Boxplots and risk management

I’m having more conversations around risk management ever since the beginning of this year with smaller app developers. With recent tightening happening in the capital markets, app developers are starting to feel there are better ways to manage the capital that they deploy into marketing campaigns.

A simple way that app developers are evaluating “risk” is the variance in x-day LTV or ROAS that they observe over a daily time series. A back of the envelope mechanism for doing this is creating a box plot from daily LTV or ROAS data. For a specific return, advertisers evaluate an appropriate level of variance they are willing to take based on historical risk vs reward outcomes.

These analyses only matter in cases where capital is limited and an app hasn’t yet created a flywheel for reinvestment, so often times risk measurement is only used for scenarios that involve launching new apps. Most big developers who have the ability to allocate larger amounts of financing only need to prove LTV consistency to obtain additional flywheel financing, but as things tighten up, some of the methods for understanding and allocating risk could end up becoming more relevant.