App marketers must show the effectiveness of their ad spend on long term revenue, often through performance metrics like ROI and ROAS. While this kind of measurement is vital, it is not always easy since performance can be skewed by improper attribution, fraud, and the use of incentivized traffic (sometimes inadvertently).
One of the easiest ways to avoid these problems is by using a DSP that focuses on revenue-based metrics. With a performance metric in place, campaigns get optimized towards sources that drive quality users: if a particular publisher or audience doesn’t deliver good ROAS, it is quickly removed from the buying mix. Fraud and other issues associated with low-quality media are automatically avoided as well because those channels don’t perform over time.
When platforms are not optimized for ROI or ROAS, it’s evident. The chart below compares click to install data between four prominent mobile marketing channels.
|Installs / Clicks||Aug-16||Sep-16||Oct-16|
Such a large discrepancy in performance can be an indication that sources B and D are using a blend of both high-performing and cheap, low-performing incentivized media: click to install rates of 25% to 40% are difficult to believe. Over time, the lifetime value (LTV) of a user base made from this blended traffic will be less.
Running audience-targeting campaigns using mobile user IDs is another way to avoid paying for inflated performance. DSPs and other data providers can help advertisers use detailed user data to target specific people likely to become big spenders.
The key is app usage data that’s tied to mobile device IDs. App usage data has proven to be the best predictor of future app behaviors — more so than demographic or publisher data. And using IDs known to show valuable behaviors avoids fraud.
There are two key ways to use mobile IDs for audience targeting:
- Some providers offer pre-built audiences or personas based on observed app activity and organized into groups such as known casino users or RPG fans. Advertisers can syndicate those audiences to their ad networks and target those specific individuals, greatly reducing the opportunity for fraud.
- Lookalike modeling, where a provider takes a seed audience of known high-quality users, identifies the most significant shared characteristics, then finds more users who have the same characteristics, also provides a list of specific individuals to target. While most familiar on Facebook, lookalike modeling is particularly effective on Open RTB as well, reducing fraud by focusing on specific mobile IDs that are likely to be high-value users.
The fast growing and lucrative casino app market will always be a target for many kinds of fraud and deceptive practices. Fortunately advertisers can use a combination of data and technology to make sure they’re limiting their exposure and maximizing their ROI.
If you have questions about measuring performance or just want to review your campaigns with us, please reach out.