Most billing conversations start and end with two models. You pay a flat monthly fee, or you pay per action. A subscription gets you in the door. A usage meter tracks what you consume.
Time-based billing is a third thing. It charges for time spent with a product, a service, or a piece of content. A session opens, the clock runs, and the customer pays for what they used.
For decades this belonged to lawyers and consultants billing hours. That is changing. AI tools running agents with real compute costs, content platforms trying to price fairly, coaching marketplaces, and developer tools are all reaching for time-based billing because it tracks value more closely when time is the variable that matters.
This covers what time-based billing is, the payment mechanic that makes it work, the three models actually in use, who is adopting them, and how to pick one.
What is time-based billing?
Time-based billing is a pricing model where customers are charged according to how long they actively use a product or service. The billing unit is time: minutes, hours, or a bounded window of access.
It sits apart from the two models it gets confused with. A subscription charges a fixed recurring fee regardless of use. Usage-based billing charges per event, such as an API call, a token generated, or a file processed. Time-based billing charges for elapsed time.
The model is old in professional services. A lawyer bills 400 dollars an hour, a consultant bills 250, and the client pays for time rather than deliverables. What is new is the move into software, content, and AI, where products increasingly deliver value tied to active engagement rather than access.
How time-based billing works
The mechanic that makes time-based billing viable is not a charge at the start. It is an authorization hold, followed by a capture for the actual amount at the end.
Session start. The customer opens a session. The platform authenticates them and places an authorization hold on the card for the session ceiling. Nothing is captured. The hold confirms the card is valid and the funds are available before any delivery begins.
Time tracking. The clock runs from session start. The platform tracks active time, handles interruptions, and records the session end when the customer exits or the ceiling is reached.
Access control. For the duration of the paid session, the customer has full access to the gated product or content. When the session closes, access is revoked automatically.
Capture. At session close, elapsed time multiplied by your rate produces the final amount. The platform captures exactly that and releases the rest of the hold.
That last step is what separates time-based billing from a fixed-price product with a duration attached. A session booked for 30 minutes and ended at 18 is charged as 18 minutes. The customer is never out of pocket for time they did not use, and the seller never delivers unpaid work.
The complexity sits between the steps. What happens when a session drops mid-way? How do you handle a partial capture on an interrupted session? How do you stop ghost sessions from running against a hold indefinitely? These look trivial on paper and are not.
The three models
Time-based billing runs in three configurations, distinguished by what the customer is buying.
Per-minute metering
The purest form. A session opens, a clock runs, and the customer pays for the exact time consumed. Common in AI tools with real compute cost per agent-minute, and in content platforms where creators want readers to pay for what they actually engage with.
The rate can be flat, where every minute costs the same, or tiered, where the first stretch is cheaper and later time is priced higher.
Session-based billing
The billing unit is a single session with an agreed ceiling: a 30 minute coaching call, a 60 minute tutoring session, a 15 minute consultation. The reserve and capture flow does the work here. The hold at session start covers the full booked block, and the capture at the end reflects real elapsed time.
This is the standard model in service marketplaces and scheduling-heavy products, where sessions have a shape both sides agree on in advance but rarely run to the exact minute. The customer gets a fixed price to evaluate before booking, which reduces abandonment, and an accurate charge afterwards, which reduces disputes.
Passes
A pass sells a window rather than a meter. A day pass, a weekend pass, a seven day pass. The customer buys access for a fixed period and uses the product as much as they like inside it. When the window closes, access ends.
Time is still what is being sold, but the customer is buying duration rather than consumption. That suits products where your cost per active minute is low and the value of access is high: research tools, archives, event coverage, seasonal products. Passes also convert well as an entry point, because the decision is smaller than a subscription and the pass expires on its own rather than requiring a cancellation.
What does not count: prepaid time packages
Selling a pool of hours or minutes upfront, drawn down session by session, is frequently grouped with time-based billing. Mechanically it is a one-time purchase. The billing event happens once, at the moment of sale, and everything afterwards is entitlement drawdown rather than charging. Time is the unit of consumption but not the unit of billing.
The distinction is worth holding, because prepaid balances behave differently in revenue recognition, refunds, and support load. Unused balances become a customer conversation. Time-based billing, where the charge follows the delivery, does not create that liability.
Who uses time-based billing
AI and developer tools. Agents do real work at real compute cost. Charging by the minute of agent runtime aligns price to cost in a way a flat subscription cannot, particularly for products where agents run multi-step tasks or long-horizon workflows.
Content creators and independent media. A standard paywall charges the same whether a reader engages daily or opens two emails a year. Paid subscription revenue on beehiiv reached 19 million dollars in 2025, up 138 percent from 8 million in 2024, which tells you how fast reader-paid media is growing. Almost all the tooling behind that growth is flat subscription tooling. Flexible time-based monetization for content is still largely unbuilt.
