3 May 2026
Most subscription businesses fail not because the product is bad, but because the founder cannot tell when the unit economics stop working. Revenue grows. Downloads grow. The deck looks better every quarter. Then the cash runs out and everyone is surprised. The cause is usually not visible in the headline numbers. It is hidden in six metrics that determine whether a subscription business can grow profitably or only grow.
Most founders track three of the six. The ones they leave out are usually the ones that decide the outcome.
Subscription is not a pricing model. It is a financial structure. A customer pays a small amount, repeatedly, over a long time. The business has to spend up front to acquire the customer and recover that spend slowly through ongoing payments. The whole question of whether the business works is whether you can afford the up-front cost and whether the customer stays long enough to pay it back, with margin to spare.
Six numbers describe whether that loop closes:
The first five are accounting questions. The last is a leading indicator that decides whether the first five will improve or get worse.
The total cost of getting one paying customer through the door. The denominator is paying customers, not signups, not downloads, not free users. The numerator is everything you spent to acquire them: paid ads, sales salaries, marketing tools, content production, referral incentives, the proportional cost of the people running these activities.
Two versions matter. Blended CAC divides total acquisition spend by total new paying customers (organic + paid). Paid CAC is the spend on a paid channel divided only by the customers acquired via that channel. The difference is large and consequential. A founder reporting blended CAC of $20 may have a paid CAC of $200 if 90% of customers came through word-of-mouth. The day you have to scale beyond word-of-mouth, paid CAC is what determines whether scaling works.
The trap most founders fall into: assuming CAC stays flat as you grow. It does not. The cheap channels saturate first. The first 1,000 customers from a single niche subreddit cost almost nothing. The next 10,000 require Meta and Google and a content team and a partnership manager. Brad Feld put it plainly in his term-sheet series at feld.com: the marginal cost of the next customer is almost always higher than the average cost of the previous one.
What kills you: CAC silently rising as you scale, while you are still pricing the product as if CAC were what it was at the beginning.
What one paying customer is worth, in net revenue, over their entire time as a customer. The standard formula is:
LTV = ARPU × gross margin × (1 / churn rate)
Each input matters. ARPU determines how much the customer pays per period. Gross margin determines how much of that you keep. Churn determines how many periods they stay.
The number that founders most often quote is gross LTV — ARPU divided by churn — without the gross margin term. That number is meaningless if your gross margin is 40%, because you only keep 40 cents of every revenue dollar. For a consumer subscription paying $10/month, with 4% monthly churn (a typical good number), and 70% gross margin after platform fees: LTV = $10 × 0.70 × (1 / 0.04) = $175.
David Skok, in his classic "SaaS Metrics 2.0" essay at forentrepreneurs.com, made the LTV/CAC ratio canonical. The folk-wisdom rule is LTV/CAC > 3. The rule has problems. It ignores time value of money, it does not account for gross-margin variability, and it assumes you can actually realize the lifetime — which you can only do if your forecast for churn holds. A 3:1 ratio is not a green light; it is a starting point for the harder questions about whether the lifetime is real and whether the churn rate will hold as you scale.
The percentage of paying customers who cancel in a given period. Two distinctions matter.
Gross vs. net. Gross churn is the percentage who leave. Net churn includes the offsetting effect of expansion revenue from existing customers (upgrades, seat additions, upsells). For B2B SaaS at scale, net churn can be negative — the cohort grows revenue over time even as some customers leave. For consumer apps with no upsell path, gross and net are nearly the same.
Monthly vs. annual. Monthly churn rates compound. 5% monthly churn is not "5% per year" — it is roughly 46% annual. 10% monthly churn is 72% annual. The number sounds small per month and ruinous per year. Most founders quote monthly because it sounds better. Buyers and investors should annualize on the spot.
The shape of churn matters as much as the rate. The cohort retention curve plots what percentage of a signup cohort is still paying after one month, three months, six months, twelve months. Healthy curves flatten — most of the churn happens early (people who signed up but were never going to stay), and the customers who survive the first few months become a stable base. Unhealthy curves keep declining linearly: every cohort loses customers at the same rate forever, and the business cannot accumulate a customer base.
