The Short Answer
The right pricing model isn't the one with more hype behind it. It's the one that matches your cost basis and how your customers actually expand. Twilio and monday.com both grew revenue 22% in their June quarters and still finished 109 stock-performance points apart, as reported by SaaStr on September 6, 2026, because one prices by consumption and the other prices by seat. This is a decision framework, not a how-to: if you have already decided to move to usage-based pricing and need the Stripe mechanics, the migration steps live in the usage-based pricing implementation guide. The 4-question diagnostic below tells you which side of that 109-point gap your business sits on.
Get frameworks like this diagnostic before your next pricing review, not months after. Subscribe to the SaasFlywheel newsletter.
Usage-Based and Seat-Based Pricing, Defined in One Paragraph Each
Seat-based pricing charges per named user or licensed seat, the model Slack, monday.com, and HubSpot built their revenue on. Usage-based pricing, also called consumption-based pricing, charges for what a customer actually consumes: API calls, data processed, or workflow runs, the model Twilio, Snowflake, and ServiceTitan use.
| Dimension | Seat-based | Usage-based |
|---|---|---|
| Definition | Price per named user or seat | Price per unit consumed (API call, credit, workflow run) |
| Margin behavior | Stable margin per seat, regardless of usage | Margin tracks cost as usage scales |
| Customer psychology | Predictable bill, easy to budget | Bill scales with value, can trigger bill anxiety |
| Best-fit company shape | Value scales with headcount, collaboration tools | Cost scales with usage or inference, API and AI-native products |
ServiceTitan proves consumption pricing works at real size: usage revenue grew 29% against 24% for subscriptions and now makes up about 22% of platform revenue, on $21.7 billion in gross transaction volume, up 23%, as reported by SaaStr on July 10, 2026. Neither model is inherently better. Switching between them in Stripe is covered in the implementation guide, not here. A third option, hybrid pricing, blends both inside one plan, and we'll get to it later.
The Market Already Priced This Question
The clearest evidence that billing unit now sorts SaaS valuations arrived in a single week. Twilio grew revenue 22% in its June quarter and closed up 73.8%. monday.com also grew 22% and closed down 35.5%. Same growth rate, opposite pricing model, a 109-point spread, as reported by SaaStr on September 6, 2026, which named the mechanism directly: “Every name in the winners group prices by consumption, by commitment pool, or by protected surface. Every name in the losers group prices by seat.”
Figma sharpens the case: the fastest revenue grower in SaaStr's entire comparison set at 48%, with 136% net revenue retention, yet its stock still fell 35.1%, as of the September 2-3, 2026 close. Figma is seat-priced, and AI credits added a real inference line to its cost of goods sold, compressing gross margin five points year over year. Growth this strong is supposed to buy you slack. Here it didn't, because the pricing model was quietly eating the growth from underneath.
The market keeps repricing this way for a mechanical reason. Redpoint's March 2026 survey of 141 CIOs, per SaaStr, August 25, 2026, found 45% pull AI budget from existing software line items rather than net-new spend, and 46% expect usage or outcome pricing to grow while 29% expect seat-based pricing to shrink. Seat-based pricing is structurally exposed wherever AI adds a real cost underneath it, and that is a mechanism, not just a stock move.
The 4-Question Diagnostic: Which Model Fits Your SaaS
Tomasz Tunguz's seat-vs-usage framework sorts pricing into three descriptive buckets, a useful starting taxonomy that predates the 2026 evidence above and offers no way to score your own business. The diagnostic below asks four questions, each pointing toward seat-based, usage-based, or hybrid. Answer honestly and count which column you land in most; that count, not a gut feeling, is your signal. Split down the middle? Our 7-signal scorecard for a broken SaaS pricing model narrows it further.
| Question | Points to seat-based | Points to usage-based | Points to hybrid |
|---|---|---|---|
| Cost-basis volatility | Cost per customer is stable, headcount-driven | Cost scales with usage, API calls, or inference | Some cost is stable, some is AI or usage-driven |
| Customer expansion behavior | Customers expand by adding seats | Customers expand by using more of the product | Customers expand on both seats and usage |
| Sales motion | Sales-led, buyer wants a budget-line number | Self-serve, product proves value before a call | Sales-led core, self-serve usage add-on |
| Company stage | 0-$50K MRR, still learning your cost curve | Cost curve known, usage is measurable | Post-PMF, adding a usage layer to a working core |
How volatile is your underlying cost basis?
Cost-basis stability is the single most important diagnostic in pricing model choice, and most articles skip it entirely. If cost per customer is stable and headcount-driven, as with most non-AI B2B tools, seat-based pricing has little to fear. If cost scales with usage or inference, a flat per-seat price recreates Figma's problem: your busiest customers cost more to serve while their bill stays flat. That gap only grows the longer you leave it unpriced.
What does your customer actually expand on?
Look at how net revenue retention actually grows, not how you assume it grows (see our NRR benchmarks piece for the breakdown by stage and motion). HubSpot is the seat-based cautionary case: NRR slipped a point to 102%, net customer adds came in at 7,000 against a 9,000-10,000 target, and the stock finished down 33.3% year to date, as of the September 3, 2026 close, per SaaStr. monday.com shows the opposite failure: AI ARR doubled quarter over quarter to 17% of net new ARR after it moved AI products to consumption pricing, yet the stock still fell 35.5%, because the core plan underneath stayed seats. A usage layer bolted onto a seat-based chassis doesn't fully fix the chassis.
Are you self-serve or sales-led?
