SaaS pricing is how a software-as-a-service company charges for ongoing access to its product. The main models are flat rate, tiered, per seat, usage-based, credits, hybrid (a base fee plus usage) and outcome-based, usually with a free trial or freemium plan on the front. In 2026 the default is hybrid: a subscription tier with an included allowance and paid overage for heavy users.
Key takeaways
- There are eight SaaS pricing models in common use. Most successful companies blend them: ICONIQ's 2026 survey found AI builders use an average of 1.7 models.
- Per-seat pricing is losing ground. Consumption pricing rose from 35% to 42% of AI builders in ICONIQ's survey, and WiseTech removed seat fees entirely.
- Your value metric matters more than your price. It should grow as your customer gets more value.
- AI changed the cost side. AI-native gross margins run around 50% to 60% against 75% to 80% for classic SaaS, so heavy users need an allowance and overage.
- Watch net revenue retention, gross retention, ARPU and gross margin to know whether your pricing works.
Before Salesforce, buying a CRM meant a project.
SAP or Oracle would come in, install the software on your servers and charge you a fortune up front. When you needed an update, you paid them to come back. As I wrote in the book, many companies ran on antiquated systems because they couldn't afford, or wouldn't pay for, the upgrade.
Salesforce changed all that. It put CRM in the cloud and charged a monthly subscription per user. It wasn't the first company to try, but it was the first to get it right at scale, and it became the blueprint for every SaaS company that followed.
Per user, per month. For twenty years that was SaaS pricing.
Then, in December 2025, Australia's WiseTech Global did something remarkable. It removed seat and user fees from CargoWise entirely and moved to transaction-based pricing. Its explanation was blunt: "For SaaS businesses that monetize based on seats or users, AI will disrupt them... our revenue is not diluted by customers becoming more productive."
That's the story of SaaS pricing in 2026. The subscription is here to stay. What you charge per is up for grabs. Let me walk you through the models, how to choose, and the numbers that tell you whether it's working.
What is SaaS pricing?
SaaS pricing is how a software-as-a-service company charges customers for ongoing access to its software. It has three parts:
- The model: the structure (flat, tiered, per seat, usage, credits, hybrid, outcome).
- The value metric: the unit you charge for (users, projects, messages, transactions, resolutions).
- The price points and packaging: the actual numbers and what each plan includes.
Most founders start with part 3. I'd start with part 2. Get the unit right and the numbers are easy to adjust later. Get the unit wrong and no amount of price testing will save you.
In the book's terms, SaaS is Model #7 of the eight subscription business models, and it's usually paired with the Regular Fixed Fee pricing model, with usage-based pricing layered on top.
What are the main SaaS pricing models?
Here are the eight models you'll see in 2026, with real examples.
| Model | How it works | 2026 example | Best for | Main risk |
|---|---|---|---|---|
| Flat rate | One price, everything included | Monarch: $14.99 a month or $99.99 a year | Simple products, one customer type | Leaves money on the table with big customers |
| Tiered | Two to four plans with rising features or limits | Xero AU: Ignite, Grow, Comprehensive, Ultimate | Most SMB software | Too many tiers confuse buyers |
| Per seat | Price per user per month | Slack Business+: $15 per user (annual) | Collaboration tools where each user gets value | Revenue shrinks when customers cut headcount |
| Usage-based | Pay for what you consume | Twilio: per message and per minute | Infrastructure and APIs | Unpredictable bills and revenue |
| Credits | Prepaid units spent on actions | Lovable: from $25 for 100 credits | AI apps with varied task costs | Abstract units, backlash when allowances change |
| Hybrid | Base fee plus included usage plus overage | Anthropic Enterprise: $20 a seat plus usage at API rates | Most AI-enabled SaaS | More complex to explain |
| Outcome-based | Pay per result | Fin: $0.99 per outcome | Work with a countable result (support resolutions) | Disputes over what counts |
| Freemium / trial (acquisition layer) | Free plan or free period before paying | Linear: free plan with unlimited members | Products with big markets and low cost per free user | Free users who never convert |
Freemium and free trials aren't really pricing models on their own. They're how customers enter one. I've kept them in the table because they shape everything else.
Let me give you the short version of each.
Flat rate
One price, one plan. It's the easiest pricing page in the world to read. Monarch, a personal finance app, runs paid-only at $14.99 a month or $99.99 a year and reports more than one million members. Flat works when your customers are similar in size and need. It breaks when a ten-person firm and a thousand-person firm pay the same.
