practical pricing playbook for founders — value-based, usage-based, freemium, tiers, and how to migrate without losing customers
- Most SaaS companies underprice by 2–3x. Patrick Campbell's ProfitWell data has been showing this for a decade and founders still don't believe it.
- Usage-based pricing is the fastest-growing model in 2026. Snowflake, Stripe, OpenAI, Anthropic, Twilio — the largest infrastructure businesses charge by consumption, not seats.
- The default modern stack is three published tiers (Starter / Pro / Business) plus a custom Enterprise tier with "contact sales." Anything else is friction.
- Raise prices yearly. Grandfather existing customers for 12 months. Communicate the value increase, not the cost increase.
- AI-native tooling reset everything in 2024–2025: per-seat collapsed for individual contributors, credits and tokens became normal, and the "$20/month for unlimited everything" anchor is dead.
Most SaaS founders charge $20 a month for software that saves enterprise teams $50,000 a year. They look at a competitor's price, drop a few dollars below it, and call that strategy. Then they spend two years confused about why growth feels capped and why their best customers tell them, unprompted, that the product is too cheap. Pricing is the highest-leverage lever in the SaaS business — Madhavan Ramanujam's research at Simon-Kucher shows a 1% improvement produces an 11% lift in operating profit, more than any other lever — and the one most founders refuse to touch. This playbook walks through the models that work in 2026 and how to migrate between them without burning your base.
What changed in SaaS pricing in 2026
The 2010s playbook — three flat per-seat tiers, $9 / $29 / $99, 14-day trial, contact us for enterprise — was built for a world where software cost almost nothing to scale. AI changed that. Inference is expensive, so flat-rate pricing now bankrupts the company. Value is enormous and uneven, so usage-based pricing finally makes sense outside infrastructure.
Four trends dominate. Usage-based momentum: OpenAI, Anthropic, Snowflake, and Stripe all charge by consumption — none of them could exist on per-seat. Credit packs: Notion AI, Cursor, Linear, and Loom sell AI features as monthly credit allowances on top of base subscriptions, because token usage is too unpredictable for flat pricing. The post-ZIRP reset: software valuations crashed in 2022, survivors raised prices, killed unprofitable freemium tiers, and started charging for what used to be free. AI-driven personalization: Stripe and ProfitWell now use ML to personalize trial, expansion, and retention offers in real time. Founders who haven't reset their pricing since 2021 are leaving the most money on the table they will ever leave. Fixing it is mostly mechanical.
The 7 SaaS pricing models
Seven pricing models actually ship in production SaaS. The rest — pay-what-you-want, perpetual licenses, lifetime deals — are marketing stunts or niche plays. The seven below cover ~99% of B2B SaaS revenue. Most companies eventually run two or three in combination.
| Model | Best for | Pros | Cons | Examples |
|---|---|---|---|---|
| Flat-rate | Single product, single buyer, simple value prop | Easy to understand, fast to buy, predictable revenue | Leaves money on the table, no expansion path | Basecamp ($299/mo flat), early Linear |
| Per-user (per-seat) | Collaboration tools where every user gets value | Predictable, scales with team, easy to forecast | Buyers share logins to avoid seats, kills land-and-expand | Slack, Notion, Linear ($8/$14/seat) |
| Tiered | Multiple buyer segments with different feature needs | Anchors middle tier, segments by willingness to pay | Decision fatigue if more than 4 tiers | Notion ($10/$15/$20), HubSpot, Mailchimp |
| Usage-based | Bursty workloads, infrastructure, variable value | Aligns price with value, no cap on expansion | Unpredictable revenue, finance teams hate it, bill shock | Stripe (2.9% + 30¢), Snowflake, OpenAI, Twilio |
| Freemium | Bottom-up adoption, viral product, zero-touch onboarding | Massive top-of-funnel, product-led growth fuel | 2–5% conversion, free users cost real money | Notion, Figma, Canva, Loom, Slack |
| Per-feature | Modular products with distinct buyers per feature | Clear upsell path, customers buy what they need | Confusing pricing page, expensive to maintain SKUs | HubSpot Hubs, QuickBooks add-ons |
| Hybrid | Mature SaaS with mixed predictable + bursty usage | Captures both base value and expansion | Hard to communicate, hard to forecast | HubSpot (seats + contacts), Datadog (hosts + events), Vercel (seats + bandwidth) |
Most successful B2B SaaS in 2026 runs tiered subscriptions with a usage component on top. Pure flat-rate is rare past Series A. Pure per-seat is collapsing wherever AI agents, automations, or external collaborators count as "users."
