Cian
Cian
Aug 9, 2017 · 2 min read

Hello again. So here you are assuming there is a large distribution of CPAs for a Target (Average) CPA goal. Maybe / maybe not — we actually see it quite tightly distributed to the average which counter to your argument I feel a bit sad about.*

Regardless if you are setting a Target CPA based on a robust LTV calc you should be already factoring in that LTV has a distribution too. (spend £10,000 to get 1000 users which overall generate £10,000 in margin over next 12Ms. Some users are worth £1, some are worth £20. Maybe most are worth roughly £10 — you shuld be able to know this.). On paper I’m happy with that outcome if my goal is break-even in aggregate vs profit maximising (there is a linked debate of objective function of marketing and marginal ROI/CPAs).

If you are granular with both your CPA bidding and your LTV calcs and the latter is robust then an ROI /payback goal on all your activity (eg only spend money if I make it back in next 12Ms)= to set CPAs to match LTV at a campaign level for example and it’s all gravy in aggregate. It might mean CPA overall goes up but you are moving money to high quality users and still delivering your objective function of your marketing £.

If my target CPA is £10 and most my campaigns are £10 then maybe I am missing some £20 CPL dudes who are worth twice as much but I’m too scared of moving my average CPA to explore that option.

LTV calc. Gift and a curse.

    Cian

    Written by

    Cian

    strongman robot ninja assassin.