Tiered
Good-better-best. This is still the most common shape for SMB software. Xero collapsed its Australian plans into four in July 2024, named for the customer's stage, and has raised prices on 1 July each year since. Tiers work when you can separate customers by need.
Per seat
Price per user. Easy to understand, easy to forecast and, for collaboration tools, roughly tied to value. The problem in 2026 is obvious: if AI lets a customer do the same work with fewer people, your revenue shrinks while your value grows. More on that below.
Usage-based
Charge for what customers consume. In the book I called usage-based pricing the natural evolution of the subscription model, using Procore, Snowflake and JFrog as examples. The upside is that you grow as your customers grow. The downside, as I learned as a national sales manager in telecommunications, is feast or famine. (One government customer ran up a huge conference-call bill during an election. We celebrated. The next month their spend dropped off a cliff and we panicked.)
Credits
Credits are prepaid units that customers spend on actions of different cost. They're the main meter for AI products in 2026, because one task might cost a cent and another a dollar. Credits keep a predictable monthly price while passing real costs through. They also caused most of the backlash of 2025 and 2026.
Hybrid
A base subscription with an included allowance and paid overage. This is now the default for AI-enabled SaaS. ChatGPT Business sells seats plus purchasable credits once allowances run out. Atlassian includes 25, 70 or 150 Rovo AI credits per user per month, pooled across a site. The hybrid pricing guide goes deeper.
Outcome-based
Charge per result. It works best in customer support, where a "resolution" can be counted. Fin charges $0.99 per outcome, Zendesk $1.50 to $2.00 per automated resolution, and HubSpot moved its Customer Agent to $0.50 per resolved conversation in April 2026. It's real but narrow. Decagon offers both models and says the vast majority of its customers choose per-conversation pricing instead.
Freemium and free trials
The book's rule still holds. Freemium works when you have an enormous addressable market, a naturally shareable product and a free version that's useful without gutting the paid one. Otherwise, use a free trial. I'll come back to this.
Which SaaS pricing model is most common in 2026?
The honest answer: a blend.
ICONIQ's 2026 survey of about 305 AI builders found consumption pricing rose from 35% to 42% of companies and outcome pricing from 18% to 23%. Companies now blend an average of 1.7 pricing models.
The fastest growers lean the same way. Among 44 companies reported to have crossed about $100 million in ARR in 2026 (mostly company-reported figures summarized by ARR Club), 18 of the 33 with visible pricing charge mainly on consumption: usage, credits, a usage hybrid or outcomes. Only four rely mainly on seats.
And the plumbing is following. Stripe completed its acquisition of usage-billing company Metronome in January 2026, with Patrick Collison saying "the shift toward usage-based models will be a defining feature of the next decade."
So if you're asking "seats or usage?", the 2026 answer is usually "a base fee for access, plus a meter for the expensive part."
Why is per-seat pricing under pressure?
Because AI breaks the link between headcount and value.
Here's what WiseTech saw coming, in numbers.
What happens when your customer's team shrinks
A logistics customer has 20 users processing 1,000 shipments a month. A year later, AI tools let them process 1,300 shipments with 12 users.
| Pricing | Year 1 | Year 2 | Change |
|---|---|---|---|
| Per seat at $30 per user | 20 × $30 = $600 | 12 × $30 = $360 | −40% |
| Per shipment at $0.60 | 1,000 × $0.60 = $600 | 1,300 × $0.60 = $780 | +30% |
| Hybrid: $200 platform fee + $0.40 per shipment | $200 + $400 = $600 | $200 + $520 = $720 | +20% |
Same customer, same value delivered (more, in fact). Under seat pricing you lose 40% of the revenue. Under a value metric tied to the work, you grow with them. The hybrid keeps a floor under the account and still grows.
The numbers are illustrative, but the mechanism is exactly the one WiseTech described.
Procore shows the other side. It never charged per seat. It prices on a customer's annual construction volume and offers unlimited users. For full-year 2025 it reported revenue of $1.32 billion, up 15%, with gross revenue retention of 95% and net revenue retention of 106%. It isn't exposed to customers cutting headcount, because it never charged for heads.
None of this means seats are finished. Linear charges $10 to $16 per user and reports 177% net revenue retention. Seats work when each new user is genuinely new value. But if your customers are using your product to need fewer people, read seat-based vs usage-based pricing before your next renewal cycle.
