LTV:CAC is the lifetime value of a paying subscriber divided by what it cost to acquire them. For a subscription app it is a useful summary and a dangerous operating metric. It is useful because it asks the right question: is a subscriber worth more than they cost? It is dangerous because the answer depends on three choices the ratio hides: how far into the future you count revenue, whether that revenue is net of store fees and tax, and which cost you divide by. Change those choices and the same cohort of users can report anything from 1.0x to more than 10x. Our view: report LTV:CAC to investors with its assumptions attached, and run paid acquisition on realised value, payback and marginal cost.
Below is the formula, a worked example that produces six different ratios from one hypothetical cohort, the case investors make for the metric, the case against it set out on RevenueCat's blog, and the set of numbers we use instead when real Meta budget is being allocated. Every number in the worked example is hypothetical and labelled as such. Published figures come from RevenueCat's State of Subscription Apps 2026, with definitions checked against the report text.
The LTV:CAC formula for a subscription app
LTV:CAC = net value of one paying subscriber over horizon H ÷ cost to acquire one paying subscriber
Textbook LTV = net revenue per period ÷ churn rate per period
Subscriber CAC = acquisition spend ÷ new paying subscribers from that spend
Each term needs a decision before the ratio means anything.
- Horizon. Is "lifetime" twelve months of realised revenue, three years of forecast revenue, or the infinite sum the textbook formula implies? Nothing in the ratio tells the reader.
- Netting. Revenue has to be what reaches you after store commission, VAT and refunds. Most subscription dashboards report a gross figure by default.
- Which CAC. Paid media only, blended with free organic installs, or fully loaded with creative production and agency fees? We set out the five versions of CAC and when each is appropriate in our guide to calculating subscription app CAC.
- Same unit on both sides. LTV per paying subscriber must be divided by cost per paying subscriber, not by cost per install or cost per trial. This sounds obvious and is one of the most common errors we see in founder decks, because the cost the ad platform reports is usually per install or per trial.
The textbook formula has a specific weakness for subscription apps. It assumes a constant churn rate, and subscription retention curves are steep early and flatter later. Take a hypothetical monthly plan netting £5.65 per payment. Plug in a 45% monthly churn rate, typical of the first renewal, and LTV is £12.56. Plug in 15%, closer to what survivors show months later, and it is £37.68. Same app, same users, three times the LTV, depending only on which month's churn you chose to divide by.
One cohort, six LTV:CACs: a worked example
The following app is hypothetical. It is a UK app selling an annual plan through the App Store, buying installs on Meta, with a 7 day free trial. These are not client figures and are not offered as benchmarks. They are chosen so the arithmetic is easy to follow.
| Input (hypothetical) | Value | Note |
|---|---|---|
| CPI on Meta | £1.50 | Cost per install |
| Install-to-trial | 12% | Share of installs starting a 7 day free trial |
| Trial-to-paid | 35% | So 4.2% of installs become subscribers |
| Cost per trial | £12.50 | £1.50 ÷ 12% |
| Paid media CAC per subscriber | £35.71 | £1.50 ÷ 4.2% |
| Fully loaded CAC per subscriber | £44.64 | Media plus 25% for creative production and fees |
| Annual price (UK, incl. 20% VAT) | £79.99 | £66.66 before VAT |
| Refunds on each payment | 3% | Applied to every billing period for simplicity |
| Annual renewal rate | 30% observed, 40% forecast | The forecast assumes 40% renew every year |
For context on the renewal input, RevenueCat puts median first renewal rates for annual plans between 23% and 40% depending on category, so 30% observed is mid-range and a 40% forecast sits at the top of the median range. The gross first-year payment is £79.99 less refunds, £77.59. The net first-year payment, after VAT, Apple's 30% first-year commission and refunds, is £45.26. Renewals net £54.96, because Apple's share drops to 15% after a subscriber accumulates a year of paid service.
