Building a marketing P&L using LTV and ROAS | Mobile Dev Memo

Building a marketing P&L using LTV and ROAS

Note: this post is adapted from content from the upcoming Modern Mobile Marketing at Scale workshop series taking place in NYC, SF, and London in October. More information here; places are still available for the SF workshop.

Performance marketing costs tend to be the largest, or at least one of the largest, operating expense line items for mobile-first companies. Unless a company uses a model like Blended-Same Month Return to recoup cohort revenues within the same calendar month as acquisition, its cohorts will feature what I call a “Projected Receivable” over their lifetimes: some amount of revenue that is expected to be contributed by the cohorts in the future, based on the LTV model that was used to value them for acquisition.

In the diagram above, where a cohort (“Cohort A”) tracks precisely to its predicted cumulative revenue model, the curve is roughly split in half into “Recorded Revenue,” which is the revenue that has been generated by the cohort to-date, and “Projected Receivable,” which is the amount of money that the advertiser believes that cohort will contribute through some future date (Day Z on the X axis). This amount of money, for any given cohort, can be thought of as a current asset: it is revenue that has been paid for (through acquisition marketing) but not yet received.

Applying some real data, imagine a projected cumulative monetization curve for some type of traffic that tracks to $1 over 90 days and realizes 67% of its 90-day value after 20 days ($0.67):

At any given point in time through 90 days after acquisition, any cohort acquired against that cumulative monetization curve has an attendant Projected Receivable. For instance, a 20-day old cohort has a per-user Projected Receivable of $0.33 ($1 – $0.67):

If that 20-day-old cohort was the first cohort acquired in an acquisition campaign, and a subsequent 19 cohorts were also acquired, the total Projected Receivables for all of those cohorts would be the sum of each cohort’s per-user Projected Receivable times the number of users acquired each day:

Why does this matter? Because juxtaposing this concept against acquisition costs provides some insight into the total amount of money an advertiser needs available, up front, to finance acquisition campaigns before the daily P&L looks positive:

Characterizing these campaigns:

  • They are being run to break even at Day 90: a constant 1,000 users is being acquired each day for a $1 CPI at a projected per-user cumulative revenue of $1 at Day 90;
  • The recoup on these campaigns is very aggressive: 67% ROAS by Day 20 means that much of the recoup is front-loaded;

The take-away here is that even with a very short recoup window for campaign spend (100% ROAS by Day 90) and a front-loaded recoup curve (67% of spend is recouped in just 20 days), after 20 days of campaigns, less than half of all spend has been recouped. Which begs the question: after how many days of running the campaign under the above conditions ($1,000 in daily spend against a $1 CPI) would the advertiser recoup its spend on these 20 cohorts — when would the total revenue received from them exceed the $20,000 paid to acquire them?

Obviously the timeline would be more than 90 days because each cohort only recoups at Day 90. If the cumulative monetization chart is projected out to 180 under the same parameters, then the Day 180 value reaches $1.15 (meaning the cohorts being acquired at $1 CPI values see 15% profit margins by Day 180):

In order to understand when the campaigns reach break even, the revenue of each cohort needs to be projected out along the cumulative monetization curve. But first, it’s necessary to break the cumulative revenue curve out into actual daily contributions, ie. to map not the total running sum of revenue contributed by day, but the actual revenue contributed each day along the curve timeline:

Once the cumulative revenue curve is broken out into actual daily contributions (eg. if the Day 1 cumulative revenue amount is $0.05 and Day 2 cumulative revenue amount is $0.15, then the Day 1 actual revenue amount is $0.05 and the Day 2 actual revenue amount is $0.10), each cohort’s actual revenue contribution can be mapped and these values can be compounded or “stacked” onto one another to get to a total, cross-cohort revenue map by day:

With these values broken out by cohort by day, a full picture of cumulative revenue by day vs. cumulative spend by day can be drawn. In the graph below, Cumulative Profit / Loss (the green / red lines) are on the right Y axis and Daily Revenue and Daily Cost (the blue / magenta lines) are on the left Y axis. Daily cost goes to $0 after Day 20 since the campaigns are no longer being run. Daily Revenue also spikes at Day 20 because, since the cumulative monetization curve is so front-loaded (ie. the bulk of the curve is generated in the first few days), new cohorts contribute more money than old cohorts on a daily basis.

From the graph below, we can see that the 20 cohorts achieve break even in terms of cumulative revenue contributed versus cost at Day 100, after which all revenue is profitable:

In this scenario, we know that total marketing spend is $20,000 (20 cohorts * 1,000 users per cohort * $1 acquisition cost). But since money is coming in as the cumulative monetization curve per cohort is realized, the maximum amount of money at risk at any given point in time — that is, the amount of money that wouldn’t be recouped if, for some reason, all cohorts immediately churned and the entirety of the cumulative Projected Receivables balance went to $0 — is $10,530. In other words, the advertiser would need $10,530 in order to spend $20,000 across 20 days, given that early campaign revenue recoup would make up the difference.

And with a total profit of $2,820 after 180 days, the advertiser makes a 14% profit margin on the $20,000 in spend:

It’s important to note here that these recoup curves only apply to the 20 cohorts that were acquired; this is not an ongoing recoup map. If this campaign was left running — that is, if the acquisition of new users against the same cumulative monetization curve and at the same volume and cost per install didn’t stop — then the campaign would hit Daily Marketing P&L Break Even (ie. compounded cohorts would generate more revenue each day than was being spent on acquisition each day) at day 92, and the campaigns would hit Cumulative Break Even (ie. the overall campaign cumulative revenue would exceed the overall campaign cumulative spend) at Day 257. Over the course of 360 days, the campaign would generate $375,590 against $360,000 in spend for a profit margin of 8.6%:

It’s vital to go through this kind of exercise in planning an advertising budget. These advertising and recoup dynamics dictate how a business is run and how quickly and to what extent it can grow: without making considerations for recoup timelines and overall cash at risk, a company isn’t being systematic with its advertising spend. It’s critical for an advertiser to understand not just when a cohort pays back, but also when a set of cohorts — an entire block of advertising spend — becomes profitable, if ever, and how that schedule of cash flows aligns with their available capital.

Photo by Eric Rothermel on Unsplash