Value-based pricing (the only real model)
Every model above is a billing mechanism. Value-based pricing is the philosophy that decides what numbers go into them. As Ramanujam writes in Monetizing Innovation, price is set by the value the customer captures, not by what it costs you to deliver. Cost-plus pricing — "our infra costs $3, so we charge $9" — is how founders systematically underprice. The customer doesn't care what your AWS bill looks like.
The anchor calculation: estimate the customer outcome in dollars (revenue gained, cost saved, time saved at fully-loaded labor cost), divide by effort to deliver, and price between 10% and 30% of value created. A product that saves a marketing team 20 hours a week at $75/hour is worth $78,000 a year. Charging $99/month for that outcome leaves 65x on the table. Patrick Campbell at ProfitWell calls this the "pricing iceberg" — most SaaS companies never raise it.
Tiering that converts free → paid
The dominant 2026 tier structure is four tiers: Starter for individuals, Pro for professionals, Business for teams, Enterprise custom. The trick is making each tier feel inevitable for the next buyer up — a Pro user should hit a wall only Business solves, not one they can route around.
- Starter / Free — single user, limited workspace, watermarked exports, no integrations. Goal: prove value, not generate revenue. Conversion target 2–5% within 90 days.
- Pro ($15–$30/user/mo) — full features for one professional. Unlimited workspaces, API access, priority support. Sweet spot for solopreneurs, freelancers, and small teams.
- Business ($40–$80/user/mo) — collaboration features, admin controls, SSO (basic), audit logs, role-based permissions. This is where most B2B revenue lives.
- Enterprise (custom) — SSO/SAML, SOC 2, DPA, custom contracts, dedicated CSM, SLA. Starts at roughly 3–5x your Business tier per seat. Always "contact sales."
The wrong move is putting SSO in the cheapest paid tier. SSO is the most reliable enterprise upsell trigger in SaaS (there's an entire site, sso.tax, dedicated to shaming companies that price it correctly) and giving it away costs tens of thousands per enterprise account. Audit logs, SCIM, custom contracts, and DPA belong above Business.
Freemium vs free trial
Freemium and free trials look similar on a pricing page and are completely different businesses. Freemium gives away a permanent tier and bets on long-term conversion. Free trial gives away the full product for 7–30 days and bets on conversion at the end. Picking the wrong one wastes years.
Freemium works when
- Product is viral or collaborative (every free user invites paid users)
- Marginal cost per free user is near zero (no heavy compute, no support)
- Product proves value within minutes of signup
- You can build a moat from data or network effects on free users
- Realistic conversion: 2–5% of active free users to paid
Free trial works when
- Product is for a specific job-to-be-done with measurable outcomes
- Onboarding takes effort (data import, integration setup) and users won't repeat it
- Marginal cost per user is non-trivial (AI inference, storage, bandwidth)
- Decision-makers want to verify ROI before signing
- Realistic conversion: 10–20% of trial starts to paid (60%+ if reverse trial)
The reverse trial — sign up, get full Pro for 14 days, auto-downgrade to free instead of getting locked out — is the highest-converting onboarding pattern of 2024–2025. Notion, Loom, and Vercel all moved to it. It captures freemium top-of-funnel and trial conversion mechanics in one structure.
Usage-based vs subscription
The question isn't "which is better." The question is "what is the shape of value delivery?" Predictable, recurring value belongs in a subscription. Bursty, variable, or infrastructure-shaped value belongs in usage. Most modern SaaS uses both.