Now calculate LTV:CAC six ways.
| Version | Value | Cost | Ratio | What is wrong with it |
|---|---|---|---|---|
| 1. Forecast gross LTV ÷ cost per trial | £129.32 | £12.50 | 10.3x | Divides a per-subscriber value by a per-trial cost. Wrong denominator. |
| 2. Forecast gross LTV ÷ paid media CAC | £129.32 | £35.71 | 3.6x | Clears 3:1. Still counts VAT and Apple's commission as yours. |
| 3. Forecast net LTV ÷ paid media CAC | £81.90 | £35.71 | 2.3x | Netted correctly, but ignores creative and fees. |
| 4. Forecast net LTV ÷ fully loaded CAC | £81.90 | £44.64 | 1.8x | Honest about cost. 45% of the value is renewals that have not happened. |
| 5. Realised net value at month 13 ÷ fully loaded CAC | £61.75 | £44.64 | 1.4x | Cash actually collected after the first renewal at the observed 30%. |
| 6. Realised net value before the first renewal ÷ fully loaded CAC | £45.26 | £44.64 | 1.0x | What the business has in the bank for the first eleven months. |
The forecast LTVs use the standard geometric sum: a first payment, then 40% renewing every year after that, so the renewals are worth 0.4 ÷ 0.6 of one renewal payment. None of the six calculations contains an arithmetic mistake. Each is a defensible answer to a slightly different question, and a founder could put any of them on a slide titled "LTV:CAC".
Three things stand out.
- The version that clears 3:1 is the one with the most assumptions in it. Row 2 counts revenue the business never receives and renewals that have not happened, and ignores the cost of the creative that bought the installs.
- Netting alone removes over a third of the ratio. Moving from row 2 to row 3 changes nothing about the customers. It only stops counting VAT and Apple's commission as revenue.
- The honest year-one answer is 1.0x. This app gets its money back almost exactly, and everything above that depends on renewals. That is not a bad business. It is a business whose growth is funded by the second year, which is a cash-flow fact the 3.6x hides completely.
Why investors like LTV:CAC
The case for the metric deserves a fair hearing, because it is not a stupid metric. It became standard in B2B SaaS, where David Skok's SaaS metrics guide observed that the best SaaS businesses have an LTV to CAC ratio higher than 3, alongside a second guideline that they recover CAC within about a year. Skok was explicit that these are guidelines, not laws.
Investors use it for good reasons:
- It compresses unit economics into one number that can be compared across a portfolio.
- It forces the company to state what a customer is worth, which many early teams have never written down.
- It answers the question that matters most to someone funding growth: if we give you more money for acquisition, will each customer return more than they cost?
- In diligence it is normally rebuilt from raw cohort data, so the definitions get tested, which is the opposite of how it is used inside many companies.
Used that way, with a stated horizon and an analyst checking the inputs, LTV:CAC is a reasonable viability test. The problem starts when the same number moves from the board deck into the ad account.
Why operators misuse it: the case against
The sharpest version of the argument against appeared on RevenueCat's blog in an article by Nathan Hudson, founder of Perceptycs, titled Stop focusing on LTV to CAC. His core point is that customer lifetime in a subscription app can only be known retrospectively, and that churn in consumer subscriptions is often not final: seasonal users cancel and come back, others pause, try a competitor and return. If lifetime is unknowable while the customer is still paying, lifetime value is an estimate dressed as a fact.
RevenueCat's 2026 report supports the "churn is not always churn" part with data. It finds that 20% of monthly plan churners reactivate within a year, against 9% for weekly plans and 5% for annual plans. A textbook LTV that treats every cancellation as final will understate some apps and, where the churn input is taken from a good month, overstate others.
Hudson's alternative is to stop estimating lifetime and take snapshots of what customers have actually paid at fixed ages, which he calls periodic ARPPU snapshots and others call realised LTV, at points such as day 0, 7, 30, 90 and 365. From those he tracks ARPPU:CAC, payback period and gross contribution after CAC. We agree with the direction. Where we would add something is on the operating side, which is the next section.
Four further problems show up specifically in paid acquisition.
1. The ratio has no clock
A 3:1 ratio earned over three months and a 3:1 ratio earned over three years are different businesses. The first can fund its own growth. The second needs a lot of cash to survive long enough to find out whether the forecast was right. Payback period is the missing clock, and it is why we treat it as the binding constraint on Meta spend for most apps.