How do you choose a SaaS value metric?
The value metric is the unit you charge for. In the book I said it must grow consistently with your customers and help you communicate value. I listed five tests: it needs to be value-based, scalable, flexible, feasible and predictable.
Snowflake charges on data. Attentive, an SMS platform, charges on messages sent. Procore charges on construction volume. Fin charges on outcomes. In each case the metric is something the customer already understands and cares about.
Here's a quick way to test a candidate metric:
| Question | Good sign | Warning sign |
|---|---|---|
| Does it grow when the customer gets more value? | Yes, closely | It grows with your costs, not their value |
| Can the customer predict it? | They can estimate next month | They'd need a spreadsheet |
| Can you measure it reliably? | It's already in your database | You'd need new tracking |
| Does it stall at some point? | Keeps growing with big customers | Caps out once they're "set up" |
| Would a customer explain it to their boss in one sentence? | "We pay per shipment" | "We pay per compute unit-hour" |
If a metric stalls once the customer reaches a certain level, I warned in the book, you end up giving away more than you intend. Score your candidates with the tool below, or read the full value metric guide.
Value Metric FinderScore candidate value metrics and pick the one that grows with your customers.Open the free tool →How should you package SaaS tiers?
Once you have a metric, build two to four tiers around it. My rules:
- Name tiers for who they're for. Xero's Ignite, Grow and Comprehensive beat Basic, Standard and Premium.
- Use your best feature as the upgrade reason. Notion keeps full AI for its $20 Business plan. Slack put advanced AI into Business+ and raised that tier from $12.50 to $15.
- Bundle, don't bolt on. Add-ons are easy to refuse. A better tier at a higher price isn't.
- Keep the entry price stable. It's your front door.
- Leave a lane for enterprise. Custom pricing, annual contracts, security features.
The packaging guide covers what goes in each tier.
Freemium, free trial or reverse trial?
Here's the table from the book, which still holds.
| Freemium works when | A free trial works when |
|---|---|
| Your addressable market is enormous | Your addressable market is smaller |
| Your product is naturally shareable | Your product isn't naturally shareable |
| The free version is useful while the paid version still has plenty of extra value | A free version would strip out everything that makes the product desirable |
Jason Lemkin's maths, which I shared in the book, explains why. At $10 a month you need about a million subscribers to reach $120 million a year. With freemium conversion of 2% to 5%, the conservative case needs 50 million free users. Most markets aren't that big.
In 2026, many self-serve AI products still lead with free entry. Thirteen of the 44 fast growers above have a permanent free tier. Others use credit trials instead, such as Deepgram's $200 of free credit. A reverse trial (full features for a period, then down to free) sits in between. See freemium vs free trial for the full comparison.
Should SaaS plans be monthly or annual?
Offer both, and price the annual plan to pull people in.
In the book I noted that about 70% of subscription services bill monthly, because the commitment is smaller. Annual billing gives you certainty and cash up front, but every renewal becomes what a sales colleague of mine called "poking the bear."
Real annual discounts in October 2026 cluster between about 15% and 26%: Slack Business+ $18 vs $15, Notion Business $24 vs $20, Zendesk Suite Professional $149 vs $115, Shopify Basic $39 vs $29. Apollo takes a different route: annual plans get a full year of credits granted up front, rewarding commitment without a deeper discount.
Before you set your annual discount, check it against your churn. A big discount only pays when monthly customers leave early.
Which SaaS pricing model fits you?
Answer a few questions about your product, costs and buyers and the Pricing Model Picker recommends flat, tiered, seat, usage, credits, outcome or hybrid.
How has AI changed SaaS pricing?
Three big shifts, all in the last two years.
1. AI became the reason to raise prices
Slack, Notion, Xero, Canva and Microsoft all bundled AI into existing plans and lifted prices. Slack folded AI into all paid plans in June 2025. Notion retired its $10 AI add-on in May 2025. Atlassian pools AI credits across a whole site so light users balance out heavy ones.
2. Allowances replaced "unlimited"
GitHub moved Copilot from premium requests to dollar-denominated AI Credits from 1 June 2026, charged at each model's token rate, while keeping plan prices the same. GitHub's own reasoning was that a short chat question and a multi-hour agent session had cost the same under the old model, which was "no longer sustainable." Cursor, Replit and Figma made similar moves. Most of the anger came from quietly shrinking what a plan included, not from headline price rises.
