Subscriptions Will Survive in Exactly Two Places

There is a particular sound every subscription makes on your credit card statement. A low, recurring drone — steady, patient, indifferent. It doesn't start when you use the product. It doesn't stop when you don't. It just charges. Month after month. Whether you opened the app once or a thousand times.

For the last decade, this was the sound of the greatest business model innovation in software history. Recurring revenue. Predictable cash flows. The magic metric that turned one-time sales into lifetime value. The entire SaaS industry — and most of the consumer software industry — was built on the premise that subscriptions are superior to every other pricing model.

That premise is breaking.

RevenueCat's 2026 State of Subscription Apps report, covering more than 115,000 apps and $16 billion in revenue, shows that median renewal rates are declining. Enterprise procurement teams are actively consolidating subscriptions and pushing back on recurring costs. Consumer surveys consistently show that "too many subscriptions" is now a top-three financial concern alongside rent and groceries.

The backlash isn't against paying for software. It's against paying for software you're not using. And the data suggests that subscriptions — the universal pricing model of the 2010s and early 2020s — will survive in exactly two places.

Where Subscriptions Work

Subscriptions are structurally sound when two conditions are met simultaneously:

Condition 1: The value is continuous. The product delivers value every day, not intermittently. Disconnecting from the product means losing access to something the customer uses constantly.

Condition 2: The value is indispensable. The product isn't optional. Canceling isn't "I'll miss this." It's "things stop working."

Only two categories consistently meet both conditions:

Category 1: Continuously Refreshed Content Platforms

Netflix, Spotify, The New York Times, Bloomberg Terminal. These products charge for access to a catalog that is continuously updated. The value proposition isn't "use this tool" — it's "access this library." The library changes every day. If you cancel, you lose access to new content.

This model works because the content is the product, and the content is always new. A Netflix subscriber who doesn't watch for a month still has a reason to resubscribe: there's new content they haven't seen. The library refreshes independently of the user's behavior.

The structural requirements: you must continuously produce or license new content at a pace that justifies ongoing payment. This is why subscription models work for streaming services and news publications but fail for most content creators — an individual creator cannot refresh their catalog fast enough to justify a monthly charge.

Category 2: Always-On Infrastructure and Platform Tools

AWS, Slack, Okta, Datadog. These products are always on. They run in the background. They are wired into the customer's operations. Canceling doesn't mean "I'll miss a feature." It means "my servers go down," "my team can't communicate," or "my employees can't log in."

This model works because the product is infrastructure. It's not a tool you choose to use — it's a system that must be running. The subscription isn't paying for access to features. It's paying for continuous operation of something the business depends on.

The structural requirement: the product must be woven into the customer's operations deeply enough that removing it requires significant effort. This is why Slack can charge per seat indefinitely — removing Slack means migrating years of conversation history, rebuilding integrations, and retraining the entire organization.

Where Subscriptions Are Dying

Everything outside these two categories is experiencing subscription decay — the gradual erosion of renewal rates, increasing price sensitivity, and growing willingness to cancel and switch.

Design and Creative Tools

Adobe switched to subscriptions in 2013 with Creative Cloud. For a decade, it worked — designers needed Photoshop and Illustrator daily, and the switching cost to alternatives was high.

In 2026, the landscape is different. Figma dominates UI design with a freemium model. Canva serves most non-professional design needs for free. AI image generation tools (Midjourney, Flux, DALL-E) produce outputs that previously required hours in Photoshop. The continuous, daily-use justification for a $55/month Adobe subscription is eroding.

The tell: Adobe's net-new subscriber growth has decelerated in every recent quarter. The installed base is large, but the growth is increasingly driven by price increases on existing subscribers, not new adoption. This is the classic late-stage subscription pattern — squeezing existing customers because new ones aren't arriving.

Productivity and Project Management

Notion, Asana, Monday.com, ClickUp. These tools charge monthly subscriptions for project management and productivity features. The problem: AI is making the core features — task management, document creation, note-taking — trivially reproducible.

ChatGPT can manage a project plan. Claude can generate a product requirements document. A Lovable-built internal tool can replace most of what Monday.com does for a specific team. The unique value of these platforms is declining as AI makes the underlying capabilities generic.

The deeper problem: most project management tools are used intermittently. A team might use Asana heavily during a sprint planning week and barely touch it for the next two weeks. Paying $10.99/seat/month for software used 10 days per month feels increasingly wrong to procurement teams doing subscription audits.

AI Tools With Discrete Outputs

This is the category where subscriptions make the least structural sense, yet most companies still charge subscriptions.

An AI image generator that charges $20/month for a certain number of generations. An AI writing tool that charges $30/month for unlimited access. An AI coding assistant that charges per seat per month.

The mismatch: these products deliver discrete outputs — an image, a paragraph, a code suggestion. The value is in the output, not in continuous access. A developer who uses an AI coding tool for one intense week and then doesn't need it for two weeks is paying for three weeks of unused access.

Usage-based pricing aligns incentives perfectly here: pay per image generated, per document created, per code suggestion accepted. The customer pays for value received. The vendor earns revenue proportional to value delivered. Both sides win.