2. Forecast retention is the most fragile input
RevenueCat reports median year-one retention of 20% to 40% for annual plans, 6% to 14% for monthly plans and 1% to 2% for weekly plans, with the range set by category. It also reports that year-one retention fell slightly between the 2023 and 2024 cohorts, from 31% to 28% for annual plans and 10% to 8% for monthly. A forecast LTV built on last year's renewal rate inherits whatever has changed since, and most of the forecast value sits in the years with the least evidence behind them. In our worked example, 45% of the forecast net LTV arrives after year one.
3. Dashboard LTV is usually gross
RevenueCat's own documentation says its realised LTV is calculated from cohort revenue minus refunds, and therefore still includes the commission, taxes and fees the stores deduct before paying you. Dividing that by a cash CAC overstates the ratio. How much depends on market and store:
| Situation | Developer receives | Calculation |
|---|---|---|
| UK price including 20% VAT, App Store, first year | About 58% | 70% of the price before VAT (0.70 ÷ 1.20) |
| Same, App Store, after a year of paid service | About 71% | 85% ÷ 1.20 |
| UK price including 20% VAT, Google Play subscription | About 71% | 85% (a 10% service fee plus a 5% billing fee) ÷ 1.20 |
| EU storefront, App Store In-App Purchase, first year, from 1 October 2026 | 74% of the price before VAT | 26% commission; 15% for auto-renewing subscriptions after their first year |
The UK and US App Store rates come from Apple's subscription documentation. The EU row reflects Apple's new EU business terms, which took effect on 1 October 2026 and set a 26% commission for App Store apps using In-App Purchase, falling to 15% for auto-renewing subscriptions after their first year. VAT rates vary by EU country, so net the price before applying the commission. The full netting walk-through, applied to install costs, is in our guide to break-even CPI for subscription apps.
4. An average ratio rewards underspending
LTV:CAC is usually reported as an average across all spend, and averages are highest when budgets are small, because the first pounds buy the cheapest subscribers. Extend the hypothetical example. At £20,000 a month, fully loaded CAC is £44.64 and the app buys about 448 subscribers. Raise spend to £30,000 and suppose the extra £10,000 buys subscribers at a marginal £70 each, about 143 of them. Average CAC rises to £50.77 and the forecast net ratio falls from 1.8x to 1.6x.
A team managing to the ratio would cut the extra spend. Yet on the forecast those 143 subscribers are each worth £81.90 against a £70 cost, about £1,700 of extra contribution. On realised year-one cash, each loses £24.74 until they renew. So the ratio says no, the forecast says yes eventually and the bank balance says not this year. All three are true. Only a payback window that the business has chosen and can fund turns that into a decision.
What we track instead
Our position, after presenting both sides: LTV:CAC is a fair question and a poor dial. We do not stop clients reporting it. We stop them steering by it. Different questions need different numbers, refreshed at different speeds:
| Question | Number that answers it | How often |
|---|---|---|
| Is this business viable in principle? | LTV:CAC, labelled with horizon, netting and CAC type | Quarterly, for the board and investors |
| Can we afford to wait for the money? | CAC payback month on realised net revenue | Monthly, by cohort |
| Is this cohort better or worse than the last one? | Realised net value per subscriber ÷ CAC at fixed ages (day 30, 90, 365) | Monthly, compared at the same age |
| Should the next £10k go in? | Marginal CAC against value within the payback window | Whenever budget changes |
| Is this week's creative working? | Cost per trial or per subscriber against a target derived from the above | Daily to weekly, inside the ad account |
The operating layer at the bottom of that table is where most of the budget is decided, and it is the layer LTV:CAC cannot reach. A new creative concept has no lifetime. It has three days of cost per trial and, a week later, the first read on trial-to-paid. We derive a cost-per-trial and cost-per-subscriber target from realised net value and the payback window, then judge creative against that target. Which of those costs to act on, and when, is covered in CPI vs CPA vs CAC for subscription apps, and the earliest return checkpoint in our guide to day 7 ROAS.
The comparison at the middle of the table matters more than it looks. Comparing this month's cohort with last month's at the same age, on realised net value per subscriber, is the fastest honest signal that acquisition quality has changed. If a new set of ads lowers CAC by attracting subscribers who renew less, the ratio will look better for months before it looks worse. Realised value at day 30 and day 90 shows it much sooner.
