In my last post I shared my own and proven approach to help you calculate LTV. Now it’s time to take your model out “into the wild” and subject it to some key “reality tests” before you apply it to your business at large.

It all starts with recognizing and establishing the characteristics and commonalities among the countries (and audiences) where you have chosen to focus your efforts. Knowing your LTV globally is a certain number (say, $0.25) is data that is neither helpful nor insightful if UA costs in your target market (say, the U.S.) are $1.90 on iOS.

In practice you need to know UA costs on a country and operating system level.

In plain text this is the input, drawn from a statistically meaningful sample of users that are similar in key characteristics to what you have determined is the perfect fit with your app, that will allow to construct a meaningful model that applies to your app and app business. This is also a step where you want data, but you also have to accept it’s where you sometimes have to improvise.

In some markets it may be too tough — or too expensive — to establish a statistically meaningful sample of cohorts. It’s also a trade-off since waiting to gain higher confidence in your sample may costs you important momentum in a critical stage of your app. But changing your frame of reference just may allow you to close the gap and get on with your business.

Let’s imagine you want to create statistically meaningful cohorts in order to understand the LTV of Android users in Germany, and let’s imagine that you don’t have the cohort sample that gives you the level of confidence you need and want as you apply your LTV model.

In this case, you could look at other markets in Europe that are similar to Germany. The idea here is to be flexible and be aware that you are creating a model. Making adjustments and improvisations along the way are par for the course and part of the process.


Image 1: Be smart about how you increase your sample size and be aware of the imprecisions


Be sure that you create a model in which you can be quite confident in based on numbers that are fairly accurate. To this end you need critical mass. My point is: if you don’t have that critical mass in a given country on a specific operating system, then you have to cast your net a bit wider.

Just keep in mind that you’re looking for commonalities. Use your common sense and only improvise if it makes sense for your app, your audience and what you are seeking to understand. Obviously there are limits. Granted, North America and Europe are quite similar when it comes to retention. But if your app is about soccer, an interest that unites Europe but leaves North America the “odd man out,” then it’s it might be best to leave that region out of equation altogether.



Once you have framed your model based on what you have determined are statistically meaningful cohorts for your app, then you have to ask and face some tough questions about how much money you want to make (recoup) by when.

Knowing this is the only way to realistically know when your UA efforts and spend are on the mark, or missing your targets completely. The timeframe depends on your app category, app business and app goals.

In my capacity as a consultant I have worked with clients who wanted to recoup their money within three years, and I have also worked with clients where I would consider 90 days an eternity. Whatever you decide, make sure your timeline is realistic given the kind of app you offer. In other words, if you have an app packed with all the bells and whistles that encourage engaging and addictive gameplay, then it may make sense to think in 6-month or even 3-year timeframes. However, if you have a one-trick pony app, where the novelty is sure to wear quite thin quite fast, then even a 90-day timeframe can be stretching your luck.

As I noted in my earlier post, the longer you predict the future, the greater the risk. Too many things can happen. Tastes can change, technology can change and  markets can change. More variables means more uncertainty. A good rule of thumb in my opinion is to aim to recoup your spend in 90 days.



Clearly, your model rests on the realistic timeframe you have chosen. Once you have this important piece of the puzzle you can map out what you want to recoup by when. But don’t assume it’s written in stone. In other words, be prepared to adjust this aspect of your model depending on what your audience is telling you.

Imagine you have an app that is high in retention, but not hitting the benchmarks you set for LTV. Don’t read it as an immediate signal to pull the plug on your UA spending because it may just be that your highly engaged users are just a tad slow in spending in your app at equally high levels.

It’s always going to be a tough one to call.


Image 2: Example of tolerance window for ROI after x days


My personal rules are simple: Define a tolerance window and define the minimum return you want to achieve by a certain day and stick with it. If my retention is on par or better and it’s just my LTV that isn’t yet at the level I want, then I will give it more time. Why? Because if users love the game, then spend can still follow. In any case, it’s worth waiting a little longer to see how it will play out.

But I can be decidedly more brutal if my retention metrics are playing out the other way around. Even if numbers are within my tolerance window, if my retention is not where I need it to be, then I will likely turn off a channel than risk wasting more spend.



UA spend, retention, revenues, in-app purchases and purchases that are not in-app – incorporating this into your LTV model is really important, and applying it to the real-world is a real challenge.

There are various tools on the market that will allow you to monitor UA spend, and there are others that excel at attributing installs or giving me a accurate snapshot of my app revenue data. But multiple dashboards can be a hassle, and can only give me part of picture.

My personal pick to get the 360° view into what I need to test and adjust my LTV model is Tenjin, a company that prides itself on offering app companies a modular full stack that covers it all (attribution, aggregation, analytics and a data warehouse) giving me a single view into how well my app is really doing — without a lot of Excel calculations.


Image 3: Monitoring ROI on campaign level


More importantly, I have the data that allows me to focus on applying my model, not reworking it. I have my percentages I want to hit and I have the data that allows me to do my daily job. It’s a plus; especially for app companies with small (or no) UA teams to task with the never-ending task of optimising UA spend.


But it’s not just about saving time and man-hours. It’s about understanding and accepting the hardest truth around UA: It’s just not where you want to spend equity, and it’s not where your VCs will want to see you spend all their cash unless you can demonstrate that it fuels exponential growth.

Your number one priority in applying your LTV model in the real-world is to map out a UA strategy and spend that will allow you identify what you want to recoup by when. Some app companies live by the numbers and will pull the plug when they feel they are missing their targets. Of course, if they are losing patience too early in the app funnel, then this safe choice can suffocate chances of achieving real success.

Other app companies take a different approach, recycling money on a loss to buy a little more time in order to close another funding round. The reasoning is clear if you consider that the alternative is doing nothing, which amounts to a death stroke for your app and – worse – shows no turnover to possible investors.

As a marketing person at heart, I have to stress that no company has won a market only with UA. It’s crucial to build a brand, not just grow your app audience. I am confident that spending to get your brand out there and in front of your target audience, will pay off (particularly when Word Of Mouth hits the point that it ignites a lift in organic installs).

But whether its UA spend or brand spend — I recommend you try to stay away from using your equity to fund it unless you are already on a fast growth trajectory. That is where alternatives like Pollen VC come in to help you tap into your earned revenues faster, so you don’t have to sell off a part of your company to cover bank loans or credit card overdrafts – or both.

Read more about further LTV models and how portfolio management are saving loss-leading UA campaigns in my upcoming e-book and follow me on twitter @_oliverkern_