What is LTV cohort analysis?
LTV cohort analysis groups customers by when they were acquired or by which channel acquired them, then tracks how their revenue accumulates over time. Instead of looking at a single total revenue figure, you see the revenue curve for each group over months or years, which reveals which cohorts and which channels actually produce sustainable economics.
The shift from total revenue to cohort revenue is the shift from a snapshot to a trajectory. Two businesses with identical current revenue can have wildly different futures depending on whether their customer base is growing or churning. LTV cohort analysis makes that trajectory visible before the business notices it in the aggregate.
Why first-order ROAS is a trap
First-order ROAS measures the revenue from a customer's first purchase against the cost of acquiring them. It is easy to compute, popular in reporting, and systematically misleading for any business with repeat customers.
The pattern that recurs across industries: channels that attract deal-hunters show excellent first-order ROAS, because those customers buy the discounted first order and never come back. Channels that attract brand-loyal customers show weaker first-order ROAS because their customers buy at full price and without a promotion, then repeat for years afterwards.
A marketing team optimising for first-order ROAS will systematically scale the channels that destroy lifetime value and starve the channels that build it. This is the single most common measurement mistake in consumer subscription, DTC, and SaaS businesses. For the foundation on ROAS itself, see what is ROAS. For the deeper case on incremental value, see reported vs true incremental ROAS.
Worked illustration
Two channels, each costing $50 CAC. Channel A brings customers who spend $80 on the first order and never return (first-order ROAS 1.6x, LTV $80). Channel B brings customers who spend $60 on the first order and $180 over the next year (first-order ROAS 1.2x, LTV $240). Channel B is three times more valuable, and a first-order ROAS optimiser would cut it.
How cohort curves work
A cohort curve plots cumulative revenue per customer against time since acquisition, usually with separate lines per acquisition group. Read from left to right, the curve tells the full economic story of each cohort:
- Height at day zero is the first-order value (or trial-to-paid conversion for SaaS).
- Slope in the early weeks shows repeat rate. Steep = customers buy again quickly. Flat = one-and-done.
- Curvature around months 3 to 6 shows retention stabilising. A curve that flattens means churn is eating the cohort; a curve that keeps climbing means customers continue to expand.
- Height at 12 or 24 months is the LTV milestone used for unit economics decisions.
Side-by-side curves (cohort A vs cohort B, or channel A vs channel B) reveal which is delivering stronger economics. Two curves with identical day-one values can diverge wildly by month six, and that divergence is usually the single most important signal for acquisition strategy.
LTV by acquisition source
The operational payoff of LTV cohort analysis is segmenting customers by the channel that acquired them and comparing LTV curves per source. This is where strategic budget decisions actually happen.
Every acquired customer carries the source that brought them in (captured at signup or first purchase via first-party attribution). Revenue over time is tracked per customer. Rolled up by source, the result is an LTV curve per channel: Meta vs Google vs organic vs email vs referral. The differences are usually stark.
Common pattern: TikTok / Meta prospecting
Higher acquisition volumes, moderate first-order, moderate LTV. Healthy balance in most mixes.
Common pattern: discount codes and deal sites
Strong first-order ROAS, weak retention. Scale carefully; the LTV curve rarely justifies a big bet.
Common pattern: organic search
Moderate day-one, excellent LTV. High-intent customers who found you after research and tend to stick.
Common pattern: referrals and word-of-mouth
Often the highest LTV channel, even when first-order values are not remarkable. Customers arrive pre-trusted.
Retention curves and churn
Retention curves are the other side of LTV. Instead of cumulative revenue, a retention curve shows the percentage of customers still active at each time point. LTV rises when retention is high; LTV flattens when retention degrades.
The two curves read together tell the full story. A cohort with high day-one revenue but sharp churn produces a steeply-rising, then flat LTV. A cohort with modest day-one revenue but high retention produces a gradually-rising LTV that keeps climbing. Neither is universally better; which one you want depends on unit economics, cash flow, and strategic positioning.
For subscription businesses especially, retention by cohort is where product-market-fit and pricing decisions get validated. A cohort showing improving month-over-month retention after a pricing change is a different business from one where retention is degrading quietly.
LTV for DTC, SaaS, and services
DTC and ecommerce
First-order ROAS is the single most abused metric in DTC. LTV by acquisition channel is the honest decision lens. Typical stabilisation: 6 to 9 months of tracking before 12-month LTV estimates are trustworthy. For the vertical detail, see ecommerce attribution.
SaaS
LTV is the core strategic metric. Trial-to-paid conversion, early churn, expansion revenue, and annual retention all live inside the LTV curve. Stabilisation requires 18 to 24 months because annual contract cycles take that long to resolve. See SaaS attribution.
Services and healthcare
First-job ROAS misses years of recurring service contracts, maintenance, and referrals. LTV cohort analysis reveals which acquisition channels produce long-term client relationships. Stabilisation can take multiple years. See services attribution and healthcare attribution.