At 0-$50K MRR you are almost always self-serve or founder-led sales, answering your own tickets between deploys, which means every hour spent negotiating a seat count is an hour you didn't spend shipping. Usage-based pricing removes a sales conversation about headcount and lets the product prove its value before you get on a call, saving time you don't have. Seat-based pricing stays easier to forecast and quote for an early sales-led motion, where the buyer wants one number for a budget line, not a usage estimate.
What stage are you actually at?
Most pricing advice online targets a team well past $1M ARR that already knows its cost curve. At 0-$50K MRR, you often do not. Cost-basis instability is a real risk, but often not your biggest one yet: revenue predictability while you find product-market fit can matter more than a margin problem still two years out. Our stage-gated SaaS pricing strategy guide covers when that changes.
When Seat-Based Still Wins
Seat-based pricing is not dead. Bain's analysis of more than 30 SaaS vendors that added generative AI found roughly 35% simply raised per-seat prices outright (Zoom's path), while about 65% layered a hybrid AI meter on top of seat pricing instead (Adobe and Salesforce's path), per Bain & Company, October 9, 2025. A large share of the market still runs pure seat pricing successfully.
Collaboration and internal-workflow tools are the clearest fit: value genuinely scales with headcount, so more seats mean more value delivered. If you cannot yet predict your own usage or cost curve, seat-based pricing buys forecasting stability while you learn your numbers, a legitimate reason to stay put, not a failure to keep up with a trend. Chasing usage-based pricing before you understand your own costs just means guessing at a metered price instead of a predictable flat one.
When Usage-Based Wins
Usage-based pricing wins when your marginal cost per customer action is real and variable: inference calls, API usage, data processed. Flat per-seat pricing on top of that cost base eventually charges your happiest, heaviest users for headcount your product just helped them shrink, turning your best customers into your least profitable accounts, a pattern Willingness to Pay describes from its own 200-plus pricing redesigns, per SaaStr, July 11, 2026 (a vendor self-report, not an independent benchmark).
a16z's framing is sharper: price the layer you actually sell. “If you sell model access, price tokens. If you turn models into useful work, price the recognizable value unit, often through credits. If you deliver a clear and attributable business result, price the outcome,” write Tugce Erten and Sarah Wang (a16z, August 27, 2026). In a16z's own small, unpublished-methodology survey of 50 technical AI buyers, 27 preferred credits over raw tokens, roughly two to one. The per-token, per-agent, and hybrid taxonomy lives in our AI-native SaaS pricing guide. Usage-based pricing also lowers friction between trial and paid for a self-serve product, since the bill only grows once the product proves its worth.
The Hybrid Middle Path (and the Switching Mistake That Costs the Most)
Hybrid is where most companies actually land, and it is usually correct once a product has a stable core job plus a variable AI or usage layer, not a compromise. Bain found about 65% of the 30-plus vendors it analyzed introduced exactly this pattern, an AI meter layered on a seat-based core (Bain & Company, October 9, 2025). The common shape: a base subscription with an included usage tier plus metered overage above it, the retrofit pattern Notion AI and the Vercel AI SDK both use structurally.
The most expensive mistake is switching models without giving customers a number they can independently verify against their own usage. “Durable AI pricing ties revenue to something a customer can independently verify, not to a claim,” is the framing Bessemer Venture Partners uses in its piece “AI pricing models that survive renewal: 4 founder lessons” (Bessemer Atlas, July 30, 2026). If you have decided to move toward usage-based pricing, the Stripe mechanics and migration steps live in that guide.
What Founders Are Actually Saying Right Now
Mark Cuban asked the question plainly on X on February 25, 2026: “doesn't this eliminate the need for per seat pricing by the software companies,” the exact doubt this article exists to resolve. Marty Kausas put the mechanism in one sentence on X on May 20, 2026: “seat-based doesn't work when an agent does variable amounts of work per user,” a direct match for the cost-basis question above.
Across the public discussions we reviewed on this exact question (45 items, Reddit and X, scanned September 14, 2026, not a formal survey), neither model has settled the debate. This Week in Startups argued on X in September 2025 that the next frontier is neither pure seat nor per-call pricing. Fynn Glover noted on X that SaaS pricing metrics still lack standardized definitions. If experienced operators are still arguing this in public, your own hesitation is not a sign you are behind. It's a sign you're paying attention.
FAQ
What does usage-based pricing mean for SaaS? Usage-based pricing, also called consumption-based pricing, means a customer's bill is tied to what they actually consume, API calls, data processed, workflow runs, or credits, rather than a fixed number of seats. It carries more revenue volatility per account but keeps margin closer to cost as usage scales.
What are the four types of pricing? The four types most cited are flat-rate, tiered, per-user, and usage-based pricing. For a SaaS business today, the real decision is seat-based versus usage-based, or a hybrid, which the diagnostic above resolves.
What is the Rule of 40 in SaaS?The Rule of 40 says a healthy SaaS company's revenue growth rate plus profit margin should add up to 40% or more. Pricing model is one lever toward that number, not the only one: usage-based or hybrid pricing that captures more expansion revenue can lift growth, but it does not replace sound unit economics.
What are the five types of pricing? Add value-based pricing to the four above and you get the five-type list some sources cite. It is useful for a textbook, not the question a founder choosing between seat-based and usage-based needs answered, which is why this article stays on that narrower decision.
Get the Pricing Decision Playbook
The core move is simple to state and hard to skip: name your own cost-basis volatility, then run it against the table above, not whatever model is trending this quarter. We track shifts like the Twilio/monday.com divergence as they happen. Subscribefor pricing and market-structure coverage before it shows up in everyone else's recap.