How to Analyze Subscription Business Models: Churn, NRR, and Cohort Analysis

Subscriptions Are Everywhere — But Most Investors Analyze Them Wrong

The subscription economy is no longer a SaaS story. It has metastasized into virtually every sector of the economy. Netflix and Spotify pioneered media subscriptions. Peloton and ClassPass brought subscriptions to fitness. HelloFresh and Blue Apron attempted to subscription-ize groceries. Dollar Shave Club and Stitch Fix extended the model to consumer products. Adobe, Microsoft, and Autodesk converted their entire software businesses from perpetual licenses to recurring revenue. Even legacy industrials like Caterpillar and John Deere now generate meaningful subscription revenue through telematics and precision agriculture platforms.

Wall Street loves subscriptions because recurring revenue is predictable, which supports higher multiples. But “recurring” is a spectrum, not a binary. A Salesforce enterprise contract with three-year terms and 95% gross retention is recurring in a fundamentally different way than a Netflix subscription that can be cancelled with two clicks. An Adobe Creative Cloud license embedded in a professional photographer's daily workflow has different retention characteristics than a HelloFresh meal kit that competes with the grocery store, restaurants, and the customer's own cooking fatigue. Treating all subscription revenue as equally durable is one of the most common analytical errors in public market investing.

The framework that separates high-quality subscription businesses from those masquerading as one comes down to five metrics: net revenue retention, gross versus net churn, cohort retention curves, CAC payback period, and ARPU trajectory. Master these, and you can distinguish a compounding subscription machine from a leaky bucket that will eventually drain itself dry.

Net Revenue Retention: The Metric That Separates Elite From Average

Net revenue retention measures how much revenue a company generates from its existing customer base over a given period, after accounting for expansion, contraction, and churn. The formula is straightforward: take beginning-of-period revenue from a cohort, add expansion revenue from upsells, cross-sells, and price increases, subtract revenue lost to downgrades and cancellations, then divide by the beginning-of-period revenue.

An NRR of 100% means the existing customer base is spending exactly what it spent a year ago. Above 100% means the business is growing organically from within — existing customers are spending more. Below 100% means the business is shrinking from its installed base, and new customer acquisition must fill the hole before growth can even begin.

Why 120%+ NRR Is Exceptional

An NRR of 120% means existing customers collectively spend 20% more this year than last year. At that rate, even with zero new customer acquisition, revenue compounds at 20% annually. Snowflake posted NRR of 127% through most of its high-growth phase because customers' data workloads expanded as they migrated more analytics to the platform. CrowdStrike maintained NRR above 115% by selling additional security modules — the average enterprise customer went from using 3 modules at initial purchase to 7+ over time. Datadog achieved 125% NRR through a similar land-and-expand motion across observability, security, and developer tools.

What drives NRR above 120% is the combination of low gross churn and high expansion. The business must retain the vast majority of its customers (gross retention above 90%) while simultaneously growing revenue within those accounts through seat expansion, usage growth, tier upgrades, or cross-sell of additional products. This is why consumption-based pricing models — Snowflake, Datadog, Twilio — tend to produce the highest NRR: customer spending scales automatically with their underlying business growth without requiring a discrete upsell conversation.

The NRR Benchmarks by Business Model

Gross Churn vs. Net Churn: The Metrics Companies Don't Want You to Compare

Gross churn measures the total revenue or customers lost in a period, with no offset for expansion. Net churn subtracts expansion revenue from the losses, producing a smaller (and often negative) number. The gap between these two metrics tells you something profoundly important about a subscription business: how much heavy lifting expansion revenue is doing to mask underlying customer attrition.

Consider a SaaS company that starts the year with $100 million in ARR from 1,000 customers. During the year, 120 customers churn, taking $8 million in ARR with them. Another 80 customers downgrade, reducing their spending by $3 million. That is $11 million in gross revenue churn — an 11% gross churn rate. But the 800 remaining customers expanded their spending by $18 million through seat additions, tier upgrades, and new product adoption. Net revenue churn is negative 7% — meaning the base actually grew by 7%, which translates to 107% NRR. The headline NRR looks healthy. But gross churn of 11% means the company replaces its entire customer base roughly every nine years, and it must maintain that $18 million expansion pace just to stay above 100% NRR.

This is exactly the pattern Salesforce exhibited through much of its growth phase. Annual logo churn ran 8–10% for years, but NRR stayed above 110% because enterprise customers continuously expanded their Salesforce footprint. The model worked because Salesforce's expansion engine — additional clouds, seats, and higher-tier editions — was genuinely powerful. But when expansion revenue decelerated in 2023–2024 as enterprise IT budgets tightened, the underlying gross churn became visible and the stock underperformed. The expansion was masking, not solving, the churn problem.

