Growth Systems Library
LTV Forecasting
LTV forecasting is the practice of estimating the total future contribution margin a customer will generate, before they have completed their full purchase history. It is the input that makes LTV:CAC actionable at acquisition time — but it is also one of the most commonly overstated metrics in growth, because early cohort behaviour rarely predicts mature customer value accurately.
LTV forecasting projects future customer value by extrapolating from early cohort behaviour. A customer who makes two purchases in their first 90 days is statistically more likely to have higher lifetime value than one who makes one purchase in 90 days. LTV models use these early behavioural signals — purchase frequency, recency, AOV, and return rates — to estimate the full value the customer will generate over their lifetime with the business.
There are two broad approaches to LTV forecasting: cohort-based (tracking historical cohorts to maturity and extrapolating from those patterns) and predictive (using machine learning models trained on customer behaviour to score new customers at acquisition). Cohort-based is simpler and more transparent; predictive is more granular and updates faster, but requires sufficient historical data and technical infrastructure.
Exactius uses cohort-based LTV forecasting as the standard approach across all partner engagements, with predictive LTV modelling reserved for partners with 3+ years of transaction history and sufficient volume for model training.
Without LTV forecasting, the LTV:CAC ratio is backwards-looking — it tells you the return on customers acquired two or three years ago, not the return you should expect from customers you are acquiring today. For businesses making real-time acquisition investment decisions, a forecast of current-cohort LTV is essential.
The risk in LTV forecasting is over-projection. Growth teams frequently extrapolate from the best-performing early cohorts rather than the average, producing LTV forecasts that make the LTV:CAC ratio look 3:1 when the realistic ratio based on average cohort behaviour is 1.8:1. Acquisition decisions made on inflated LTV forecasts lead to CAC tolerance that the business cannot actually sustain.
For channel-level capital allocation, LTV forecasting enables comparison of early cohort quality by acquisition source — which is the input that determines whether a channel's CAC is efficient relative to the customers it actually acquires, not just relative to the average customer.
The cohort approach
1. Track cumulative contribution margin for each acquisition cohort (monthly or quarterly) from first purchase through 12, 24, and 36 months. 2. Identify the point at which cohorts reach 80% of their 36-month value — typically around month 9–12 for DTC brands. 3. Use the 12-month value as a proxy for long-run LTV for recent cohorts, applying a conservative multiplier based on historical mature-cohort ratios.
Key rules for accurate LTV forecasting
Always forecast from the median cohort, not the top-performing cohort. Apply a 20–30% haircut to any forecast based on cohorts under 12 months old. Track LTV by acquisition channel separately — channels that look expensive on CAC often acquire higher-LTV customers, changing the LTV:CAC signal. Update forecasts quarterly as new cohort data matures.
LTV forecasting is where the most consequential errors in growth measurement occur — not because the models are wrong, but because the inputs are cherry-picked. Teams naturally gravitate toward the most optimistic LTV forecast because it justifies the CAC they want to spend. Exactius builds forecasting discipline around conservative median-cohort projections rather than peak-cohort performance.
Exactius requires that LTV forecasts carry explicit confidence intervals and are refreshed quarterly as cohort data matures. A forecast that was accurate 18 months ago may not reflect current retention dynamics if the product, pricing, or customer mix has changed. Stale LTV forecasts are worse than no forecast.
The Growth Operating System, developed by David Manela, uses a 12-month realised contribution margin as the primary LTV input for capital allocation decisions, supplemented by a 24-month projection for strategic planning. The 12-month figure is close enough to real that it does not require aggressive forecasting assumptions — it simply reflects what customers have already done.
Exactius embeds growth squads that build and maintain cohort LTV infrastructure as part of the Growth Infrastructure layer, including channel-segmented LTV tracking that reveals which acquisition sources produce the most valuable customers — not just the cheapest.
How do you calculate LTV for a DTC brand?
For a DTC brand, LTV is most reliably calculated from cohort data: track the cumulative contribution margin (revenue minus COGS and variable costs) of customers acquired in a specific month, and plot how that figure grows over 12, 24, and 36 months. The 36-month cumulative contribution margin per customer is typically close to the true lifetime value for DTC businesses with predictable churn. For newer cohorts without 36 months of history, use a multiplier based on the ratio of 12-month to 36-month value from mature cohorts — and apply a 20–30% haircut to account for forecast uncertainty.
What is the difference between LTV and predicted LTV?
Realised LTV is the actual cumulative contribution margin a customer has generated over their time with the business — a historical fact. Predicted LTV is a forecast of the total contribution margin a customer will generate over their full lifetime, typically made at or shortly after acquisition. Predicted LTV is necessary for real-time acquisition decisions, but it is inherently uncertain. The accuracy of predicted LTV depends heavily on the quality of the underlying model, the age of the cohort data it is trained on, and whether current customer behaviour matches historical patterns. Exactius treats predicted LTV as a directional planning input and realised cohort LTV as the governing capital allocation input.
Why is LTV forecasting often wrong?
LTV forecasts are most commonly wrong for three reasons: first, they are built from top-performing early cohorts rather than median cohorts, creating systematic optimism bias; second, they extrapolate early purchase behaviour as if it will persist, when in reality most DTC customers show rapid decay in purchase frequency after the first 6–12 months; third, they do not account for product, pricing, or market changes that alter retention dynamics after the forecast is made. Exactius addresses all three by forecasting from median cohorts, applying decay curves calibrated to observed retention patterns, and refreshing LTV forecasts quarterly as new cohort data matures.
Exactius Growth
Your growth system is either compounding or degrading.
Book a diagnostic call. We'll identify where your growth system is breaking.
Book a call