Appsflyer introduces new Protect 360 features: Our thoughts.
A few weeks ago, Appsflyer completely revamped their Protect 360 (P360) product and introduced some new interesting features to help advertisers in the fight against fraud. On top of a complete UI overhaul they also launched two new features: Post-Attribution Reporting and Anomaly Insights.
For the most part the UI is similar to the previous version of the product with the biggest difference being they no longer have the “advanced detection” tab. This has been completely merged with the main P360 screen. It’s apparent that many advertisers will welcome this update as confusing to continuously toggle back and forth while evaluating media .
In additional the user interface is now more intuitive to operate with interactive charts and graphs that are more user friendly to navigate than the previous version. Previously it was difficult to understand exactly what you were viewing when interacting with the visual representations of the data and now it seems to have improved along with the loading times.
Image source: https://hub.appsflyer.com “Learn about the new Protect360.pdf”
Post Attribution Reporting
By far the biggest update with the latest release is AppsFlyer’s new Post-Attribution Reporting. According to AppsFlyer 16% of fraudulent installs can’t be detected in real-time, rather they can be flagged after attribution has been made when sufficient data for that site/device id has been gathered. Appsflyer isn’t sharing exactly how this tool is classifying post-install attribution fraud for a few reasons. One, they are protecting their intellectual property when it comes to fraud detection and prevention. Two, given the state of the current eco-system if fraudsters possessed this information they would be able to exploit it. In general, however they are looking at all the main types of fraud (bot traffic, CTIT abnormalities, Click Flooding, etc).
This is going to be a great feature for all advertisers and the industry moving forward as it will put more pressure on networks to clean up their traffic as they will see larger amounts of installs scrubbed from billing. Previously it may have been a game of finger pointing in excel sheets and now it’s all centralized from the MMP.
There are however a few drawbacks with the recent update.
First of all, as these post-attribution flags come after attribution the main Appsflyer dashboard continues to report the installs as they have originally been attributed. It would be logical to at least be able to layer both data sets directly in the dashboard as opposed to exporting them to some sort of BI software to complete the task.
Secondly there are no real-time notifications for media partners that a post-install attribution has been flagged for fraud. As this is the first release of this feature, it’s understandable that it isn’t included but it should be considered for future releases. It would be highly beneficial for media partners to get notified for these types of events so they can reconcile on their side quickly and optimize accordingly rather than relying on manual data dumps from the AF platform.
Included in this new tab are additional visual insights so you can compare media sources against a chosen baseline for click time to install (CTIT) abnormalities. Although the updated visuals on this tab are easier to navigate than previous versions for optimal media operations , there could be some improvements.
For starters the biggest drawback on this feature is the inability to look at site level data. At the moment you only have the ability to look at aggregate data across each media source you are running on. So right off the bat you can’t see the entire picture for each source.
It is convenient when it automatically tells you when there is an “anomaly,” however it doesn’t point out if this anomaly is a negative or positive indicator. For example, depending on what media source you pick as your baseline, P360 will report anything that deviates from this CTIT distribution as an anomaly. Unless you’re an experienced user you might assume the anomaly has negative connotations when in reality it just means “different from the baseline” which could mean a lot of things.
This feature could use some additional attention from the Appsflyer team. It would be nice to be able to take more actionable insights from this view.
Overall, the principal updates by AppsFlyer are a big step forward with fraud protection. It is very encouraging that they are starting to use their global attribution data to flag fraud post-attribution
Questions about the latest Appsflyer update? Mobile attribution in general? Feel free to reach out for a chat.
Comments are closed.