1. Monthly traffic optimization options: Appsflyer + Protect360
Real-time and post-attribution fraud
The fraud activity is monitored throughout the placement period using Appsflyer tools, including Protect 360. Evaluation, optimization, and disabling is done by Side ID and Publisher ID.
Parameters* for optimization through Protect360:- Real-time fraud up to 20%, post-attribution fraud up to 15%
- New devices. The number of new devices in the AF should not be > 10%
- LAT Devices % - not less than 70% iOS
- Assists installs % - not more than 80%
Parameters* for manual optimization:- CTIT > 60 minutes not more than 30%
- CTIT > 5 hours not more than 20%
- CTIT < 10 seconds not more than 10%
- Is Primary attribution UA - False not more than 5%
- Is Primary attribution Retargeting - True not more than 5%
- Retention Rate less than organic up to 15% and not more than on 10%
- Partners from the blacklist below are strictly prohibited
- Tracking links without erid/token won’t be paid
- Uninstall rate not more than 10%
- Devices without Advertising ID – not more than 60%
Additional options of optimizations (once a month)- CR from click to install not less than 0.1%
- Correlation analysis of organic installs with clicks/installs by in-app sources
- Device model anomalies
- Blacklisted IP presence
- Versions of installed apps (for new users)
- Monitoring motivated third-party apps for the availability of an offer and attribution links
- Revenue anomalies analyses
In case of exceeding the totality of these indicators, in agreement with the client, the partner may be refused an offer at the end of the month.
- Monthly traffic optimization parameters: Appmetrica
Monitoring compliance with the fraud activity during the entire placement period is carried out using reports from Appmetrica. Verification takes place at the publisher ID level.Parameters for downloading a report on installations:- publisher_name, publisher_id, click_id, click_datetime, click_ipv6, install_datetime, install_receive_datetime, install_ipv6, is_reinstallation, ios_ifa, google_aid, device_manufacturer, device_model, connection_type
Parameters* for manual optimization:- CTIT > 60 minutes not more than 30%
- CTIT > 5 hours not more than 20%
- CTIT < 10 seconds not more than 10%
- Devices without GAID/IDFA - not more than 60%
- Match IP click and install
- CR from click to install not less than 0.1%, not more than 5%
- It is forbidden to send impressions to a click link. It leads to a partial loss of information in Appmetrica due to the technical difficulties of the tracker, consequently, some conversions may not be displayed by the system, which may lead to a decrease in CR and failure to achieve Hard KPI.
- Monthly traffic optimization parameters: Appmetrica + Fraudscore
Parameters for optimizing traffic during the month:The fraud activity is monitored throughout the placement period using reports from Appmetrica and analytical data from Fraudscore. Checking takes place at the level of Appmetrica_device_id, id_2 in Fraudscore.
Parameters for manual optimization:- 100% of installations that fall into the High category are removed
- 100% of installations and actions that were categorized as LowTTI, FlatTTI, Hi-jacked device, clickspam, Blacklist, Fake device, Bad Local are removed
- The number of installs and actions that hit the "OK" green zone must not be below 10% of the total number of installs.
- The number of installs and actions that fall into the blue "Low" zone must not be more than 30% of the total number of installs.
- The number of installs and actions that fall into the orange "Mid" zone must not be more than 20% of the total number of installs.
- The number of installs and actions that fall into the red zone "High" must not be higher than 20% of the total number of installs. This is the main criteria, if it is not fulfilled, the partner can be completely disconnected.
It is forbidden to send views to a click link. It leads to partial loss of information in Appmetrica due to technical difficulties of the tracker, i.e. some conversions may not be displayed by the system, which may lead to lower CR and failure to achieve Hard KPI.
1. Monthly traffic optimization parameters: Adjust + Fraud Prevention Suite
Parameters for optimizing traffic during the month:
The fraud activity is monitored throughout the placement period using Adjust reports, including Adjust antifraud system. The checking takes place at the racker level.