3. Gross margins came under pressure
This is the one most founders underestimate. Classic SaaS runs on gross margins of about 75% to 80%. SaaStr puts AI-native gross margins at 50% to 60%. ICONIQ found the average AI product gross margin was 45% in 2025, projected to reach 53% in 2026. Bessemer found its fastest-growing AI "Supernova" companies averaged about 25% gross margin, "often negative."
Token prices keep falling, but agents use far more tokens per task. A lower listed price can still mean a higher bill.
Why a flat AI seat needs an allowance
You sell an AI-enabled seat at $30 a month. Your app calls Claude Sonnet 5.5, priced at $2 per million input tokens and $10 per million output tokens (Anthropic, checked 27 October 2026). Hosting and support cost $2 per user.
| User type | Tokens a month | Token cost | Gross margin on $30 |
|---|---|---|---|
| Average user | 1M in, 0.2M out | $2 + $2 = $4 | ($30 − $4 − $2) ÷ $30 = 80% |
| Heavy user | 10M in, 2.5M out | $20 + $25 = $45 | ($30 − $45 − $2) ÷ $30 = −57% |
If one user in ten is heavy, your blended gross margin falls from 80% to about 66%: (0.9 × $24 − 0.1 × $17) ÷ $30. If one in five is heavy, it falls to about 53%. The fix is the hybrid: include an allowance that covers the average user comfortably, then charge overage, offer a higher tier, or route heavy work to a cheaper model.
Model your own mix in the calculator below, and read adding AI to a SaaS product before you price an AI feature.
AI Gross Margin CalculatorModel token costs, heavy users and margin for an AI product.Open the free tool →Which metrics tell you your SaaS pricing is working?
Pricing shows up in five numbers.
| Metric | What it tells you about pricing | Real reference points |
|---|---|---|
| Net revenue retention (NRR) | Are existing customers paying you more over time? | Book: JFrog 139%, Snowflake 158% at IPO. 2026: Linear 177%, TechnologyOne 115%, Procore 106% |
| Gross revenue retention (GRR) | Are customers staying, ignoring expansion? | Procore 95% (2025) |
| ARPU | Is your average customer worth more? | Xero ARPU up 23% in fiscal 2026 |
| Gross margin | Can you afford your heaviest users? | Classic SaaS ~75–80%; AI-native ~50–60% |
| CAC payback | How fast does a new customer repay acquisition cost? | Depends on price level and sales model |
In the book I called net dollar retention a miracle, because it shows growth before you sign a single new customer. Usage-based and hybrid pricing are the main reason the best SaaS companies run above 100%. Learn to calculate it in the NRR guide.
One more thing from the book that people forget: sales compensation. Picture two salespeople. John makes twenty-five sales to small businesses. Anna makes five, to bigger companies that will scale their usage for years. Pay commission on deal count and John wins. Under usage pricing, Anna is worth far more. If you move to usage or hybrid pricing, pay your team on starting usage and on how that usage grows, or your best people will leave.
What about SaaS pricing in Australia?
Australian SaaS companies face the same model choices, plus a few local issues. As at October 2026, and as general information:
- GST. Consumer prices should show the total including GST. B2B prices are often quoted ex-GST and should be labeled. Overseas SaaS sold to Australian consumers can also attract GST. See GST on subscriptions.
- Subscription rules from 1 July 2027. Disclosure at sign-up, no unreasonable barriers to cancellation, and ongoing notices at times set by regulations (trial-end and renewal reminders are expected; the regulations weren't made as at October 2026). Some protections extend to small-business subscribers.
- Unfair contract terms. A standard form term that lets you change the price without letting the customer leave can be unfair.
- Valuation. Recurring revenue is what buyers pay for. Pricing that lifts retention and margin lifts what your business is worth.
Free Subscription Profit Review
Australian business owners can book a free 30-minute review with a BPAG Accountants adviser. An honest conversation about:
- whether a subscription model could work for your business
- how recurring revenue could cover your fixed costs each month
- how it could lift what your business is worth
No cost, no obligation. General information only until we understand your situation.
What are the most common SaaS pricing mistakes?
- Pricing from competitors only. Their price reflects their costs and customers, not yours.
- Charging on a metric that stalls. Customers stop paying more just as they get the most value.
- Too many plans. Five or more plans usually means nobody is sure where they belong.
- "Unlimited" with real costs underneath. It invites your most expensive users.
