Every founder who calls me about rising CAC has a spreadsheet ready. Almost none of them have a cohort chart. That is the whole diagnosis, most of the time.
CAC is loud. It arrives daily, in a dashboard, with a number attached. Retention is quiet. It fails slowly, months after the money was spent, and by then the auction gets the blame.
The five-minute test
Pull your last six monthly cohorts. For each, chart cumulative revenue per customer at day 30, 60, 90 and 180. Then answer one question: is the day-180 number for the most recent cohorts lower than it was a year ago?
- If day-180 value is flat or rising and CAC is up 20%: you have an acquisition efficiency problem. Go fix creative and channel mix.
- If day-180 value is falling: your CAC is not the problem. You are refilling a bucket with a hole in it, and every efficiency gain you make will be eaten by the leak within two quarters.
In my experience the second case is roughly two-thirds of brands between $5M and $50M. Nobody wants it to be, because acquisition has vendors who will sell you a solution and retention mostly has homework.
Why retention decays without anyone deciding it should
It is rarely one thing. It is a discount that trained people to wait. A product line extension that diluted the reason anyone bought. A fulfillment slip that never got measured against repeat rate. A second-purchase moment nobody owns, because acquisition owns week one and lifecycle owns month six and the gap between them belongs to no one.
Retention is not an email program. It is the sum of every promise your marketing made and your operation had to keep.
What I do in the first 90 days
- Name the second-purchase window. For most consumables it is 21–45 days. Everything in that window gets one owner and one weekly number.
- Kill the discount reflex. Replace sitewide promos with a reason to come back that is not price: replenishment timing, access, a product ritual, membership.
- Interview 20 lapsed customers. Not a survey. Phone calls. You will hear the same three sentences and they will be more useful than any attribution model.
- Instrument repeat rate by acquisition channel. Some of your cheapest customers are your worst. Reallocating toward channels with better second-order rates typically moves blended economics faster than any bid change.
- Fix the one operational thing everyone already knows about. There is always one. Ask a customer service rep, not a dashboard.
None of this is clever. That is the point. The clever work is in acquisition because acquisition is where the vendors are. The money is in the part nobody is selling you.
Get day-180 value moving and your CAC problem becomes a rounding error — because you can suddenly afford to outbid everyone who is still optimizing the cheap half of the equation.