Our GDC 2024 presentation "How not to f**k your studio: Forecasting done right" now available. Download here >
.png)
You cannot maximise the full potential of your free-to-play title unless you're really really good at UA. That is a fact. No game would perform better without significant UA investment. However, if you are one of the lucky ones in a position to be spending large amounts on UA, spend it responsibly.
As marketers, we have the responsibility for what is often the largest cost line item in the business. What we do can make or break a studio, and often does. It's why I'm advising many companies these days that they should be spending much less than they are. Because at the end of the day, the real job of a marketer isn't just to spend money but to make money.
"The job of a marketer is not to spend money but to make money."
80% of my time at King was spent on CLV prediction and modelling, not on making ads look good or getting my CPIs down. The way to focus on the former is obviously by investing in analytics.
Analytics is gold dust. It is particularly under-invested in terms of headcount and the analytics stack. And when it comes to analytics, there are capabilities that I think you absolutely must have:
Everybody's cumulative CLV curve is different - and not all games are created equal. Using the below graph to illustrate this, you could have two games which are working to the same payback period of one year. Given the shape of the curve, Game A is 66% more profitable after three years. Yet, somehow everybody defaults to "we're buying to one year". It's important do your own modelling and choose the payback periods which are right for you.
Another thing about paybacks is that up until recently, the mobile gaming industry was quite content with two-year paybacks. Two years! Absolute madness. For two reasons: firstly, predicting a CLV curve two years out is extremely difficult. Secondly, targets are hard to hit. They're volatile, especially if you're working in a small company where data isn't always sufficient. Sometimes you miss, and when you miss, it becomes more damaging the further out you aim.
If you miss your target by 10% at a payback of 365 days, it's only going to blow out by 126 days, which isn't the end of the world. But if you miss aiming for two years, it's going to blow out another 242 days beyond your two-year period. So the time to recoup gets significantly worse.
People inherently get that as you're spending more on UA, your CPIs are going up. Below is a real world example of a game that has rapidly dropped their spend over a period of time. The red dots represent individual days.
What we've done here is we've plotted the relationship between the amount of spend on the x axis and the amount of installs on the y. You can clearly see that there is a diminishing return.
However, if you believe what is being showing in the graph above, then the below must also be true: cost per incrementality.
The blue line represents your CPI rising over time, as you spend more money. But if you think about it in terms of the cost per incremental, what's the additional budget being spent? Divided by the incremental amount of new installs that we achieved, that relationship is a lot steeper. Essentially it shows that you're paying double for those incremental users. It doesn't make any sense to do that.
Things could get even worse when you convert it into a payback period - you may never make that money back!
The reason why subsidising paid with organic is is so dangerous is because the organic installs you are getting today, are most likely the result of paid installs you bought years ago. The relationship between organic installs isn't off paid one to one; it's off active usage, which has a much closer relationship. Active usage takes time to build, so there's a massive lag there.
This is why a lot of games tend to underinvest during their early life cycle of the game and over invest later. Below is an example of a real game.
Looking at the above, this studio is spending to a sub 12-month payback (grey bar). The problem with this is, if you look at the split of installs represented by the pink (paid) and the blue (organic), the free to play split is reversed compared to what you'd normally expect if you were spending at scale. This happens a lot later in the life cycle of a game.
If you look at their results on a paid-only basis, it would be catastrophic. This suggests that you are destroying value with current spending patterns.
There's typically reluctance to cut spending because of the fear that it will lead to fewer organic installs, potentially crushing the game’s growth. However, examining the actual results of reducing spending can be eye-opening (see below).
For the game in question, reducing spending led to making half the money but doubling the profit, which, for a small team in a small studio with limited cash flow, is a necessary trade-off.
Unfortunately, this studio wasn't even aware that this was an option until the analysis was presented. This highlights the importance of not only closely monitoring the impact of your spending on both paid and organic installs but also being willing to adjust strategies based on empirical data.
"If everybody were disciplined with UA spend, everyone would be better off."
At the end of the day, if everybody were disciplined with UA spend, everyone would be better off.
--
This blog post was adapted and condensed from Angus's GDC 2024 presentation: You're probably wasting money on user acquisition. You can download his full presentation here.
