Meerkats AI
← All posts
Retention·2 Jul 2026·9 min read

Repeat rate, LTV, payback: the math that decides if your ads can ever be profitable

Most D2C brands lose money on the first order — that's fine, if you know exactly when the customer pays you back. The cohort math, the formulas, and the point where retention says 'stop acquiring'.

TL;DR
  • First-order losses are a strategy, not a problem — but only if you know your payback window. Most brands assume it instead of measuring it.
  • LTV must be contribution-based (margin after COGS, shipping, returns), never revenue-based. Revenue LTV overstates by 2–3× and funds bad acquisition.
  • The bridge metric is payback period: how many days until a cohort's cumulative contribution covers its CAC. Under 90 days you can scale aggressively; over 180, acquisition is a loan you may never collect.

Here's the uncomfortable arithmetic from a real audit: average order value ₹586, contribution after product cost ~₹340, cost per acquired customer ₹1,746. Every first order loses ~₹1,406. That brand isn't broken — most D2C brands at scale lose money on order one. The question that decides everything is: does order two, three and four arrive fast enough to pay the loan back?

The three formulas, defined honestly

Contribution-based LTV (the only kind that counts)LTV = Σ (order contribution) per customer over horizon — NOT Σ revenue
Payback periodpayback = days until cohort cumulative contribution ≥ cohort CAC
The ratio everyone quotesLTV : CAC — useful only when LTV is margin-based and the horizon is stated

"LTV:CAC of 3" is meaningless without two disclosures: margin-based or revenue-based, and over what horizon? A 3:1 revenue-LTV over 24 months can be a 0.9:1 contribution-LTV over 6 — the same brand, bankrupt in the second framing.

Worked example: when does ₹1,406 come back?

MilestoneCumulative contributionAgainst ₹1,746 CAC
Order 1 (day 0)₹340−₹1,406
Order 2 (median day 38)₹720 (repeat AOV runs higher)−₹1,026
Order 3 (median day 84)₹1,130−₹616
Order 4 (median day 141)₹1,560−₹186
Order 5 (median day ~200)₹2,010+₹264 — payback

Two hundred days to break even. Whether that's fine depends entirely on repeat probability: if 55% of first buyers reach order two, this machine compounds. If 22% do, the cohort never pays back — the averages above only materialise for a sliver of buyers, and scaling acquisition scales the loss. Same CAC, same AOV, opposite verdicts. Only the cohort table knows.

The four retention levers, ranked by leverage

  • Second-order rate — the whole game. Moving 30% → 40% repeat does more than any CAC optimisation you will ever run. Post-purchase flows, replenishment reminders timed to consumption, and a reason to return (new flavour, refill pack).
  • Time-to-second-order — compresses payback directly. A day-30 nudge that pulls order two from day 55 to day 35 shortens the loan by three weeks at zero media cost.
  • Owned-channel share — email/WhatsApp/SMS revenue carries ~zero acquisition cost, so every order routed through owned channels is pure contribution. In India, WhatsApp open rates make this lever unusually strong.
  • Subscription / bundling — converts repeat probability into a contract. Even 15% of buyers on subscription changes cohort math structurally.
When retention math says stop
If contribution-LTV at 12 months is below CAC and the repeat curve has flattened, more ad spend is not growth — it's renting revenue at a guaranteed loss. The correct moves are price/AOV, product that earns a second purchase, or channel mix — not a new agency. The brands that die usually knew their blended numbers and never looked at a cohort.

Why platforms can't tell you any of this

Ad platforms see the click and the first conversion. They cannot see order two (it comes via email), returns (your 3PL), or margin (your cost sheet). So platform-optimised acquisition systematically selects for cheap first orders — which often means discount-hunters with the worst repeat behaviour. The fix is feeding your own definition of a good customer back into bidding (value-based audiences from repeat buyers), which requires the cohort join to exist first.

FAQ

What horizon should LTV use?
State one and hold it: 6 and 12 months are the operating pair for most D2C. "Lifetime" without a horizon is a pitch-deck number.
Our repeat rate looks stable — are we safe?
Blended repeat rate is a lagging average of old and new cohorts mixed. Track repeat by acquisition month; blended stability routinely hides three quarters of new-cohort decay.
Is 90-day payback a universal target?
It's a working-capital statement, not a law. Funded brands stretch to 6–12 months deliberately; bootstrapped brands financing inventory on cash need 60–90 days or the growth eats itself.
Do discounts on the first order ruin cohorts?
Measurably, often yes — discount-acquired cohorts repeat less and repurchase only on discount. Run the cohort split before scaling any voucher campaign; sometimes the effect is small, sometimes it's the whole margin.
What's the minimum data to do this properly?
Order history with customer IDs (Shopify has it), per-order contribution (needs your cost sheet), and CAC per cohort (needs the ad-spend join). It's plumbing — the same spine every other honest metric needs.

See your real numbers, not the platform's.

Connect Shopify, Meta and Google. Reconciled truth in 20 minutes.