Effective ways to fight mobile ad fraud
As for now, every app marketer is able to choose between 3 different options: in-house fraud detection, tracker-analytics fraud detection or third-party anti-fraud solutions. Below are pros and cons to every mentioned option, so you can compare and choose the most appropriate tool for your specific needs.
1. In-House Fraud Detection.
In-house fraud prevention usually means the anti-fraud team, consisting of technology professionals, such as fraud investigators, rules analysts, chargeback analysts, and fraud supervisors. But in practice, it’s rather a UA manager(s), responsible for the purchase of traffic and further analyzing of traffic received.
1. A customized solution — company’s business logic will be embedded, providing an exact fit to company’s needs.
2. This option allows to incorporate data from various sources.
3. The high level of data visualization.
1. The high cost of development and maintenance of your service.
2. The small volume of analyzed data (only from company’s app(s)) and as a result — the level of fraud detection not above the average.
3. Because of using custom rules, the level of accuracy of fraud detection is not above the average. Custom rules can not detect such sophisticated types of fraud as “smart bots” or “mixes”.
Summary: No one knows the business better than its own employees, so in-house team might show good results in tuning business operations for the specific needs of the company. But at the same time, in-house teams usually able to recognize fraud and chargeback patterns on the primitive level only. A deep analysis, consisting of reviewing thousands of combinations of publishers, sub-publishers, sub-sub-publishers and so on takes a lot of time and efforts for them. Cost-effectiveness of such teams is also a matter of discussions.
2. Tracker-Analytics Fraud Detection.
The most of well-known tracker-analytics systems began to offer the built-in anti-fraud function since the last year. But before purchasing such solution, it is worthwhile to figure out what pros and cons are in it.
1. Anti-fraud function is built-in in a tracker and does not require additional implementation activities.
2. Some trackers offer an automatic rejects function, which is very convenient for rejection of primitive fraud types.
3. Such solution has access to a large amount of data as a result of tracker’s activities.
1. A conflict of interest, since the main monetization model of such solutions — payment for tracking paid installs. In the case of rejects in connection with fraud, they risking to reduce their incomes for attribution.
2. Low personalization of data — during analysis system relies on the same algorithm and same metrics for all apps.
3. Trackers detect just a few types of fraud (due to automatic rejects they do not analyze the post-install metrics).
Summary: Perhaps, a built-in solution from a tracker-analytics system is the most convenient and easy-to-go way to start your own fighting with mobile app-install fraud. But such solutions, in their majority, are not able to give a univocal decision whether we deal with fraud or not. It only indicates ‘a suspicious traffic’. The responsibility and the final decision are laid on UA manager, as in the case of an in-house anti-fraud team. Also, the accuracy and completeness of fraud detection on the average level due to simplified analysis.
3. Third-Party Anti-fraud Solutions.
This is an independent solution, invented specifically for fighting against fraud and all company resources are aimed at improving and training algorithms for detecting different types of fraud. Let’s take a look at the main pros and cons of third-party anti-fraud solutions.
1. The highest level of accuracy and completeness of fraud detection (if we talk about solutions that using true ML)
2. Great data personalization due to data accumulation from the wide amount of different apps and system learning over time.
3. Best-in-class third-party solutions are based on deep ML algorithms to analyze data considering every detail in order to detect all types of mobile ad fraud.
1. Sometimes, third-party solutions can cost more than solutions from tracker-analytics systems.
2. Data consolidation and decision-making (fraud/not a fraud) can take up to 1–2 days, if they work with post-install events analysis. On other hands, that helps to increase accuracy dramatically.
3. These solutions with their initial resemblance may highly vary in their characteristics, so here you may need more time for making your choice.
Summary: With variety solutions available on the market in this category, it’s easy to get lost. But the main feature of the worthy solution is implementing of true Machine Learning algorithms and analyzing post-install metrics (retention, engagement, custom events, financial events) during fraud identifying. At the moment, Third-party Anti-fraud Solution showing the best result in fraud-fighting, taking into account the fact that it is precisely capable of detecting fraudsters with the highest accuracy and completeness.