Finding the most viable, cost-efficient way to acquire loyal users is the holy grail for mobile publishers and advertisers. The question of whether campaigns should be run on a cost-per-install (CPI), cost-per-click (CPC), or cost-per-mille (CPM) basis is a source of confusion in the press and industry as a whole.
As a mobile games marketing platform with the accrued experience of running thousands of performance marketing campaigns for our clients, we at AppLift have our own opinions on the issue: CPI is best viewed as a tradeoff rather than an optimal solution, and is most effective when combined with LTV (user lifetime value) optimization.
A happy medium between volume and risk
A recent VB Insight study, ‘Mobile User Acquisition: How the most successful developers get better users for less money,’ found that CPI, though the most-favored user acquisition method among mobile advertisers, is actually the least successful. Chasing the lowest price installs may just wind you up with low value users.
John Koetsier, the VentureBeat journalist behind the study, wrote in a recent email to me “CPI results in a flood of cheap, low-value users, while other methods (CPL (cost-per-lead), CPP (cost-per-purchase, a specific case of CPA), for example) result in a much higher percentage of high-value users. The overall number of users is lower, but the percentage of users who monetize well is much, much higher.”
CPI campaigns though, whether optimized for LTV or not, can still make more sense than other types of campaign model.
In this article, we will go on to compare CPI with other types of campaign and then talk about how they can be optimized further.
The mobile advertising models spectrum
Looking at the various available advertising models, we view CPI as sitting in the middle of a spectrum based on risk to the advertiser and traffic volume. CPM sits on one side; the traffic volume being high but the financial risk laying solely with the advertiser, since the publisher is paid for merely displaying an ad.
Moving to the right, CPC removes a degree of traffic volume, as publishers have to work a little harder to achieve a user click and secure a payment for themselves, but also simultaneously lifts some risk from the advertiser.
Compare this to CPA (or CPL, or CPP), which, in dropping the full weight of the risk on the publisher (who only receives payment when a defined event takes place), makes them reluctant to adopt this model when such a significant risk befalls them and hence results in much lower traffic volumes. This might make a campaign’s ROI look good to begin with, but CPI campaigns optimized for LTV bring about increased virality and organic installs, as well as longer-term traffic quality improvement.
We view CPI as a “happy medium” of risk and traffic volume; it occupies the middle ground for both.
We’re studying mobile app analytics solutions.
Share your knowledge, and we’ll share all the data with you in return.
It’s not always an optimal solution by itself (pre-LTV optimization), but it’s a balanced tradeoff and works well in many instances. You also need to remember that publishers have ad inventory which they provide to different advertisers, so advertisers and pricing models compete both against and between each other. For instance, high-quality inventory from publishers requires high CPIs to guarantee that these well-performing publishers keep promoting the ad.
Mitigating the trade-off
In order to strike the balance between traffic and risk, it’s necessary to optimize campaigns towards high LTV users.
LTV optimization means investing ad spend where it has proven to yield the users that perform defined in-app events, which we call proxies. These proxies include user actions and behaviors like level completion, usage patterns, social shares, in-app purchases and other monetization and retention indicators.
Storm8’s head of user acquisition and monetization James Peng confirmed to us that CPI combined with LTV optimization works best for them:
“For me, CPI media buying is crucial because it limits financial exposure per unit, and it allows me to focus on optimizing user value through each channel instead of managing my CPC or CPM towards a goal. By dynamically optimizing campaigns for LTV, I can exclude publishers and media niches that aren’t delivering value and ultimately drive a consistent return.”
Our experience includes running LTV-optimized campaigns for hundreds of games across a myriad of genres and publisher sizes. One example where our continuous optimization of traffic sources has been particularly effective is for Aeria Games’ Immortalis campaign, we achieved a 150 percent ROI after recouping the initial investment. The organic downloads garnered as a result of the game’s elevated position in the app store charts have added extra unattributed income, amounting to additional profit.
In a campaign for a different client, our focus on relevant retention and monetization metrics for a simulation game saw 0.4 percent of generated users go on to become whales, i.e. users who spent over 100 USD in a three-month period.
Data we have collected from a one-week cohort of our top five highest-performing media partners during yet another of our campaigns for an action game shows that these traffic sources recouped the initial user acquisition investment just four weeks afterwards.
CPI campaigns, then, are best seen as a compromise between traffic volume and risk to the advertiser, both of which are middling for this particular model. Without LTV optimization, incentivized CPI campaigns can effectively serve as a virality tool for certain apps; a way to get new users onboard fast. Combined with LTV optimization, we can benefit from the high traffic volume but also black- and whitelist the relevant traffic sources and invest ad spend only where it brings in the most valuable users.
AppLift is a mobile games marketing platform. The company was founded in August 2012 by Kaya Taner, Tim Koschella and Hitfox Group. AppLift’s platform helps mobile game advertisers acquire loyal and paying gamers at scale on a perfor... read more »
Powered by VBProfiles
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.