For consumer subscription apps, month-1 churn of 40–70% is normal. By month 6, many cohorts are at 80–90% churn. This is the freemium-mobile reality and is the reason most consumer apps cannot sustain venture economics regardless of top-line growth. For B2B SaaS, the bar is much higher — most healthy companies show <2% monthly gross churn at scale.
The mean revenue you collect, per paying user, per period — net of platform fees, refunds, and discounts. The "net" matters.
A consumer app charging $10/month on iOS is not making $10. Apple takes 30% in year one, 15% from year two onward. After their cut, the net is closer to $7 in year one. After payment processing, refunds, and currency conversion, closer to $6.50. If you are forecasting LTV from a $10 ARPU number, you have already over-estimated the business by 35%.
Pricing structure shifts ARPU in non-obvious ways. Annual subscriptions usually carry a 15–25% discount versus the monthly equivalent. The customer pays less per year, you collect less per year, but churn is structurally lower because there is no monthly cancel decision. The trade is: lower headline ARPU, more durable LTV. Whether that is good or bad depends on the churn shape — if your monthly churn is high, the discount on annual is worth paying because it converts unstable revenue into stable revenue. If your monthly churn is already low, the discount is just margin you gave away.
Upsells and tier upgrades raise ARPU over time within a cohort, which is the single most powerful lever for offsetting churn. A B2B SaaS company with 5% monthly gross churn and 7% monthly expansion has 2% monthly net retention growth — the cohort is worth more next year than this year. Few consumer apps have this lever.
The number of months it takes for the gross profit from one customer to equal the CAC for that customer. The formula:
Payback = CAC / (ARPU × gross margin)
If CAC is $100, ARPU is $10/month, and gross margin is 70%: payback = $100 / $7 = 14.3 months.
Why payback matters: it determines whether you can afford to grow with paid acquisition. A business with 12-month payback can spend $1 to acquire a customer and have it back in cash within a year — paid acquisition is a reasonable use of investor capital. A business with 36-month payback is borrowing from the future to fund the present and will exhaust runway long before the cohort pays back.
The venture-capital benchmarks: B2B SaaS aims for <12 months. Consumer subscription tolerates longer paybacks (up to 24 months) when retention is exceptional. Anything beyond 24 months requires unusual retention or a financing structure that can carry the gap.
The reason payback period beats LTV/CAC for operational decisions: LTV is a forecast, payback is a measurement. You know your CAC, you know your ARPU and gross margin, you can compute payback today. LTV requires you to project churn into the future, which is the most uncertain input in the entire model.
The hardest of the six to measure and the only one that, if true, makes the other five easier. Andy Rachleff, who coined the term in the 2000s, defined it as the moment when "you have a product that customers genuinely want." Sean Ellis later operationalized it: ask your active users "how would you feel if you could no longer use this product?" If 40% or more answer "very disappointed," you are at or near PMF. (Sean Ellis, "The Product/Market Fit Question," startup-marketing.com.)
PMF is not a milestone you cross once. It is a regime that holds, narrows, or breaks as you scale. A company can have PMF in one customer segment and not in adjacent ones. A company can have PMF at one price and lose it when they raise prices. The question is not "do we have PMF" — it is "do we have PMF in the segment we are currently trying to grow."
Signals that suggest you are in PMF for a segment: organic growth in that segment without paid push, retention curves that flatten in that segment, NPS above 30–40 in that segment, repeat-usage rates that suggest the product is habitual.
Signals that you are not: paid acquisition growth without organic, retention curves that decline linearly, NPS clustered near zero or negative, churn that does not stabilize after the first few cohorts.