Dollar-Based vs. Logo-Based Retention

Dollar-based retention weights each customer by their revenue contribution, so a $500K enterprise account churning matters more than ten $5K SMB accounts. Logo-based retention treats every customer equally. Most companies prefer to report dollar-based retention because it is almost always higher — large customers churn less frequently and expand more aggressively than small ones.

The analytical value comes from comparing the two. If dollar-based retention is 115% but logo-based retention is 82%, the company is losing nearly one-fifth of its customers annually while extracting enough additional revenue from survivors to post a healthy headline number. This creates dangerous customer concentration: revenue becomes increasingly dependent on fewer, larger accounts. If even one or two of those anchor accounts churn in a single quarter, the financial impact is outsized. Adobe's Creative Cloud avoids this risk by maintaining both high dollar retention (above 95%) and high logo retention (above 90%) — a sign that the product retains customers across every segment, not just the largest ones. For more on evaluating how acquisition costs relate to retention dynamics, see our analysis of customer acquisition cost and LTV.

Cohort Analysis: Reading the Retention Curve

If NRR tells you how the overall business is performing, cohort analysis tells you why. A cohort groups customers by their acquisition date — typically quarterly or annually — and tracks their behavior over subsequent periods. The resulting retention curve is the most honest representation of a subscription business's quality, because it cannot be manipulated by mixing older, stickier customers with newer, less proven ones.

The Three Shapes of Retention Curves

The “smile curve” is the ideal shape. Revenue from a cohort initially declines as poor-fit customers churn in the first three to six months, then stabilizes, and eventually curves upward as surviving customers expand their spending. This pattern is visible in the best enterprise SaaS companies — CrowdStrike, Datadog, and Snowflake all exhibit smile curves because the customers who survive the initial period become deeply embedded in the platform and expand usage over time. A cohort acquired at $1 million in ARR might dip to $900K at month six, stabilize at $880K by month twelve, then climb to $1.1 million by month twenty-four through module adoption and seat expansion.

The “flat curve” is acceptable. Revenue drops in the early months, then flattens at a stable retention level. Netflix exhibits this pattern: a cohort loses roughly 15% of subscribers in the first three months as trial users and impulse sign-ups drop off, then monthly churn stabilizes at 2–3% for the remainder. There is no meaningful expansion revenue per subscriber beyond periodic price increases, so the curve does not smile upward, but the stabilization indicates genuine product-market fit among the customers who survive the initial screening period.

The “decay curve” is the warning sign. Revenue from a cohort declines continuously without ever stabilizing. This pattern was visible in Blue Apron's cohort data: customers churned at a roughly constant rate of 10–12% per quarter regardless of tenure. A subscriber who had been receiving meal kits for two years was just as likely to cancel as one who had been receiving them for two months. The product never became embedded in the customer's routine, which meant every cohort was on a predictable path toward zero. HelloFresh exhibited a somewhat less severe version of the same pattern, though its retention improved modestly in later years through menu diversification and pricing optimization.

Cross-Cohort Comparison

Comparing curves across successive cohorts reveals whether the product is improving or degrading. If the Q1 2025 cohort retains better than the Q1 2024 cohort at the same tenure, the business is getting stronger — the product is improving, the ICP is better defined, or the onboarding process has been refined. Spotify demonstrates this pattern: each successive annual cohort has shown marginally better retention than the prior one, driven by algorithmic playlist improvements, exclusive podcast content, and the social network effects of shared playlists. If newer cohorts retain worse, the company may be acquiring increasingly marginal customers as it scales beyond its natural addressable market, or competitive alternatives may be eroding the product's differentiation.

CAC Payback and LTV/CAC: The Unit Economics of Subscriber Acquisition

Every subscription business faces the same fundamental question: how much can I spend to acquire a subscriber and still generate a positive return? The answer depends on two metrics that interact in critical ways: CAC payback period and the LTV/CAC ratio.

CAC Payback by Channel

CAC payback measures how many months of gross profit it takes to recover the cost of acquiring a subscriber. The formula is: CAC / (Monthly ARPU × Gross Margin). If acquisition cost is $150, monthly ARPU is $15, and gross margin is 65%, monthly gross profit per subscriber is $9.75, and payback is 15.4 months. But this blended calculation obscures massive variation by acquisition channel.

Netflix's organic subscribers — those who sign up through word of mouth, social media buzz around a hit show, or direct navigation — have near-zero acquisition cost. Their payback is essentially one month of gross profit. Netflix's paid acquisition subscribers, acquired through performance marketing, CTV ads, and partnership bundles (like the T-Mobile integration), carry meaningful CAC. Historically, Netflix spent $50–$80 per paid subscriber in the U.S. market, implying payback of 6–10 months at approximately $8 in monthly gross profit per subscriber. Spotify's paid search subscribers had a CAC of roughly $15–$25 with payback of 4–7 months, while partnership channel subscribers (telecom bundles) had higher acquisition costs but also higher retention because the subscription was embedded in the customer's phone bill.

