LTV / CPI vs. cashflow

Posted on November 16, 2015 by Eric

Quentin_MetsysXXThe_Banker_and_his_Wife_1514

The ability to think about mobile marketing campaigns from the perspective of an LTV / CPI delta is largely a luxury afforded to developers with substantial balance sheets (or, unfortunately, an existential mistake made by small developers without sufficient expendable cash to finance growth).

Running mobile marketing campaigns against the profitability of LTV over CPI (that is, profitable unit economics of revenues vs. marketing costs) requires some available pool of capital to spend against before cohort revenues make up for the unit costs of acquisition. In other words, because LTV is paid out over some projected time period (eg. 12 months, 18 months, etc.) yet marketing costs are born upfront (usually due within 30 days of a campaign being run), developers need to consider not only the ROI of their campaigns (ie. the degree to which LTV exceeds CPI) but also the schedule of the cashflows that their acquired cohorts generate.

For instance, consider the below logarithmic LTV curve for some cohort definition (geography / network / device / etc.) that produces an 365-day LTV of $1.52 (all charts and their underlying data can be found in this spreadsheet EDIT: link removed, please request access to the spreadsheet via email):

example_ltv_curve

Assume this developer has decided that any marketing spend must generate 50% ROI after 365 days (that is, $1 spent produces $1.50 in total cumulative per-user revenues by day 365). This developer can achieve that ROI by buying cohorts with a $1 CPI, with each cohort achieving break-even at day 48 (as per the LTV tab in the spreadsheet linked above). Each cohort is expected to remain active for 365 days.

Now assume that this developer is able to acquire cohorts of 1,000 users per day at this price ($1 CPI); the developer will run this campaign for one full year before turning it off completely.

Since the Day 1 cumulative spend of each cohort (see the LTV tab) is $0.05, the developer is paying $1,000 per day to acquire cohorts that generate $50 in revenue on that day (for a Day 1 loss of $950). Each cohort's cumulative profit / loss curve looks like this:

cumulative-cohort-profit-loss

As the cohorts accumulate, marketing spend losses accumulate to a greater degree than do revenues, until Day 48, when the revenues generated by the 48 live campaigns ($1,000.86) exceed the upfront marketing cost paid to acquire the 48th campaign ($1,000). Keep in mind that the revenue schedule of this LTV curve is specific to this example and was generated randomly.

The developer's cash inflows look like this:

cash-inflows-outflows

Which is to say, the developer is spending more money than it is making until Day 48 (revenues spike at day 366 because the developer turns off marketing spend yet still has 364 cohorts active).

The developer's cumulative profit / loss by day looks like this:

cumulative-total-profit-loss-by-day

What this example illustrates is that, despite making roughly 50% ROI on all marketing spend (earning back $1.52 for every $1 spent on marketing), this developer is spending more money than it is making for more than a month (48 days) after launch. The total spend of $365,000 produces revenues of $553,848, for an operating profit of $188,848.

But the developer had to float a loss of $11,454 before it started generating positive cash flow (eg. was making more money than it was spending). Even at a massive profit margin, this developer needed 6% of two-year profits and 2% of two-year sales up front (that is, offset by cumulative recoup) to fund its marketing campaigns. Before launching a product, every developer needs to consider its ability to do the same.

  • jokim1

    What do you think of some of the factor companies out there factoring against estimated LTV like Pollen VC? Know of any others? Word on the street is some of the big guys (top grossing not King/Supercell) do this on a regular basis.

    • ESeufert

      I wouldn't be surprised if some of the bigger tier 2 developers were using credit facilities for launches. Regarding the factoring companies: it's an interesting business model. PollenVC is the only one I know of.

      • relyes

        Banks are hopping into this financing game now too, particularly for companies with VC backing (some capital leverage) using a lines of credit / debt financing model. This is a super low risk lending function for the banks (thus, it needn't be expensive for the game company), and when combined with AR financing against store receipts, these gaps in cashflow are very reasonable to cover.

  • Hi Eric,
    I have a question about spreadsheet from the article (https://docs.google.com/spreadsheets/d/1LABbXSovs3dTyyLLhiHLZaVa7VCCSZsiT3GNHsWUgp0/edit#gid=136468388):

    The average CPI and average cohort may be different every day. In spreadsheet there is only 1 value (1000 for cohort and 1$ for CPI). Do I have to fill in a value for each day and each cohort separately? (for more accurate results)

    @ESeufert:disqus

    • ESeufert

      No, that's meant to be an average across the life of the campaign, just to illustrate the point.

  • Gennady Klyachman

    Hi Eric,

    Thank you for yet another great article that helps me to calculate my expenses.

    I’ve been through the spreadsheet and i think i now understand how i can use it in my own calculations. The only question i have is regarding the “Actual Spend” you use on “LTV” sheet. Is it basically a projected ARPDAU for a single cohort through its’ entire lifetime or am i missing the point?

    Thank you!