5 Pitfalls of Mobile App Marketing & How to Avoid Them
Kurt Geater is currently a Mobile Marketing Manager at Groupon, the 800lb shopping app gorilla, where his focus is on new user acquisition for the app. His first marketing role was at the Austin American-Statesman newspaper in Austin, TX, working as a Targeted Acquisition and Retention Analyst where he ran direct mail marketing and aided in their print-to-mobile migration.
Learn more from his Mobile Hero profile.
As many mobile app marketers can attest, one of the greatest things about working in the mobile app space is that it affords one the opportunity to work in perhaps the most dynamic, constantly evolving fields in marketing. Mobile apps are an integral part of everyday life and, as a result, businesses are investing millions of dollars in order to market their mobile apps to a global audience.
Although exciting, one downside to this ever-changing industry is there are always new pitfalls to avoid. Many times, marketers are not aware of these pitfalls until they have spent some (or all) of their budget inefficiently. Below are a few things I have learned over the last few years working in the mobile space, and my hope is that at least one of these tips will strike a chord with even the most-seasoned app marketer.
Optimize for the Right Event
One of the questions I always ask myself is, “what good is a download that does not end up converting?” Sure, net downloads may help you on the ASO side of things, but when it comes to running campaigns with the goal of making a return on your investment, do you really want total installs to be your goal? That said, one of the smartest things an app marketer can do is to start optimizing for installs that drive a purchase (or whichever lower-funnel event you are trying to achieve).
As large-scale networks continue to aggregate data on their users, they are getting smarter about how they bucket, segment, and charge for their audiences. They are able to differentiate between users that purely love to download apps vs those who are more likely to download AND make a purchase. With that in mind, if you are still optimizing towards net installs, don’t be shocked if the downstream value of your downloads does not hit your targets or starts to degrade in the long-run.
It may be useful to take a step back, re-evaluate your core optimization events, and it may even be worth paying a little more for high quality users if they continue to convert better in the long-run. At the end of the day, I would rather pay $20 for an install that results in a purchase rather than paying $2 for ten installs that never end up converting.
Know Your MTTI
One of the most under-valued metrics that is not talked about enough is the MTTI, or the median time-to-install of an app download. The MTTI is defined as the median time it takes for the user to download an app once they have clicked an ad. The shorter the MTTI, the easier it is to make the argument that it was the ad which drove the download. This is best illustrated in the table below:
|Channel||1st Quartile||Median||3rd Quartile|
|Paid Network A||1||2||6|
|Paid Network B||1||2||10|
|Paid Network C||1||60||80|
|Paid Network D||1||2||4,500|
Here we can see that the MTTI for Networks A, B, and D are ~2 minutes which means a typical user will download an app approximately two minutes after viewing an ad. In this case, we can safely assume (barring any click-spamming fraud and using ‘owned’ media as a proxy) that it was the ad that incentivized the user to download.
In contrast, take a look at network C. This MTTI is 60 minutes, which means it took the user an hour to download an app after viewing an ad from this network. Taking this into account, it is much harder to make the argument that it was Network C which incentivized the app download, and there may have been some other touch-points along the way which incentivized the user to install.
An additional metric to point out here, is the 3rd quartile MTTI for Network D. In this case, Network D is getting attributed downloads which occurred ~4,500 minutes (roughly 75 hours) after the click. In cases like this, it is much harder to make the argument that it was the ad which drove the install so you may want to consider making your look-back windows more stringent in order to avoid organic cannibalization or misattribution of your downloads.
Diversify Your Portfolio
Diversify your portfolio of channels for both UA and re-engagement. Every channel has a point of diminishing returns, where for every additional dollar spent, the amount of value you can derive from the channel will decrease due to the pool of users getting smaller and smaller.
Almost every app marketer has encountered a scenario where they spend the first $1,000 on a new network and everything looks amazing in terms of performance, but as soon as they try and scale up, things start to look inefficient. Taking this into account, the more channels you are able to get working efficiently, the better. This will not only help with potential reach of your app, but it will also provide you with other areas to shift your budget if a channel gets hyper-competitive during a given season.
Increasing the number of channels you use will also help prevent ad fatigue for potential users and allow you to keep your user ad frequency at a healthy level. Always keep testing. You never know what may end up being the next big thing.
Avoid Fingerprinting at All Costs
One of the easiest ways to burn through a budget and end up paying for downloads you could otherwise be getting for free is to use fingerprinting for your mobile app attribution. Fingerprinting is the attribution methodology which attributes downloads based on a number of identifiers, namely IP address, phone make/model, and phone OS version. Almost all MMPs support this, and some even have it set as a default.
In order to illustrate the issue with fingerprinting, let’s say we have two individuals: User A and User B. Both are at a Seattle Seahawks game, both are using the same cell-phone tower, and are therefore using the same IP address. User A has an iPhone 7 with iOS 12.4. User B also has an iPhone 7 with the same OS. User A is a high-intent user. He enjoys making app purchases and has very high downstream purchase frequency. User B, in contrast, loves downloading apps for fun, but she rarely makes in-app purchases and usually uninstalls an app within a day or two.
Let us say User A downloads an app organically without clicking an ad and User B saw an ad for the same app on a network, clicked, and decided to download. If you are using fingerprinting, there is a chance that this high-intent, organic user who never clicked an ad will be attributed to the network because they look identical in terms of IP address, phone model, and OS version.
As an app marketer who is using finger-printing, this behavior can result in your business hemorrhaging money. Just imagine the scenario where thousands of high-intent, high-value users are being misattributed to a network for which they never even clicked an ad. As you are evaluating the network, you observe that the downstream value of the downloads looks amazing (of course it does, it has been inflated by the cannibalization of organic users), and as a result, you consider pumping more money into that network, increasing the negative effects of this misattribution.
In order to avoid this scenario, disable fingerprinting whenever possible, and always use IDFA and AD ID attribution. Some MMPs use 7-day finger-printing as a default so either turn this off or shorten the fingerprinting look-back window to less than 1 hour to improve the overall accuracy.
Not All Downloads Are Created Equal
Whenever a vendor or network asks me, “what is your target CPI?” my response is always, “well, that depends on the quality of users you can drive”. One of the biggest mistakes an app marketer can make is to set the same CPI target across all networks. The best way to set your CPI targets is to look at the unique downstream value that each individual network is driving and cater your CPI targets to that value.
If a business sets a fixed CPI target of $5 across all networks, there is a chance that they are overpaying on certain networks while simultaneously underpaying and missing out on potential scale for other networks. Your CPI targets should be ever-changing and adjusted seasonally based on the downstream value of each of your network cohorts.