Improving LTV:CAC without touching media
Because the ratio has two sides, the biggest improvements often do not come from the ad account. Install-to-subscriber conversion lowers CAC directly at the same CPI, and it is usually owned by product rather than media. Our Steps & Beasts case study is a public example: alongside heavy creative testing and AI UGC, onboarding changes lifted install-to-subscriber conversion, and revenue rose 145% with active subscriptions up 118%. The case study does not publish an LTV:CAC and we will not invent one. What it shows is that the cost side of the ratio responds to work done after the install, not only to bids and audiences.
The value side responds to plan mix, pricing and the quality of users the ads attract. The benchmarks for the first two conversion steps are in our trial-to-paid benchmarks and download-to-paid benchmarks.
How to report LTV:CAC honestly
If you have to put the ratio in a deck, and many founders do, make it impossible to misread. A ratio reported with these six labels is a useful number. Without them, it is an opinion.
- Horizon: realised to a stated age, or forecast to a stated number of years.
- Netting: after store commission, VAT and refunds, with the store mix stated.
- CAC type: paid media, blended, or fully loaded with creative and fees.
- Unit: per paying subscriber on both sides.
- Cohorts: which acquisition months, platforms and markets are included.
- Payback month: shown next to the ratio, always.
Frequently asked questions
What is LTV:CAC for a subscription app?
LTV:CAC is the lifetime value of a paying subscriber divided by the cost of acquiring one. For a subscription app, lifetime value is the net revenue a subscriber produces across their initial payment and every renewal, and CAC is the acquisition spend divided by the number of new paying subscribers it produced. A ratio above 1 means a subscriber is expected to repay what they cost. The ratio says nothing about how long that takes.
What is a good LTV:CAC ratio for a subscription app?
The widely quoted 3:1 comes from B2B SaaS. David Skok's SaaS metrics guide says the best SaaS businesses have an LTV to CAC ratio higher than 3, and stresses these are only guidelines. We are not aware of a published benchmark built specifically for consumer subscription apps, and a ratio is only comparable when the horizon, the netting and the type of CAC are the same. For a subscription app, a realised net ratio above 1 at twelve months is a more useful test than a predicted ratio of 3.
Should I use predicted LTV or realised LTV?
Use realised value for operating decisions and predicted value only for planning, with the assumptions written down. Realised LTV is the revenue a cohort has actually produced by a given age, such as day 30 or day 365. Predicted LTV adds renewals that have not happened yet. In a typical annual-plan app, a large share of predicted value sits beyond year one, so a predicted ratio can look healthy while the cash has not arrived.
Is LTV:CAC or CAC payback period more important?
They answer different questions. LTV:CAC asks whether a subscriber is worth more than they cost over their whole life. Payback period asks how long the business has to fund the gap before the money comes back. A subscription app can have an attractive ratio and a payback period it cannot afford, so for spend decisions payback and realised value at fixed cohort ages are usually the binding constraints.
Why does my RevenueCat LTV make LTV:CAC look better than it is?
RevenueCat's documentation says its realised LTV is calculated from cohort revenue minus refunds, and therefore still includes the store commission, taxes and fees the stores deduct before paying you. For a UK App Store subscriber in their first year, where the price includes 20% VAT and Apple keeps 30%, the developer receives about 58% of that figure. Divide a gross LTV by a real cash CAC and the ratio is overstated by around 70% in that case.
Can an LTV:CAC ratio be too high?
Yes. An average ratio is usually highest at low spend, because the first pounds buy the cheapest subscribers. A very high ratio can mean you are leaving profitable growth unbought. The question for the next block of budget is whether its marginal CAC, the cost of the extra subscribers it buys, is still below the value those subscribers produce within a payback window you can fund.
The short version
LTV:CAC is not wrong. It is incomplete. It tells you whether a subscriber should eventually be worth more than they cost, and nothing about when, net of what, or at what scale. In our hypothetical cohort the same users produced ratios from 1.0x to 10.3x without a single arithmetic error. Investors are right to ask for it. Operators are wrong to steer by it. Run paid acquisition on realised net value at fixed ages, a payback window you can fund and the marginal cost of the next block of spend, and let LTV:CAC be the summary that falls out of those, not the number that drives them.
For where these numbers sit in the wider growth system, read our subscription app marketing strategy guide. If you want a team to run Meta acquisition against targets built this way, our consumer apps page explains how we work with subscription apps, and if your app is already spending, our pricing is published.