- Parameters* for optimization through antifraud system:
- Rejected install rate not more than 15%
- RI TME (Too Many Engagements) not more than 10%
- RI CI (Click Injection) not more than 5%
- RI DO (Distribution Outlier) not more than 10%
- RI AIP (Anonymous IP) not more than 20% on the territory of Russian Federation and not more than 5% in other countries
- Parameters for manual optimization:
- Uninstall rate not more than 25%
- CTIT > 60 minutes not more than 30%
- CTIT > 5 hours not more than 20%
- CTIT < 10 seconds not more than 10%
- LAT Devices not less than 70% (only for iOS)
- Click to install CR not less than 0,1% and not more than 5% (CR less than 0,01% is automatically blocked by tracker)
- Assist installs – not more than 20% crossings with other sources, not more than 80% of crossings with own campaigns
- Amount of events with 0 revenue – not more than 5% (no adgroup)
- Retention rate less than organic not more than on 15%, not more than on 10%
- Deviations be App store type – not more than 5%
- Deviations be OS name – not more than 15%
- The Country parameter is not RU – not more than 10% of the partner’s total traffic
- Correlation analysis of organic installs with clicks / installs from in-app sources – once a week
- Presence of empty order IDs in the CRM upload is not more than 5% per campaign
- Partners from the blacklist below are strictly prohibited:
- Tracking links without erid/token won’t be paid
- Additional options of optimizations (once a month)
- Device model anomalies
- Presence of blacklisted IP
- Versions of installed apps (for new users)
- Collation of installs amount in Adjust by OS with Google Developer and App Store Developer consoles
- Monitoring motivated third-party apps for the availability of an offer and attribution links
- Revenue anomalies analyses
- If the aggregate exceeds more than 3 indicators, in agreement with the client, the partner may be refused an offer at the end of the month.
- Monthly traffic optimization parameters: MyTracker
Compliance with the fraud threshold throughout the placement period is monitored using reports from Fraud Scanner, including the Fraud Scanner anti-fraud system. Checks are conducted at the SUBID1-5 and SUBNETWORK (site_id) levels.
Parameters for optimization by the anti-fraud system:- Low CR no more than 30%
- Strict fraud metric no more than 30%
- Matching clicks no more than 30%
- Suspicious devices (%) no more than 20%
- Malicious IP addresses (%) no more than 20%
- Bots and device farms (%) no more than 10%
- CTIT (long installations) > 5 hours no more than 20%
- CTIT (short installations) < 10 seconds no more than 10%
- Rigorous fraud metric (matching clicks, short installs, bots and device farms) no more than 10%
- Confident fraud metric (matching clicks, short installs, bots and device farms, anomalous behavior, attribution fraud) no more than 20%
- Confident fraud metric (matched clicks, short installs, bots and device farms, anomalous behavior, attribution fraud, low CR, Root/Jailbreak, long installs) not more than 50%
Parameters* for manual optimization:- CTIT (long installs) > 1 hour max 20%
- CTIT (short installs) < 10 seconds not more than 10%
- CR from click to install not lower than 0.1%, not more than 5%
* all percentages are taken according to internal benchmarks, based on the expertise of MGCom- Traffic optimization settings for all MMPs
If the sum of the parameters written above is exceeded, the partner may be paused on an offer. Exceeding the recommended values once on one parameter is not a reason for disabling, but may result in a warning.- Additional optimization options
Please note that your traffic can be additionally checked in the following way on all MMPs:- Versions of installed applications (for new users) correspond to the current ones in the store
- Correlation analysis of organic installs with clicks/installs from in-app sources
- Anomalies in Device Model and Device Manufacturer
- Traffic coming from Blacklisted IP (IPs are collected from Protect360 databases)
- Reconciliation of the number of installations in MMP by OS with the Google Developer and App Store Developer consoles;
- Monitoring third-party motivated applications for the presence of an offer and attribution links.
- Tests of new partners/network
- 14 calendar days are given for a network test and optimisation.
- 1 month is given for the test of a new partner. Within 2 weeks, it is allowed to exceed the fraud metrics specified in the regulation, then all points of the regulation apply to the partner.
- Appsflyer: For the period of testing a new partner, we evaluate the CTIT and Assists indicators for the first 7 days (see paragraph 1), if the indicators for these parameters are higher than acceptable, then the partner is banned from an offer
- Appmetrica: For the period of testing a new partner, we evaluate the CTIT and CR indicators for the first 7 days (see point 2), if the indicators for these parameters are higher than acceptable, then the partner is banned from an offer
- Adjust: For the period of testing a new partner, we evaluate the RI TME and RI CI indicators for the first 7 days (see paragraph 3), if the indicators for these parameters are higher than acceptable, then the partner is banned from an offer.
The test is considered passed successfully if after 2 weeks of the test period the fraud indicators do not exceed the values described in the regulations.Payment to the partner during the test period is made only for traffic that meets the requirements of the regulation.- Interaction with partners when fraudulent traffic is detected
- The partner is issued 3 warnings in a period of 2 months on the quality of traffic with a request to optimize a Publisher.
- In case of the fourth warning, the partner is disconnected from the offer.
- In case of disputable situations regarding targeted actions, information from the advertiser's CRM can be used to evaluate the quality.
- If the advertiser identifies the fact of fraudulent traffic, all traffic may not be paid for.