LTV & cohorts: are customers worth more over time?
A single average LTV blends your loyal buyers with your one-and-dones and lags months behind reality. Cohorts, grouping customers by the month they joined and tracking their cumulative value, show whether retention is actually improving.
AI can build the cohort table and contribution-based LTV from your data; you supply the real orders, because AI can't see them and must never invent them.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are a retention analyst. Use MY data only (invent nothing).
Paste what I have: for the last 6 monthly cohorts, the number of customers who joined and their cumulative revenue at 30/60/90 days (or raw order data).
My contribution margin %: [%]
Do this:
1. Build a cohort table: rows = join month, columns = cumulative contribution per customer at 30/60/90 days.
2. Compute contribution-based LTV (not revenue LTV) so it ties to what I can spend on CAC.
3. Tell me whether retention is improving, flat, or declining across cohorts, and what that implies for affordable CAC.
4. Flag if I'm relying on an average LTV that a cohort view contradicts.
If data is missing, tell me exactly what to pull; do not estimate.
Output: cohort table + contribution LTV + retention trend + CAC implication.
Or do it in 4 steps
- Use contribution-based LTV, not revenue. LTV in revenue overstates what you can spend, only the contribution margin (see contribution-margin) is money you actually keep. Your affordable CAC is anchored to contribution LTV, not headline revenue per customer.
- Group customers into cohorts by join month. Then track each cohort's cumulative value at 30, 60, 90 days. This separates "we grew" from "our customers got more valuable", two very different kinds of health.
- Read the trend across cohorts. If recent cohorts hit a higher 90-day value than older ones, retention is improving and you can afford more CAC. If they're declining, growth is masking a leaky bucket, fix retention before scaling spend.
- Don't trust a single blended LTV. It averages your best and worst customers and updates slowly. A cohort view catches a retention problem months before the blended number does, when you can still act on it.
Worked example (labeled): the January cohort reaches $70 cumulative contribution per customer by day 90; March reaches $85; May reaches $95. Retention is compounding, so your affordable CAC can rise with it.
If instead they fell 70 → 60 → 50, your blended LTV would still look okay for months while new customers quietly become less valuable. The cohort view is what exposes it. Build it from your real orders.
Refresh cohorts monthly; let the trend, not a single average, set how aggressively you spend.