundwork to ensure you have the base infrastructure to help minimize fraud and brand risk.

  • Set initial expectations of client concern over fraud and brand risks
  • Outline large budget opportunities if these risks are mitigated
  • Request initial campaigns to start with top-performing whitelisted publishers for our category
  • Request details around specific placement, including content on property, demographics, and other types of publishers
  • Contractually exclude traffic sources like pop-up, incent, adult, tobacco, firearm, and politics, including specific language around exclusion of any source with Offshore book advertising, partnership, or placement.

2. ONGOING MONITORING

Alongside standard campaign management, Flatiron implements an additional process to look for data patterns that indicate potential fraud may exist and that further analysis is required. This process must be utilized at the network level, publisher level, and sub-publisher level to ensure effectiveness.

  • Abnormally low click-to-install rates
  • This often indicates a publisher is engaging with “click stuffing” or “cookie bombing” a process where a mobile network tags the user as if the user had clicked on their ad when in reality the user converted due to another marketing channel

Impressions

Clicks

Installs

Click-to-Install

100,000,000

50,000,000

1

0.000002%

  • Randomly distributed time from click-to-install
  • With both “click stuffing” and “cookie bombing”, a pattern emerges with a randomly distributed time from click-to-install because the timing of the user clicking or being tagged and actual install are not related
  • In valid marketing activity, the pattern shows a relationship between the timing of a click and that of an install since they are correlated and predictable.

Time

Expected Time from Click-to-Install

Fraudulent Time from Click-to-Install

0 min. - 10 min.

40%

10%

10 min. - 20 min.

25%

10%

20 min. - 30 min.

15%

10%

30 min. - 300 min.

10%

10%

300 min. - 1440 min.

10%

60%

  • Consistently low amount of time from click-to-install
  • When mobile networks have “click injections”, there’s a consistently low amount of time between the click and the install
  • This is generally executed by having a widely distributed utility app that injects a click onto another app right before the user is able to open

Time

Expected Time from Click-to-Install

Fraudulent Time from Click-to-Install

0 min. - 1 min.

15%

100%

2 min. - 20 min.

50%

0%

20 min. - 40 min.

20%

0%

40 min. - 1440 min.

15%

0%

  • Abnormally high registration to first-time-deposit (FTD) rates
  • Exceedingly high registration to FTD rates can be an indicator of misaligned traffic or traffic to an audience that is disinterested in the product
  • Sometimes mobile networks will show users an offer for trying out a product which is often beneficial to all parties (e.g., if you deposit $10 then you’ll receive $5 of gems for a game in return)
  • The problem arises when the user has little genuine interest in the actual product because the audience targeting is ineffective in identifying the most interested users

State

Registration-to-FTD Rates (Single Platform)

Registration-to-FTD Rates (Cross Platform)

Expected

<75%

<125%

Requires Further Analysis of Deposit and Play Metrics

>75%

>125%

  • Abnormally low post-FTD play rates
  • Abnormally low play rates post-FTD can be another sign of misaligned traffic. This often occurs on affiliate deals, for example, save $50 on a $75 subscription for placing a $10 deposit
  • The logical action by a user is to take the deal to save money net of deposit for a service they wanted to purchase, either way, the issue arises if the user is only taking the offer for the discount and does not have a genuine interest in the product
  • The post-FTD play rates will be a clear indicator in determining how effectively targeted the offer is to the value the user brings to the business

State

FTD-to-Play Rates [7days] (Single Platform)

Expected

>90%

Requires Further Analysis of Deposit and Play Metrics

<90%

  • ADDITIONAL METRICS TO MONITOR
  • 30-day, 60-day, 90-day redeposit rates
  • Distribution of deposit sizes

Monitor metrics at the network, publisher, and sub-publisher levels.

3. INVESTIGATING ABNORMAL TRAFFIC SOURCES

Once you have discovered abnormal data, then you should:

  1. Build a case around traffic source by analyzing these metrics at the publisher and sub-publisher levels:
  • Click-to-Install rates
  • Registration-to-FTD rates
  • FTD-to-play rates
  • FTD-to-redeposit rates
  • Distribution of click-to-install rates
  • Distribution of FTD amounts

2. Reach out to the network and present the analysis

3. Request and review any additional materials the network offers to share

4. REFUTING FRAUDULENT TRAFFIC

Once you have discovered abnormal data, then you should:

  1. Determine if the materials sent by the network convince you of:
  2. No fraud
  3. Partial traffic fraudulent
  4. All traffic fraudulent
  5. If convinced that partial or complete fraud then determine the amount of damages to ask the networks to refund
  6. Remove all fraudulent sources from the mobile networks