The channel-level distinction matters enormously for assessing growth sustainability. A company reporting 8-month blended payback might have 3-month payback on organic channels and 18-month payback on paid channels. If organic acquisition plateaus as brand awareness saturates, the marginal payback deteriorates rapidly. Dollar Shave Club experienced exactly this: early viral growth produced spectacularly low blended CAC, but as the viral effect faded and the company shifted to television and digital advertising, blended CAC more than doubled and unit economics collapsed.

LTV/CAC: 3x Minimum, 5x Ideal

The lifetime value to customer acquisition cost ratio measures total expected gross profit from a subscriber over their lifetime divided by the cost to acquire them. The 3x minimum threshold exists because after accounting for overhead costs not captured in gross margin — R&D, G&A, infrastructure — a business needs approximately 3x gross profit recovery just to break even on a fully loaded basis. At 5x+, the business generates substantial economic profit per subscriber, funding reinvestment and delivering returns to shareholders.

Expansion Revenue vs. New Logo Revenue: Where Growth Actually Comes From

One of the most revealing analyses you can perform on a subscription business is decomposing total revenue growth into its two components: expansion revenue from existing customers and revenue from new customer acquisition. The mix between these two sources tells you about the durability and capital efficiency of growth.

Expansion-driven growth is more valuable per dollar because it carries near-zero incremental acquisition cost. When a Salesforce customer adds 200 seats to their existing deployment, or when a Snowflake customer's query volume increases 40% because they migrated more workloads to the platform, that revenue growth requires no sales commission, no marketing spend, and no onboarding investment. The incremental margin on expansion revenue is substantially higher than on new-logo revenue. This is why companies with high NRR — where a large portion of growth comes from expansion — tend to have structurally better margins than companies of similar scale that rely primarily on new customer acquisition.

Datadog provides an instructive example. In its early years, roughly 60–70% of incremental ARR came from new logos. By FY2024, the mix had shifted: approximately 50% of incremental ARR came from expansion within existing accounts, largely driven by customers adopting additional products from Datadog's expanding suite. This shift corresponded with meaningful margin expansion because the company was generating more revenue per sales and marketing dollar. Adobe's transition to Creative Cloud followed a similar pattern — once the installed base reached critical mass, expansion revenue from price increases, tier upgrades, and new product bundles (adding Firefly AI, Adobe Express) drove an increasing share of growth at minimal incremental cost.

Conversely, businesses where growth comes almost entirely from new logos face a perpetual cost escalation problem. Every dollar of growth requires another dollar of S&M investment. Netflix, despite its scale, still derives the majority of its revenue growth from new subscriber additions in international markets and the ad-supported tier, not from expansion within existing subscribers. This is because Netflix has limited expansion levers beyond periodic price increases — there is no usage-based upside, no module cross-sell, and no seat expansion. The economics still work because Netflix's brand drives relatively low CAC, but it is a structurally different growth profile than a company like Adobe where the existing base compounds organically.

ARPU Trends, Downgrade Risk, and Pricing Power

Average revenue per user (ARPU) is the pulse rate of a subscription business. Rising ARPU indicates pricing power, successful upsells, or favorable mix shift toward higher-value customers. Declining ARPU signals commoditization, competitive discounting, or a shift toward lower-value segments that may grow subscriber counts but dilute unit economics.

ARPU Expansion: The Pricing Power Signal

Spotify raised its U.S. Premium individual price from $9.99 to $10.99 in mid-2023, then to $11.99 in mid-2024. Monthly churn increased by an estimated 20–30 basis points temporarily but reverted within two quarters. The net effect was a roughly 20% ARPU increase with minimal subscriber loss — textbook pricing power. Netflix executed a similar playbook, raising U.S. Standard plan pricing from $13.99 to $15.49 to $17.99 over consecutive years while growing total subscribers. The ability to raise prices without triggering material churn acceleration is the strongest evidence that a subscription product is genuinely essential to the customer rather than merely convenient.

Adobe demonstrates the extreme end of pricing power. Creative Cloud All Apps pricing has increased from $49.99/month at launch in 2013 to $59.99/month currently, with annual churn remaining below 10%. Creative professionals have no viable alternative for the integrated Photoshop/Illustrator/Premiere Pro workflow, which gives Adobe near-monopoly pricing power in its core segment. This is the kind of structural advantage that supports LTV/CAC ratios of 8x or higher.

The Downgrade Risk: When ARPU Compresses

ARPU compression is the canary in the coal mine for subscription businesses. It often appears before headline churn metrics deteriorate because customers downgrade to cheaper tiers before cancelling outright. Peloton saw this clearly: as engagement declined among pandemic-era subscribers, many downgraded from the full $44/month All-Access membership to the $12.99 app-only tier before eventually churning. The result was ARPU declining from $39 to under $30 while logo retention appeared temporarily stable — a misleading combination that masked the true rate of customer disengagement.

