4 common attribution mistakes (and how to avoid them)

Attribution allows digital marketers to learn about the user journey and what brings users to their app. But despite the advanced capabilities of mobile attribution tools available to marketers, it’s still important to try and avoid the headaches that come from data discrepancies. Here we’ve outlined some of the most common attribution pitfalls marketers need to avoid.

1. Avoid paying for the same install more than once

When more than one media source claims an install, marketers can be charged multiple times. The last thing marketers need is additional costs for an acquisition, so being charged twice needs to be avoided wherever possible.

Double attribution occurs when there is not a unified payment system. Media partners’ SDKs are added, measurement begins and these companies claim installs based on clicks. But because they have no data on the activity of other media partners (and therefore can’t see the entire user journey) a network can claim an install for which they only partially contributed. There could be another click (from another partner) before the install, but they wouldn’t know it. A great way to avoid these discrepancies is to work with a mobile measurement partner (MMP) that provides last-click attribution, meaning only the network with the last click receives payment. With a clear-cut payment method, marketers can be sure they won’t shell out more money for an install than they first anticipated.

An additional benefit is that MMPs are a third-party, making them unbiased in their reportage. Unlike self-attributing partners (who will gain monetarily from claiming installs), a third-party measurement partner’s business relies on accuracy and transparency.

2. Don’t compare downloads with installs

Another common error marketers make is to compare downloads with installs. If you’re looking at two different reports and treat downloads and installs as the same, you’ll notice that the numbers don’t add up.

Downloads and installs have fundamentally different meanings: a download is registered when a user clicks to download, while an install requires the user to open the app. In short, this discrepancy occurs when some users have downloaded the app, but haven’t opened it.

This can leave marketers scratching their heads, but the reason for this distinction is fairly simple. If an MMP doesn’t have access to Google or Apple’s download data, they will keep their reports independent by only reporting installs. During your own analysis, confusion can be easily avoided by only comparing installs with installs.

3. Measure Facebook the right way

There are two ways to measure your campaigns on Facebook. Firstly, you can embed Facebook’s SDK, or you can use one of Facebook’s official MMPs. If you are using another method to measure the performance of your Facebook campaigns, you’re likely missing out on vital information that could otherwise be gathered through Facebook’s SDK (or with a third-party MMP). For example, if you’re using deep links to see where the user came from, you still won’t have a timestamp for that install. This can be avoided by setting up Facebook measurement via either method mentioned above.

4. Don’t let ad fraud waste your resources

Ad fraud is a widespread problem facing the mobile ecosystem recent reports show that it’s still rising. Juniper Research have estimated an industry-wide $42 billion loss of ad spend in 2019, a 21% increase from the $35 billion lost in 2018. Besides taking money out of your pockets, ad fraud also comprises data and ruins your analytics.

Whether its click injection, click spam or fake users (bots), every app vertical is susceptible to fraud – particularly those with high Cost per Install (CPI) campaigns. Look out for the following scenarios to ensure you are vigilant against these types of attacks:

Fake users and jailbroken devices
If your MMP doesn’t have the same insight into Apple and Google’s data on jailbroken devices and fake accounts, they won’t be filtering out the same activity. That being said, your MMP will have their own fraud prevention methods available (usually as a separate package) to give greater protection against fraudulent claims.

Incentivized campaigns: Resetting device IDs
If you’re running campaigns (particularly on Android) with incentivized traffic sources, fraudsters will try to avoid paying for in-app purchases. Their aim is to receive multiple incentives mainly by refreshing their advertising IDs, then repeatedly uninstalling and reinstalling the same app. Every time a new advertising ID is produced, they may be counted as individual installs by an MMP (but not by Google), creating a discrepancy issue.

Fraudulent in-app purchases
If you’re tracking in-app purchases without purchase validation, your recorded revenue may not correlate with App Store/Play Store revenue. This is due to fraudsters hacking apps that allow them to make free in-app purchases. For gaming, this could be a fraudulent (fake) transaction for in-app currency. To avoid this, talk with your MMP about implementing a purchase verification system.

With these potential issues in mind, you can save valuable time and energy that would otherwise be wasted trying to find out which measurement is accurate. Ensuring that data is integrated (and knowing the measurement capabilities of each partner) will result in a swift resolution – so you can focus on what matters.

4 common attribution mistakes (and how to avoid them)
Illustration: Freepik

Read next on DIW: Recent Marketing Trends Are Changing Consumers Expectations from Brands (infographic)

No comments:

Post a Comment