Buying Installs: Quality Trumps Quantity
Over the years I have developed distribution networks and partnerships that account for well over 200 million application installations. I have learned some painful and expensive lessons along the way, so I hate seeing the mistakes I’ve made repeated by others.
So if I could only share two pieces of advice to anyone looking to distribute apps or manage performance campaigns it would be this: Measure the quality of the conversion for each campaign and optimize campaigns for quality before volume.
Sounds simple, doesn’t it? You would be surprised how hard it can be. I have worked with dozens of networks and thousands of publishers -- some of the brightest minds managing some of the biggest budgets in the industry -- and I know that very few have the skills or discipline to pull it off.
If you find yourself with a team of campaign managers managing scores of creative across multiple networks, you may see that just optimizing to the conversion is hard enough. Scoring the quality of the conversion beyond that may seem pointless if you have an average lifetime or book value established. In some cases the user value is a real challenge to calculate. So why bother?
Fraud is reason enough to measure quality. Any marketer offering a CPA is at risk. Those who aim primarily to maximize distribution volume are at the biggest risk. Quality scoring is probably the best way to reduce the fraud. Bad clicks and installs do not perform at the end of the day. Automated scoring will flag high-risk channels quickly.
Because bad installs and clicks underperform, reducing poor scoring campaigns will lift the overall quality of those that remain. By lifting the value of your conversions you can lift the CPA targets and increase your payouts to the best-performing channels. Optimizing your spend in this fashion will create a more virtuous cycle -- better traffic inventory for placement, better publishers, and ultimately more in-target conversions.
So how should you score quality? For applications this can be a challenge. Lifetime value is ideal, but for some applications calculating that figure will take too much time. The process of establishing accurate and effective scores is unique for every app. Simple factors such as opens, session length, page views, play levels, and other engagement benchmarks when combined with good analytics can predict lifetime value. Associating this score to the campaign ID is old hat for the Web, and it is no easy feat for mobile apps. Fortunately, an emerging group of service providers is taking up the challenge.
Another key factor to consider is that quality will change over time as traffic evolves. Today’s best-scoring publisher may be impacted by circumstances such as keyword use surges and growth. Tailoring your CPA targets to the score of each campaign in each channel will keep your spend and the overall quality of your distribution network safe over time. It will also make managing your partnerships much easier.
The temptation to maximize distribution to volume can be difficult to resist. Your distribution may start with great top-end returns, providing you with a sense of confidence. Beyond the risk of fraud, providing your distributors with volume incentives without a quality qualifier will ultimately backfire.
Publishers will negotiate higher CPAs based on volume. This provides them with an incentive to maximize reach. As they push for broader reach, your once-solid returns will change.
Without good systematic quality scoring, chances are high that you will get burned with a tanking ROI. If fraud is involved, you may find out too late and your brand may be damaged beyond repair.
The key is to think beyond the install or conversion. What does the user have to gain from the experience? What do you want from the user? What behavior are you targeting? Why are you spending all this money to capture their attention? Working backward from these basic objectives will help you build the right distribution partnerships. Take it from someone who has learned the hard way.