SaaS Pricing Models: 7 Models, With Real Pricing Page Examples

Per-seat pricing charges for each user who gets access. It is the default for collaboration software because the number of people using the tool is a decent proxy for the value the account gets.

The upside is predictability. Buyers can do the math in their head, and revenue grows as the customer hires. The downside is that it punishes adoption. Every new seat is a budget conversation, so teams share logins or keep casual users off the tool.

The best per-seat pages fix that leak by pricing seat types differently. Figma charges $16 per full seat on Professional and lets viewers in free. Linear allows unlimited members on its free plan, so the seat charge only starts once a team is committed.

Per-seat works best when nearly everyone on the team gets similar value from the product. It breaks down when one power user does most of the work, or when the product runs automatically in the background. In those cases you are charging for logins, not for value, and customers notice.

On the page, per-seat pricing almost always pairs with a monthly/yearly toggle. That toggle appears on 38.1% of the 989 pages in our catalog, and per-seat products lean on it hardest because the annual discount multiplies across every seat.

Usage-based pricing bills for consumption: requests, hosts, tokens, shipments, transactions. It is the natural model for infrastructure, APIs and AI, where your own costs scale with what customers do.

Customers like that they start small and pay only for what they use. The risk sits on both sides. Customers fear surprise bills, and you get revenue that moves with their traffic, not with a contract.

Usage pricing pages look different from plan grids. Algolia prints its rates in the card body: 10K search requests free, then $0.50 per 1K on Grow. Datadog prices at $15 per host per month on Infrastructure Pro. Groq lists cost per million tokens next to model speed. Our full usage-based pricing guide covers value metrics, free allowances and overages in detail.

Because buyers cannot see their bill on a rate table, usage pages need a calculator or slider. Only 7.5% of catalog pages have one, which is a gap most usage-based companies could close.

Usage pricing also needs a floor. A free allowance gets developers testing, and a published overage rate tells finance what happens past the limit. Leave either out and you trade signups for support tickets.

Credit-based pricing sells a bundle of prepaid units that customers spend on actions. It is a wrapper around usage pricing, common in AI tools where one action can cost very different amounts to serve.

Credits give you a single meter across many actions and let you change the cost of each action without repricing plans. Buyers get a fixed monthly bill. The catch: credits are abstract. Nobody knows what 2,000 credits buys until you translate it.

The pages that work translate it. Bardeen adds a credit cost table that explains the meter, and a dropdown inside the card sets credit volume. Macaly prices credits in real tasks in an FAQ table. Base44 splits two credit types on every card, which is honest but adds a second thing to compare.

Feature-tiered pricing sells packages. Each plan has a fixed price and a fixed set of features and limits, and customers move up when they need what the next tier includes.

This is the classic three or four card grid. It is easy to buy, easy to compare and easy to sell self-serve. The hard part is packaging: deciding which features gate which tier, and spacing the prices so each step feels earned. We cover that in depth in the tiered pricing guide.

AgentMail runs Free, Developer at $20, Startup at $200 and Enterprise at custom pricing, with tenfold jumps between tiers and an inverted card spotlighting Startup. HeyGen keeps it to three plans, Free, Creator at $29 and Pro at $99, with limits stated as concrete caps.

The weakness of pure feature tiers is the ceiling. Once a customer is on the top self-serve plan, revenue stops growing no matter how much they use the product. That is why so many tiered companies add seats, usage or an enterprise plan on top, and drift into hybrid pricing.

Feature tiers depend on a clear feature comparison table. 68.5% of catalog pages include comparison rows, making it the most common pricing page feature we track.

Hybrid pricing combines a fixed platform fee with a variable component, usually usage, seats or credits. It is where most modern SaaS ends up once it outgrows a single meter.

The base fee gives you predictable revenue and qualifies serious buyers. The variable part lets big customers pay more without forcing everyone into a giant plan. The cost is complexity. Two meters means two things to explain, and the page has to do the explaining.

Apify sells Starter at $29 per month plus pay as you go, converts the subscription into spend credit, and shows a worked example under its How pricing works section. Langfuse charges a base price plus $8 per 100k units. GitHub Copilot defines each tier by a premium request quota and publishes the overage price.

Flat-rate pricing is one price for one product. No tiers, no meter, no seat math. It is the simplest model to explain and the hardest to grow revenue on, because a customer who gets ten times the value pays the same.

Single-plan pages are rare. Only 3.1% of catalog pages show a single tier. Copy.ai puts one self-serve Chat plan on the page at $29 per month covering five seats. GrowthList runs one card, Full Access at $29 per month, with trial terms under the button.

Flat pricing can also work as positioning against per-seat competitors. Basecamp offers Pro Unlimited at $299 per month billed annually and pitches that flat price against per-user pricing, staggering card heights to rank value.

Quote-based pricing hides the number and routes buyers to sales. It fits complex deals: large implementations, regulated buyers, heavy services, or contracts where price really does depend on scope.

The cost is friction. Self-serve buyers leave, and the ones who stay arrive at the sales call with no anchor. Hidden prices show up on 9.1% of catalog pages, and a custom quote path appears on 24.6%. That second number is mostly hybrid: a public grid with a quote-only top tier.

If you go fully quote-based, explain why. Chamber's page is a single contact card, but it gives a reason for hiding prices and uses a founder-led CTA. CallZen.AI shows no prices at all, only a feature matrix, with a Value for Money badge on the middle plan.

A middle ground works better for most B2B products: publish self-serve tiers, then add an enterprise tier with a starting price. Calendly anchors its Enterprise plan at a real number, starting at $15k per year, so big buyers know the range before they book a call.

Start with the value metric, not the model. Ask what unit of your product grows as the customer gets more value. People using it, work processed, outcomes delivered. The model follows from the answer. Our value-based pricing guide walks through how to find that metric.

A good value metric passes three tests. Customers understand it without a glossary. It grows as their usage grows. And they can predict next month's bill within a reasonable range. Seats pass all three for collaboration tools. Tokens pass the first two for AI APIs and fail the third, which is why AI companies wrap them in credits or plans.

Then match the model to how your customers buy:

• Value grows with headcount and buyers want a predictable bill: per-seat.
• Value and your costs grow with consumption: usage-based, or credits if actions vary wildly in cost.
• Distinct customer segments want different capabilities: feature-tiered.
• Big accounts dwarf small ones on usage but you need a revenue floor: hybrid.
• One clear job, one buyer type, and simplicity is your pitch: flat-rate.
• Deals need scoping, security review or services: quote-based, ideally only for the top tier.

A model only works if buyers understand it in the few seconds they spend on the page. Each model needs different support. Per-seat needs a seat selector and annual toggle. Usage-based needs a calculator and a free allowance. Credits need a translation table. Tiers need a clear highlighted tier, which 56.5% of catalog pages use.

Whatever the model, the questions follow. FAQs appear on 60.1% of the 989 pages, more than any other extra we track. Write them for the objections your model creates: surprise bills for usage, seat sprawl for per-seat, what happens at the limit for tiers.

If your model is right but the page is not explaining it, that is a fixable problem. Fix the page first, then judge the model. If you want a second set of eyes, Conversion Factory will revamp your pricing page for a flat fee.