Professional service marketplaces. Lawyers, consultants, coaches, tutors, and therapists have always billed this way. What is changing is that platforms now build the billing layer into the product rather than asking practitioners to reconcile hours through invoices and external trackers.
EdTech and on-demand learning. Charge students for the hours they actually engage rather than a monthly pass they use twice. Lower commitment lowers hesitation at purchase, and revenue tracks real usage.
SaaS with cost-intensive features. Not every feature costs the same to run. Document analysis, video processing, and live data feeds can be priced on time consumed instead of being bundled into a flat plan that raises the base price for everyone.
Why the model is growing
Three forces are pushing it.
AI is the first. Agents cost per runtime minute. A flat subscription across users where some run hour-long agents and others run five-minute queries is mispriced almost by construction.
The second is a shift in what buyers will accept. People are pushing back on subscriptions they barely use. Time-based billing changes the question from whether something is worth 20 dollars a month to whether it is worth two dollars for the next hour. The second question is easier to answer, and it gets asked at the moment of value rather than at renewal.
The third is independent builders. Solo developers, newsletter writers, and small product teams cannot absorb the overhead of custom billing infrastructure. They need session handling, holds, captures, and access control to work without a backend.
Implementing time-based billing without building a backend
Most billing infrastructure was built for subscriptions. Stripe, Paddle, Lemon Squeezy, and Polar all handle recurring payments well. Time-based billing needs more: authorization holds, session state, partial capture, mid-session interruption handling, and access revocation on close.
Tiun models two product types natively, subscriptions for fixed-schedule recurring access and time-based billing for session-driven access. For the time-based path, you create a product in the dashboard and set your rate, then call tiun.start() from the frontend when a customer opens a session. Tiun handles the authorization, authentication, session tracking, and access control. Your app receives an event when access goes live and another when the session ends. Subscriptions run through the same SDK with tiun.checkout({ productId }), which matters when a product needs both.
There is no backend to build, no webhook handler to maintain, and no session state to hold in your own infrastructure.
Choosing the right model
The three models are not interchangeable, and the wrong choice misaligns the price signal with the value delivered.
Per-minute metering fits products where session length varies widely and unpredictably. One user runs a five minute summarization. Another runs a 90 minute analysis pipeline. A flat session fee overcharges the first and undercharges the second. It also fits any service where an expert's time is the product.
Session-based billing fits when delivery is structured around agreed blocks. Coaching calls, therapy, scheduled interviews. The block matches the product shape, so there is no ambiguity at the end, and the customer knows the ceiling before booking.
Passes fit when access itself is the product and your marginal cost per minute is low. They also work as a low-commitment entry point ahead of a subscription, and they are the cleanest option when metering would feel punitive to the customer.
A practical signal. If you already know your average session length and the variance is low, session-based is simpler to operate. If variance is high and your costs scale with time, per-minute metering is the more honest model for both sides. If your cost is mostly fixed and you want people using the product heavily rather than watching a clock, sell a pass.
The bottom line
Time-based billing is not a replacement for subscriptions. The distinction matters when value is tied directly to time spent, because a flat monthly fee mis-measures that value in both directions. It overcharges light users and undercharges heavy ones.
As AI products with real compute costs, independent content businesses, and on-demand service marketplaces grow, time-based billing moves from a professional services pattern into a standard option that any serious billing platform should support.
Frequently asked questions
Is time-based billing the same as hourly billing?
Same logic, different granularity and different timing. Traditional hourly billing rounds to the nearest hour and is calculated after the fact through an invoice. Time-based billing in software operates at the minute level, holds funds at session start, and captures automatically at session end with no manual step. The precision gap is the point: a 22 minute session billed hourly rounds up and overcharges.
Are prepaid credit packages a form of time-based billing?
No, even when the credits are denominated in minutes or hours. A prepaid package is a one-time purchase, and the billing event happens at the moment of sale. What follows is entitlement drawdown rather than charging. The practical difference shows up in refunds, revenue recognition, and unused balances, none of which apply when the charge follows delivery.
What happens if a customer loses connection mid-session?
The reserve and capture flow answers this cleanly. Because funds were held rather than charged, the platform closes the session at the last confirmed activity timestamp, captures only the elapsed time, and releases the remainder of the hold. Without that mechanic you get one of two bad outcomes: ghost sessions that keep billing, or blanket refunds that create an abuse vector.
Can a product run subscriptions and time-based billing together?
Yes, and many should. A subscription can cover baseline access, a dashboard, or a free tier, while time-based billing applies to the features that carry real cost per use. A research tool might offer an account with limited queries and charge by the minute for deep analysis mode. The models serve different parts of the same product.
How do customers react to being charged per minute?
Resistance is lower than most builders expect when the rate is visible before the session starts and the capture is accurate afterwards. The dynamic differs from subscriptions. A monthly fee creates a value judgement at signup that gets revisited at every renewal. A per-minute rate creates a small immediate decision at the moment of use. The friction lives on the rate card, not in the billing mechanic.