SaaS Pricing and Packaging Models That Protect NRR
Pricing is not a number, it is a decision about what you charge for and who gets what. The right structure grows revenue as customers succeed and keeps retention intact. The wrong one trains customers to stay small. This guide covers the main models, how to choose a value metric and how to test.
TL;DR
- Start with the value metric: the unit that grows when your customer gets more value, such as seats, projects, records or outcomes.
- Seat, usage, tiered and hybrid models each have trade-offs for predictability, expansion and margin.
- Packaging (what is in each tier) matters as much as price. Design tiers around customer types, not feature counts.
- AI features carry real variable cost, so usage limits or credits often belong in the model.
- Test with customer interviews and willingness-to-pay methods before changing live prices, and grandfather existing customers.
What you are really choosing
Pricing answers three questions: what unit do we charge for, how do prices change as customers grow, and what is included at each level. Get these right and revenue tends to rise with customer success. Get them wrong and you either leave money on the table or push customers into workarounds.
Find the value metric
A good value metric has three traits: customers understand it, you can measure it reliably, and it grows when they win. Test candidates by asking: “If this number doubled, would the customer be getting roughly twice the value?”
For a scheduling tool that might be active users or bookings. For an analytics product, tracked events. For accounts-payable software, invoices processed. Avoid metrics tied to your cost but not to customer value, such as storage for a tool where files are incidental.
The main models
| Model | How it works | Strengths | Weaknesses | Fits |
|---|---|---|---|---|
| Per seat | Fixed price per user | Predictable, simple | Customers limit seats, shared logins | Collaboration tools |
| Usage-based | Pay for what you use | Tracks value, low entry barrier | Unpredictable bills, forecasting | APIs, infrastructure, AI |
| Tiered | Bundles at set price points | Easy to compare, upsell path | Tiers can feel arbitrary | Most SaaS |
| Flat rate | One price for everything | Simplest to sell | No expansion path | Very small products |
| Hybrid | Base fee plus included usage | Balances predictability and growth | More complex | Many mature products |
None is universally best. Many companies change models as they learn how customers use the product.
Package for customer types
Tiers work best when each one fits a recognisable customer, not when they are lists of features cut at random.
| Tier | Customer | What they need |
|---|---|---|
| Starter | Individual or tiny team testing | Core job, low price, limited volume |
| Team | Growing team | Collaboration, integrations, higher limits |
| Business | Larger company | Admin controls, security features, support |
| Enterprise | Complex buyer | Single sign-on, audit, contracts, custom terms |
Place the features that matter to bigger buyers, such as audit logs and single sign-on, in the higher tiers. Keep the entry tier genuinely useful so people can reach the first value moment.
Pricing AI features
AI features cost money every time they run. If you offer them at a flat price with unlimited use, heavy users can erode margin. Common approaches are included credits or monthly allowances, usage-based add-ons, and higher tiers for larger volumes. Track the real cost per task, since token prices change quickly, as the token price war article explains.
How to research and test
- Interview customers about current alternatives, budgets and what they would give up before paying more.
- Run a price sensitivity survey such as Van Westendorp with a decent sample of your target segment.
- Look at behaviour: which features correlate with upgrades, and where trials stall at the paywall.
- Test on new customers, such as new visitors seeing a different price, rather than changing prices for everyone at once.
- Review after a full cycle, including conversion, average revenue per account and early churn.
Beware relying on what people say alone. Stated willingness to pay is usually higher than real behaviour.
Changing prices without breaking trust
- Give existing customers notice and a reason tied to real improvements.
- Grandfather them for a set period or offer an annual lock-in at the old price.
- Never reduce what a customer already has without telling them.
- Watch cancellations and support tickets closely for the first month.
What pricing does to the rest of the business
Pricing decides your expansion path, your hybrid sales motion and the shape of your unit economics. If the cheapest plan is enough for most customers, net revenue retention will lag, and if the jump between tiers is too large, upgrades stall.
Mistakes to avoid
- Copying competitors’ prices without knowing their costs or customers.
- Too many tiers or add-ons, which slows buying.
- A free plan with no upgrade reason.
- Charging for something customers cannot control, like internal infrastructure costs.
- Never revisiting prices, so a plan from year one still governs year three.
A pricing review checklist
Every quarter, check whether the value metric still tracks customer value, whether any plan attracts the wrong customers, which features drive upgrades, how heavy users affect margin and whether competitors have changed their structure. Write down one change to test next quarter.
Your next step
Book five customer conversations this month about what they would give up before paying more and what would justify a higher plan. Their answers are worth more than any competitor price list.
Bottom line
Choose a value metric that grows with customer success, pick the simplest model that fits, and package tiers around customer types. Price AI features with their real cost in mind, test with new customers, and change prices openly. Pricing is a product decision you should revisit regularly.
Frequently Asked Questions
What is a value metric?
The thing you charge per that scales with the value customers receive. Good ones are easy to understand, easy to measure and grow as the customer succeeds, such as active users, contacts, API calls or processed invoices.
Seats or usage?
Seats are predictable and simple but cap expansion when customers limit users. Usage tracks value but can cause bill shock. A hybrid, a base fee with included usage, balances the two for many products.
How do I research willingness to pay?
Interview customers about alternatives and budgets, and consider a Van Westendorp price sensitivity survey, which asks at what prices a product feels too cheap, a bargain, getting expensive and too expensive. Combine it with actual conversion data.
Should I raise prices for existing customers?
You can, but plan it. Give notice, explain what has improved, and consider grandfathering existing customers for a period or offering a lock-in on annual plans to protect trust.