The Three Models Replacing Subscriptions

Model 1: Usage-Based Pricing

Pay per action. Intercom's $0.99 per AI resolution. Stripe's percentage per transaction. Twilio's per-message pricing. AI image generators charging per image.

Usage-based pricing works when the product delivers discrete, measurable units of value. The customer can predict costs based on usage. The vendor's revenue scales with the customer's success. Alignment is structural, not contractual.

The challenge: revenue predictability. Wall Street loves subscriptions because revenue is predictable quarter to quarter. Usage-based revenue fluctuates with customer behavior. Companies with usage-based models trade higher alignment for lower predictability — which can impact valuation multiples.

Model 2: Outcome-Based Pricing

Pay for results, not access or usage. A lead generation platform that charges per qualified lead. A legal AI that charges per contract reviewed. A customer support AI that charges per resolved ticket.

Outcome-based pricing is the logical extension of usage-based pricing: instead of charging per action (per API call, per query), you charge per outcome (per lead, per resolution, per completed task). This is the model Intercom pioneered with per-resolution pricing, and it's spreading to other categories.

The challenge: defining and measuring the outcome. What counts as a "resolution"? What qualifies as a "lead"? The vendor and customer must agree on the definition, and the measurement must be transparent and auditable.

Model 3: Hybrid (Low Base + Usage)

A low monthly base fee for platform access, plus usage-based charges for actual value delivery. This model combines the predictability of subscriptions with the alignment of usage-based pricing.

Example: $10/month base fee for platform access, data storage, and basic features. Plus $0.50 per AI-generated report, $0.25 per automated workflow execution, $1.00 per complex analysis. The customer always has access to the platform (satisfying the infrastructure condition), but pays incrementally for value-creating actions.

This is emerging as the default model for AI-native SaaS in 2026 because it solves both problems: the vendor gets a predictable base of recurring revenue, and the customer pays proportionally for value received.

The Zombie Metric Problem

Here's what makes this transition dangerous for existing SaaS companies: many are reporting subscription metrics that mask underlying decay.

Monthly Recurring Revenue (MRR) counts revenue from active subscriptions. But if a growing percentage of subscribers are in their final month before canceling — they just haven't canceled yet — MRR overstates the health of the business.

Net Dollar Retention (NDR) measures whether existing customers are spending more or less over time. Tomasz Tunguz's recent analysis of 25 public software companies shows NDR declining across the board. This means existing customers are spending less each renewal cycle — either downgrading plans, reducing seats, or canceling outright.

Logo Retention counts the percentage of customers who renew. But a customer who renews at a lower tier or with fewer seats is technically "retained" while generating less revenue. Logo retention can be 90% while revenue from those logos declines 20%.

These metrics were designed for a world where subscriptions were the natural pricing model. In that world, they accurately reflected business health. In a world where subscriptions are being questioned, they become zombie metrics — numbers that look alive but represent a dying model.

How to Navigate the Transition

If you're running a company with a subscription model, the transition to usage-based or outcome-based pricing doesn't have to be sudden. Here's the playbook that minimizes churn:

Phase 1: Introduce a usage component. Add a usage-based element alongside the existing subscription. "Your plan includes X AI-generated reports per month. Additional reports are $Y each." This introduces the concept without eliminating the familiar subscription structure.

Phase 2: Make usage the primary value metric. Shift marketing and customer success conversations from "features included in your plan" to "outcomes delivered this month." Send monthly reports showing: "Your subscription generated X value through Y actions." This reframes the relationship from "access" to "outcomes."

Phase 3: Offer a usage-first plan for new customers. New customers get a low base fee plus usage-based pricing. Existing customers can opt in or stay on their current plan. This creates a natural transition where the customer base gradually shifts without forced migration.

Phase 4: Sunset subscription-only plans. Once 50%+ of new customers are on usage-based plans and the unit economics are proven, begin migrating remaining subscription customers. Offer generous transition terms — lower base fees, usage credits, extended grandfathering.

The transition typically takes 12-18 months. Companies that try to do it faster risk triggering mass cancellations from customers who feel forced into a new model they don't understand.

The Bigger Picture

The subscription economy was a product of its time. When software was deployed on servers and required continuous maintenance, subscriptions made sense — the vendor provided ongoing value through hosting, security, and updates. When content was difficult to produce and distribute, subscriptions made sense — the platform provided ongoing value through a continuously refreshed library.

AI changes both dynamics. Software that used to require continuous access now delivers discrete outputs. Tasks that required continuous tool access now require a single AI prompt. The ongoing relationship between vendor and customer is shifting from "continuous access" to "intermittent value delivery."

The companies that recognize this shift early will build pricing models aligned with how customers actually experience value. The companies that cling to subscriptions — because the metrics look good, because Wall Street understands them, because they're familiar — will watch renewal rates decay until the zombie metrics finally reflect reality.

Subscriptions will survive. But only in the two places where they structurally make sense: continuously refreshed content, and always-on infrastructure.

Everything else is on borrowed time.