Netflix's introduction of the ad-supported tier at $6.99/month represented a calculated ARPU trade-off. On one hand, it attracted price-sensitive subscribers who would not pay $15.49+ for Standard, expanding the addressable market. On the other, existing Standard and Premium subscribers who downgraded to the ad tier compressed ARPU. Netflix managed this transition successfully because advertising revenue per ad-tier subscriber approximately offsets the subscription revenue difference — total ARPU including ad revenue for the ad tier is roughly comparable to the old Basic tier. But investors should watch the mix closely. If the ad tier grows to 50%+ of the subscriber base while ad revenue per subscriber stagnates, Netflix's blended ARPU will compress in ways that the current growth narrative does not account for.

When Subscriptions Become Commoditized: The Churn Spiral

Not all subscription businesses compound. Some enter a death spiral where rising competition erodes differentiation, commoditizes the offering, and triggers a self-reinforcing cycle of increasing churn and deteriorating economics. Understanding when a subscription business is entering this spiral is one of the most important skills for public market investors.

The spiral follows a predictable sequence. First, competitors launch similar products at lower prices or with free tiers. The streaming wars illustrate this vividly — by 2024, consumers could choose from Netflix, Disney+, Max, Peacock, Paramount+, Apple TV+, and Amazon Prime Video, all offering broadly similar content categories at $8–$18/month. Second, the incumbent responds with promotional pricing, discounts, or bundled offers to defend subscriber counts. Third, ARPU compresses as both new and existing subscribers access lower price points. Fourth, compressed ARPU reduces LTV, making customer acquisition less profitable at the same CAC. Fifth, the company either cuts marketing spend (accepting slower growth) or maintains spend at deteriorating unit economics (accelerating cash burn). Neither outcome is good for shareholders.

Dollar Shave Club is the cautionary tale. Launched in 2012 with a viral video that generated millions of subscribers at near-zero CAC, the company looked like a subscription juggernaut. Unilever acquired it for $1 billion in 2016. But the product — razor blades — had zero switching costs. Harry's launched an identical model. Gillette responded with its own direct-to-consumer offering. Amazon entered with its Solimo brand. Dollar Shave Club's CAC rose as it shifted from organic virality to paid acquisition, while churn accelerated as customers discovered they could buy equivalent blades cheaper from multiple competitors. By 2023, Unilever wrote down the acquisition and restructured the business. The subscription model had temporarily obscured what was fundamentally a commodity product with no defensible moat.

The defense against commoditization is switching costs. Adobe's Creative Cloud survives because migrating years of Photoshop files, Illustrator templates, and Premiere Pro projects to a competitor is prohibitively painful. Salesforce survives because CRM data, custom workflows, and third-party integrations create deep lock-in. Spotify's playlists, listening history, and algorithmic recommendations create moderate switching costs that improve with usage. Blue Apron had no switching costs whatsoever — switching from one meal kit to another required changing nothing about the customer's kitchen, equipment, or skills. When evaluating any subscription business, ask: what would the customer lose by cancelling and switching to an alternative? If the answer is “nothing meaningful,” the subscription premium is borrowed time. For more on how to evaluate competitive moats and switching costs, see our framework on analyzing competitive moats.

Putting It Together: A Subscription Business Evaluation Framework

The best subscription businesses share a consistent set of characteristics that you can identify from public filings, earnings calls, and investor presentations. NRR above 110% for B2B or above 100% for B2C. Gross retention above 85% for B2B or above 75% for B2C. Cohort retention curves that flatten or curve upward after the initial churn period. LTV/CAC above 3x on a fully loaded, capped-LTV basis. CAC payback under 18 months. ARPU trending upward through pricing power and mix improvement. Expansion revenue representing a growing share of incremental growth. And meaningful switching costs that make cancellation painful rather than merely inconvenient.

Applying this framework would have identified Adobe, Salesforce, and Netflix as compounding subscription machines years before their stock prices fully reflected the quality. It would have flagged Blue Apron, Dollar Shave Club, and Peloton as structurally challenged long before the market capitulation. The metrics are not hidden — they are disclosed in filings and discussed on earnings calls. The edge comes from tracking them systematically, cohort by cohort and quarter by quarter, rather than relying on headline subscriber counts and revenue growth rates that obscure more than they reveal.

The subscription economy is not slowing down. But the companies that will create lasting shareholder value are those where the subscription model reflects genuine product stickiness and customer lock-in, not those where “subscription” is simply a billing mechanism layered on top of a commodity product. The metrics in this guide are your toolkit for telling the difference.

Frequently Asked Questions