Calculating LTV
Three common approaches, each with trade-offs:
1. Historic LTV (empirical)
Total revenue per customer, divided by the number of customers. Simple and honest; uses only observed data. The limitation is that it reflects cohorts acquired long enough ago to have accumulated revenue, which may not represent current acquisition behaviour.
2. Predictive LTV (statistical)
Use the retention curve and average order value to extrapolate LTV for current cohorts. The common form is LTV = (average purchase value × purchase frequency) / churn rate. Useful for forecasting; sensitive to the assumptions baked into each term.
3. Cohort-projected LTV
Blend of the two. Use observed cohort behaviour where possible, fit a retention model to extrapolate beyond the observed period. Typically the most reliable approach for businesses with 6 to 18 months of data.
The important practical point: LTV is an estimate, not a ground truth. Treat every LTV figure with a confidence interval, re-run estimates quarterly, and expect revisions as more data accumulates.
Using LTV for budget decisions
The practical outputs of LTV cohort analysis feed three kinds of decision.
- Channel reallocation. Shift budget toward channels with high LTV per dollar of acquisition cost, away from those with low LTV regardless of first-order ROAS.
- Target CAC revision. As LTV estimates stabilise, target CAC can rise for high-LTV channels (payback becomes less binding) and should fall for low-LTV ones. This affects bid caps, audience targets, and creative testing budgets.
- Product and pricing. Cohorts that show improving retention after a pricing or product change are a positive signal for further investment. Cohorts showing quiet retention degradation are an early warning.
The honest caveat: LTV is a lagging indicator. Decisions made today based on 12-month LTV curves assume the future cohorts behave like the past ones, which is often but not always true. Pair LTV with leading indicators (trial activation, early engagement, first-month retention) for faster feedback.
Data requirements
LTV cohort analysis needs four data streams:
- 1. Customer identifiers. A stable customer ID (email or internal user ID) that persists across sessions and purchases.
- 2. Acquisition source. The channel, campaign, and UTM that brought the customer in, captured at signup or first purchase via first-party attribution.
- 3. Transaction history. Every purchase or subscription event tied to the customer ID, with amount, date, and product where relevant.
- 4. Retention or activity signals. Login data, session data, or subscription status per customer over time, which feeds the retention curve.
For DTC businesses on Shopify, the core data flows automatically. For SaaS, the pattern is tracker plus product analytics plus billing system. For services, CRM and CSV bridges from scheduling systems cover the same ground. Attriqs' LTV cohort analysis feature works with any of these patterns.
Frequently asked questions
What is LTV cohort analysis?
LTV cohort analysis groups customers by when they were acquired (or by what channel acquired them) and tracks how their revenue accumulates over time. Instead of looking at total revenue today, it shows the revenue curve for each group over months or years. The comparison reveals which cohorts (and which acquisition sources) produce customers that stick, repeat, and expand, versus which produce one-time volume that never returns.
Why should I care about LTV, not just first-order ROAS?
First-order ROAS measures the revenue from the first purchase against the cost of acquisition. It is easy to compute and misleading for almost every business with repeat customers. Channels that bring bargain-hunters show strong first-order ROAS but weak repeat rates. Channels that bring brand-loyal customers show weaker first-order ROAS but multiples of lifetime revenue. Optimising for first-order alone scales the wrong channels systematically.
How is LTV different from CAC payback?
CAC payback is the time it takes for cumulative revenue from a customer to equal the cost of acquiring them. It is one milestone on the LTV curve, useful for cash flow planning. LTV is the broader trajectory: where does revenue end up at 12 months, 24 months, 36 months? The channels with the fastest CAC payback are not always the channels with the highest long-term LTV, and both metrics matter for different decisions.
Can you attribute LTV to specific marketing channels?
Yes, and this is the core value of LTV cohort analysis. Every acquired customer carries the attribution source that brought them in (captured at signup or first purchase). Revenue is tracked over time per customer, then rolled up by source to produce LTV curves per channel. Attriqs' LTV cohort analysis tags every customer with their source and retention curve so this view is available by default.
How long do I need to track customers to trust LTV estimates?
It depends on business type. DTC typically produces stable 12-month LTV estimates after 6 to 9 months of tracking, because a large share of repeat behaviour shows up in the first year. SaaS requires longer (18 to 24 months) because annual contract cycles and multi-year expansion play out more slowly. Services and healthcare can require multiple years to see the full picture.
What is a cohort curve, and how do I read it?
A cohort curve plots cumulative revenue per customer on the vertical axis against time since acquisition on the horizontal axis, usually split into separate lines per cohort (monthly or quarterly acquisition groups, or per channel). A steep curve means customers are generating revenue quickly; a gradual curve means revenue accumulates slowly; a curve that flattens means retention is degrading. Side-by-side curves reveal which cohorts or channels are delivering stronger long-term economics.
How does LTV fit with multi-touch attribution and MMM?
LTV, MTA, and MMM answer three different questions. LTV tells you what a customer is worth over time. MTA tells you which channels moved them along the journey to conversion. MMM tells you how much revenue each channel caused in aggregate. A complete stack uses all three: MTA and MMM to attribute the acquisition, LTV to judge whether the acquisition was worth the cost over time.