The reason PMF belongs in this list of metrics, even though it is qualitative: when PMF is real, CAC is lower (organic referral works), LTV is higher (retention is durable), churn is lower (the product is wanted), payback is shorter, and ARPU can rise over time without breaking conversion. When PMF is weak, you can spend your way to apparent growth, but the unit economics will betray you eventually.
The six are not independent. They are a system, and the system has a few defining couplings.
Bad churn destroys LTV, which destroys payback. If monthly churn doubles from 4% to 8%, LTV halves. CAC has to halve to keep the same payback. Halving CAC is hard. So bad churn forces a price increase, a positioning narrowing, or a slowdown in growth — and most founders try none of these and run out of cash.
Strong PMF makes acquisition cheaper. Customers who like the product talk about it. Word-of-mouth lowers blended CAC. The first 1,000 customers of a PMF business cost a fraction of the first 1,000 customers of a non-PMF business. This is why apparently identical companies can have wildly different unit economics — one has PMF in the segment, one is paying its way to a similar headline.
ARPU can mask both good news and bad. A business with low monthly churn but low ARPU may be uneconomical because LTV per customer is small. A business with high monthly churn but high ARPU may work if the customers who stay pay enough to cover the ones who leave. The numbers must be looked at together; in isolation each can mislead.
Payback is the operational test. It changes most directly with the things you can affect — pricing, gross margin, channel mix. Founders who want to fix unit economics usually need to fix payback first; LTV improvements come slower because churn moves slowly.
Three things.
First: if you can compute these six numbers from your own data right now, you are ahead of most pre-seed founders. If you cannot, the first investment in the business is instrumenting it so you can. RevenueCat, Amplitude, and Mixpanel make most of this trivial. There is no excuse for not knowing your churn cohort.
Second: be honest about which numbers you have and which you are guessing at. Pre-seed companies usually have CAC and ARPU and a guess at PMF; they almost never have a credible LTV because there is not enough cohort history to compute it. Saying "our LTV is $200" with three months of data is theatre. Saying "our month-1 retention is X, our month-3 retention is Y, and we are watching what happens at month-6 before claiming an LTV" is what we want to hear.
Third: the question is rarely "are these numbers good." It is "are they trending in the right direction, and what is the load-bearing assumption that has to hold for them to keep doing so." A pre-seed business with bad CAC today and a credible plan to fix it is more interesting than one with great CAC today and no understanding of why.
We disagree with the conventional founder wisdom that "you cannot measure PMF until you are bigger." You can. Most founders avoid the measurement because they are afraid of what it would say. The 40% question, asked of the 50–500 active users a typical pre-seed company has, gives you the answer.
If you are running a subscription business:
If you are evaluating one (as an investor, board member, or operator joining):
If you want a second read on the numbers, write to us at pitch@valaris.is.
Foundational SaaS metrics: David Skok, "SaaS Metrics 2.0" and the broader forentrepreneurs.com library, the canonical practitioner-level reference for SaaS unit economics.
Term-sheet and unit-economics context: Brad Feld and Jason Mendelson, Venture Deals: Be Smarter Than Your Lawyer and Venture Capitalist. 5th edition. Wiley, 2022. Also Feld's blog at feld.com.
Public-market subscription benchmarks: the 10-K filings of Spotify, Netflix, HubSpot, and Salesforce. Their disclosed retention curves, ARPU breakdowns, and CAC payback discussions are the highest-quality public data on subscription unit economics.
Product-market fit: Andy Rachleff, "The 6 Things All Entrepreneurs Need to Know About Product-Market Fit", First Round Review. Sean Ellis, "The Product/Market Fit Question", startup-marketing.com. Marc Andreessen, "The Only Thing That Matters", pmarchive.com.
Practitioner SaaS analysis: Tomasz Tunguz at tomtunguz.com, particularly his analyses of public-company NRR and CAC payback. Lenny Rachitsky's newsletter at lenny.lennysnewsletter.com for consumer-subscription practitioner data.
Consumer-app specifics: RevenueCat's annual State of Subscription Apps report — the largest dataset on consumer mobile subscription metrics, freely published.