Flat Retention vs Decaying Retention in LTV Inputs
Plugging a flat repeat-rate into an LTV formula is the single biggest source of calculator error. Here's the size of the gap, why it happens, and how to fix it.
Quick answer
A flat retention assumption (e.g. 'customers repeat at 30% per year') overstates LTV by 30-80% versus a decaying cohort curve, because real repeat rates fall fastest in the first 90 days and then flatten. Use a flat input only when payback is under 6 months and you only care about year-one LTV; otherwise feed a monthly decay curve from your actual cohorts.
Flat Retention vs Decaying Retention in LTV Inputs
Flat retention assumes a constant repeat rate every period; decaying retention models the real curve where most churn happens early and survivors stabilise.
When you compute LTV, the single biggest lever is the retention assumption you feed into the formula. A flat input — 'we keep 35% of customers each year' — is mathematically clean but ignores the shape of real buying behaviour. Decaying retention reflects what cohort data actually shows: a steep drop in the first 30-90 days post-purchase, followed by a long tail of repeat buyers who stabilise at a lower but durable rate.
The gap between the two methods is rarely small. For a typical Shopify apparel cohort with 28% twelve-month retention, a flat 28% input can overstate three-year LTV by 50% or more, because the flat model implicitly assumes the steep early churn never happens.
This page is for anyone running an LTV:CAC ratio calculator and wondering why the output feels too optimistic. The fix is mechanical, not philosophical — you just need to feed the calculator the right shape of input.
Why the gap exists
Customer retention is almost never linear. After a first purchase, the probability that a customer comes back drops sharply in the first 30-90 days, then decays much more slowly. The remaining cohort behaves like a different population — habitual repeat buyers who churn at a low, stable rate.
A flat retention input ignores both phases. It applies one average rate across every period, which over-counts the long-tail period (where actual retention is lower than the headline average) and under-counts the early period (where churn is much steeper).
The compounding trap
LTV formulas compound retention across periods. A 5-point error in your monthly retention assumption becomes a 30%+ error in 24-month LTV. Small misspecifications at the input stage produce large misspecifications at the output stage — which is why this single input matters more than most teams realise.
How to detect it in your model
The clearest signal is a mismatch between your model's predicted LTV and your actual revenue per acquired customer at the 6, 12, and 24-month marks. If the model says €240 LTV at 24 months but your finished cohorts deliver €160, the retention input is almost always the culprit.
A second signal: your model's LTV grows roughly linearly with time. Real LTV curves are concave — they grow fast in months 1-6, slower in months 7-12, and barely move past month 18. If your LTV chart looks like a straight line, you're using a flat retention input.
Flat vs decaying retention: cumulative LTV for a Shopify apparel cohort (AOV €60, 25% margin, starting month 0)
| Month | Flat 35%/yr input | Decaying cohort curve | Gap |
|---|---|---|---|
| Month 6 | €18.40 | €16.10 | +14% |
| Month 12 | €26.80 | €21.40 | +25% |
| Month 18 | €34.10 | €24.20 | +41% |
| Month 24 | €41.20 | €25.90 | +59% |
| Month 36 | €54.80 | €28.40 | +93% |
Cumulative LTV: flat assumption vs real cohort decay
Flat 35%/yr retention
Decaying cohort curve
How to fix it
Pull a cohort retention table from Shopify, your data warehouse, or your LTV measurement layer. You want monthly repeat-purchase rate for cohorts that are at least 12 months old, segmented by acquisition month so seasonality doesn't distort the curve.
Fit a two-phase decay: a steep rate for months 1-3 and a flatter rate for months 4+. A simple geometric decay (retention(t) = a * b^t) works well enough for most stores; survival-analysis models are overkill below €15M revenue.
Then feed the period-by-period retention into your LTV calculation rather than a single flat number. Most LTV:CAC ratio calculators that let you enter a monthly retention curve will give you a result within 5-10% of your actual cohort revenue — versus 30-80% off with a flat input.
When the flat input is good enough
Flat retention is acceptable in three narrow cases: payback period is under 6 months and you only care about year-one LTV; you're comparing two channels' relative LTV (the bias cancels out); or you're sizing a quick directional decision and a 30% error doesn't change the answer.
Outside those cases — and especially when LTV:CAC is the input to a budget decision — the flat assumption is the most expensive shortcut in DTC modelling. The decay shape isn't a nuance; it's the dominant feature of the curve.
Rule of thumb
If your retention curve at month 12 is less than 60% of your retention at month 1, a flat input will lie to you by 25%+ at the 24-month horizon. Always decay.
Experiment ideas to test before you trust the model
Before betting media budget on an LTV:CAC number, validate the retention curve itself. Pick the three acquisition channels driving 80% of new customers and compute a separate monthly retention curve for each — paid social cohorts often decay 30-40% faster than organic search cohorts, which flat models hide entirely.
Then test interventions that specifically target the steep early-decay window: a month-2 replenishment email, a free-shipping threshold on second order, or a post-purchase bundle offer. Even a 3-point lift in 60-day retention compounds into a meaningfully higher cohort LTV at 24 months.
Frequently asked questions
For a typical DTC cohort with 25-35% annual retention, a flat input inflates 24-month LTV by 40-60% and 36-month LTV by 70-100%. The error compounds with time, so the longer the horizon, the worse the lie.
A geometric decay — retention(t) = a * b^t, where a is your month-1 retention and b is the monthly decay factor — captures 80% of the real curve for most stores. Fit it on 12 months of cohort data using a basic regression in Sheets or your warehouse.
Less so. Subscription retention curves are flatter by design because churn is gated by an active cancel action, not a passive non-purchase. Flat inputs are usually within 10-15% of true LTV for subscriptions — but for one-time-purchase DTC, the gap is much larger.
No. A discount rate adjusts for the time value of money; it doesn't fix a misspecified retention curve. You'd need a wildly aggressive discount rate to offset the compounding error, and that breaks every other comparison your finance team cares about.
Shopify's native reports include a Customer Cohort Analysis view that gives you monthly repeat-rate by acquisition month. Export 12-18 months of cohorts, average them by tenure month, and you have your decay curve. Most warehouse setups (Fivetran + dbt) produce the same table in a few hours.
Use a public benchmark curve from your vertical as a starting point — apparel decays faster than beauty, beauty faster than home goods — and replace it with your own data as soon as you have 6 months of cohorts. A benchmark curve is still closer to truth than a flat input.
Directly. If your flat-retention LTV is €60 against a €25 CAC, your reported LTV:CAC is 2.4. With a decaying curve, the real LTV might be €38, and your real ratio is 1.5 — which crosses the threshold most teams use to decide whether to scale a channel.
Per channel, if you're using LTV to make channel-level budget decisions. Paid-social cohorts, search cohorts, and organic cohorts often have meaningfully different decay shapes — averaging them hides the channel that's actually unprofitable.
For stores under €15M revenue, usually no. Geometric or two-phase decay gets you within 5-10% of a survival-analysis model and is auditable by your finance team. Reserve the heavier models for when retention modelling itself is a strategic priority.
Quarterly is enough for most stores. Refit sooner if you've changed your product mix, launched a subscription, or shifted acquisition channels meaningfully — all three change the underlying retention shape, not just the level.
Start tracking the metric that matters
Free for 30 days. No credit card. Connect your site in 4 minutes and see the one metric driving revenue.