ROAS = Return On Ad Spend = Revenue / Ad Spend
With this KPI, you are measuring if your paid acquisition campaigns are yeilding a positive return, ie: bringing back more revenue than you've spent. ROAS is expressed a ratio (2:1) or percentage (200%).
Wording: ROAS, DROAS, ROI, eROAS, LTV:CAC...
ROAS is the equivalent of a metric investors are keeping a very close look on: "LTV:CAC ratio" (hint: they're going to look for 3+).
While wording can vary from a company to another, I personally use "ROAS" to talk about the direct measurable effect of campaigns, without factoring virality, organic uplift and other organic traffic. This is sometimes called "Direct ROAS" or "Direct ROI".
"ROI", in contrast, would be used to discuss "Blended ROI" which includes all revenue (tracked & hidden virality, organic uplift, organic baseline) and measures the overall profitability of an app for a given timeframe, not of its sole UA campaigns. Some use the term eROAS instead of ROI. For some channels, attribution may not be straightforward, and the line between Direct ROAS and Blended ROAS may be more blurry. This is particularly true for influencer marketing, TV and other offline campaigns.
While there is no exact consensus on the definition itself, there are also discrepancies on the expression of its own value. By the first definition (revenue/spend), a campaign yeilding $50 on $100 invested would be 50% ROAS, or 1:2, but some express it as (revenue/spend -1), and this campaign would be said to have -50% ROI: it indeed lost money.
Evolution of UA KPIs: from CPM to ROAS
For a long time, app marketers have taken the simplistic route and were measuring the efficiency by looking at their CPI. CPI was already a step forward measuring campaigns based on high funnel metrics like Reach, CPM or cpc, but isn't a complete representation of performance, as it says nothing about the revenue generated by those installs. As the industry matured, UA metrics moved down the funnel to CPA (cost per action, based on events count) and eventually ROAS & ROI become the primary KPI of User Acquisition (UA).
For many, ROAS is the ultimate measurement of app marketing campaigns. Advanced advertisers often contrast this pure profitability figure with additional measurement of traffic quality, including long term retention or virality, among other positive side-effects of UA campaigns.
The importance of granularity
"ROAS" can be used to describe the overall performance of UA, and gives a good top-level understanding of profitability. Yet, app business should aim at being able to observe it with the maximum level of granularity, including network (channel), campaign, adgroups/targeting (including geo & OS) and ads/creatives, to be able to act on the detailed levers of their ad spend efficiently.
Timeframe & payback period
There is no standard for the timeframe in which ROAS is measured, most of the time ROAS will be defined by the lifetime value (LTV) of these cohorts. Due to the need for immediate feedback about campaign performance, most are using projected values of LTV, or pLTV, to estimate ROAS. Some marketers use a shorter timeframe, like first week ROAS or first month ROAS (m0-ROAS) to work on exact data instead.
LTV-ROAS does not provide information about how fast money is recouped, and understanding the velocity at which LTV compounds to the business is critical in determining the payback period of adspend, to anticipate cash flow trends. A perfect profitable app business could actually shut down if payback is not factored properly:
"Companies can go bankrupt buying advertising on an LTV-profitable basis", Eric Seufert
Defining Ad Spend
UA costs are usually straightforward. It's what your company is paying for the campaigns, and there are usually no issue on taxes, fees, converting currencies, etc. Plain and simple.
It doesn't necessarily refect the full extend of return on investment (some channels require much higher maintenance & execution time), but factoring the cost of running the campaigns is a complex exercise, rarely reliable nor comparable. Employees' time and agencies fees are for usually excluded from the calculation and accounted under structure costs.
Defining revenue
In most cases, marketers are working on net proceeds: the "revenue" in the initial formula would be counted as "net revenue after tax & fees", otherwise called "proceeds" as by Apple standards. Using gross value is not helping to understand if the ad spend is actually profitable and is a mispresentation. For the same reason, you want to factor revenue as your comission, not bookings, in the case of e-commerce or marketplace for instance.
Calculating pLTV
The calculation of this LTV net revenue to estimate to its full extent the impact of campaign can vary vastly depending on the business model of the app. In particular between e-commerce, IAP-based monetization (including subscriptions), ads-based monetization, or a mix of the above. Read more about pLTV calculation and the different cases mentioned in this guide to pLTV from Appsflyer.
At the end of the day, without consensus on the terms and their calculation, the vocabulary you pick and definition you attach to the two sides of the equation may be specific to your business, but they should be consistent across channels and over time to enable comparisons and eventually surface if your UA activity are making sense from a profitability point of view.