- Never raising prices. Products improve every year. Prices should follow.
- Changing allowances quietly. Announce cuts as openly as price rises.
- Ignoring margin when you add AI. Model your heavy users before launch.
When is SaaS subscription pricing the wrong choice?
When customers use you rarely or unpredictably, a monthly fee can feel like a tax. A prepaid pack or pure pay-as-you-go may fit better. When your revenue is really a share of transactions, as with Stripe or Shopify's payments take, the subscription is secondary to the transaction model. And when buyers can't see an ongoing benefit, no pricing model will save the subscription. That's the lesson of RapidSOS's Haven app in the book, which I cover in how to price a SaaS product.
What to do next
Here's a one-week plan.
- Day 1: Write down your current value metric. Run it through the five tests above.
- Day 2: List your customer segments and what each one values most.
- Day 3: Pick a model with the Pricing Model Picker.
- Day 4: Model your heaviest users' cost to serve, especially AI costs.
- Day 5: Draft two to four tiers in the Pricing Table Builder.
- Day 6: Compare your structure with the SaaS pricing examples.
- Day 7: Decide what you'll test, and when you'll review prices again.
Then ask yourself the WiseTech question: if my best customer became twice as productive with half the staff, would I earn more or less? If the answer is less, your pricing is working against you.
More in this section
SaaS Pricing Examples
24 real SaaS pricing examples from Slack, Notion, Atlassian, Snowflake, Clay, Fin, Zendesk, WiseTech, Xero and more, grouped by model, with each lesson.
How to Price a SaaS Product
How to price a SaaS product step by step: value, cost floor (including AI tokens), value metric, model, willingness to pay and tiers, with an example.
Frequently asked questions
What is the most common SaaS pricing model?
Tiered pricing with a per-seat or usage element is still the most common shape you'll see on SaaS pricing pages. But the direction is clear: in 2026 more companies add a usage allowance and overage to each tier. Among 44 companies that crossed about $100 million ARR in 2026, 18 of the 33 with visible pricing charged mainly on consumption.
How much should a SaaS product cost?
There is no standard number. Price from the value you create, not from competitors alone. A common approach is to estimate the monthly value to a typical customer, capture roughly 10% to 20% of it, and check that your gross margin stays healthy after hosting, support and AI costs. Then test it.
Is per-user pricing dead?
No, but it's weakening. Seats still work when each user gets clear value and headcount tracks the customer's growth. They struggle when AI lets customers do more with fewer people. Many companies now keep a seat price for access and add a usage meter for the expensive work, as Anthropic, ChatGPT Business and Atlassian do.
What's the difference between SaaS pricing and SaaS packaging?
Pricing is the number and the unit you charge for. Packaging is what goes into each plan: features, limits, support and add-ons. They work together. Good packaging gives each customer segment a clear plan and a reason to upgrade, while pricing sets how much each plan costs.
Should a SaaS startup publish its prices?
Usually yes, at least for self-serve plans. Published prices let small customers buy without a sales call and signal confidence. Keep a 'contact sales' lane for large customers with custom needs. Some enterprise-focused AI companies, such as Sierra, publish nothing, but they sell six-figure contracts through sales teams.
How do I price SaaS for Australian customers?
Decide whether to show A$ prices, and whether they include GST. Consumer prices should show the total price including GST. Business buyers often see ex-GST pricing, clearly labeled. Overseas SaaS sellers may also have GST obligations in Australia, so check the ATO's rules or speak with an adviser.
What is a SaaS pricing strategy?
A pricing strategy is the set of decisions behind your price list: who you're pricing for, what value metric you charge on, how you package tiers, how you acquire customers (trial or freemium), how you handle annual plans and discounts, and when you review prices. The price itself is the last decision, not the first.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- The Builder's Economy: 10 metrics from ICONIQ's 2026 State of AI report (SaaStr, 10 Jul 2026)
- WiseTech Global 1H26 results briefing scripts (Feb 2026)
- Procore fourth quarter and full year 2025 results (12 Feb 2026)
- Inference is the new sales and marketing spend (SaaStr, 9 Feb 2026)
- Bessemer: The State of AI 2025 (13 Aug 2025)
- GitHub Copilot is moving to usage-based billing (GitHub blog, 27 Apr 2026)
- Stripe completes Metronome acquisition (14 Jan 2026)
- Anthropic Claude API pricing (checked 27 Oct 2026)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.