If your customer logs in every Monday and gets the same outcome, that's a subscription. If their usage is 10x higher in November because of Black Friday, charging them the November price all year is how you lose them in February. Stripe charges per transaction, Snowflake per query-second, OpenAI per token — because that's the unit of value. None would survive flat-rate.
The 2026 default is hybrid: a base subscription with a usage allowance, then metered overages. HubSpot does it with contacts, Datadog with hosts and events, Vercel with bandwidth. The base gives finance teams predictable revenue. The overage captures expansion.
Annual vs monthly
Annual billing fixes more SaaS problems than any other single lever. It improves cash flow (12 months upfront), reduces involuntary churn (one card decline vs twelve), increases LTV (annual customers churn at half the rate), and gives sales a reason to ask for a bigger commitment. The cost is a discount.
Standard is 15–20% off monthly when paid annually. Below 15% nobody bothers. Above 25% you signal that monthly is fake. Most B2B SaaS in 2026 prices monthly at the "real" price and offers "two months free" (16.7% off) for annual — same math, better marketing.
Enterprise pricing (the dark art)
Enterprise pricing has nothing to do with the price on your pricing page. The Enterprise tier exists for one reason: to let you charge what the deal is worth, which is usually 5–10x your highest published price. Self-serve customers pay what's published. Enterprise customers pay what their procurement team can be made to sign off on.
The mechanics: any company with 500+ employees has procurement, InfoSec review, a legal team that wants to redline your MSA, and a budget cycle. They will require SOC 2 Type II, a DPA, SSO with their IdP (Okta, Azure AD), and a dedicated CSM. They will also pay $50K–$500K per year for software your self-serve users get for $99/month, because their alternative is $2M of internal engineering. Don't list a price — "Contact sales" exists because sales is the price discovery process.
The biggest mistake on enterprise: trying to sell it before shipping what enterprise actually needs. SSO/SAML, SCIM, audit logs, RBAC, SOC 2, DPA, custom contracts, regional data residency — these are table stakes that get you past InfoSec, not nice-to-haves.
Pricing page design
The pricing page is the highest-converting page on your site after the homepage and signup, and the one founders touch the least, which is why most still look like 2018. A modern B2B pricing page in 2026 follows a tight checklist.
- Three published tiers, anchored on the middle one with a "Most popular" tag
- A fourth Enterprise tier with "Contact sales" — never a price
- Monthly / Annual toggle, default to Annual to anchor the cheaper price
- Comparison table below the cards with checkmarks for every feature, grouped by category
- FAQ section answering refund policy, cancellation, payment methods, taxes/VAT, plan changes
- Social proof — logos of customers, testimonials, "10,000+ teams use X" — placed adjacent to the cards
- One clear CTA per tier: "Start free trial," "Get started," "Contact sales" — never two CTAs on the same card
- Currency switcher if more than 20% of traffic is non-US
Linear, Vercel, Resend, and Notion all run variations of this. It works because the three-tier anchor exploits centering bias — see Starter / Pro / Business, pick Pro; see Pro / Business / Enterprise, pick Business. The middle tier is whatever you want most customers to pick.
Raising prices without churn
Most SaaS companies should raise prices every 12 months. The ones that don't leak 5–10% of potential ARR a year to inflation alone. Founders avoid it out of churn fear. Done correctly, churn from a price hike is under 2%. Done badly, you lose 20%. The difference is process.
- Announce 60 days out. Email every paying customer. Explain the new prices, what's new in the product, and exactly when their bill will change. Surprise price hikes destroy trust.
- Grandfather existing customers for 12 months. Tell them they keep their current price for the next billing year. New signups pay the new price starting today.
- Communicate the value increase, not the cost increase. "We've shipped X, Y, Z over the past 12 months and our pricing now reflects that" is the message. "Costs have gone up" is not.
- Offer an annual lock-in. Give grandfathered customers the option to lock their old price for two more years if they switch to annual billing. Cash flow win, retention win.
- Measure churn delta, not absolute churn. Compare 90-day churn before and after the increase. If it's under 2 percentage points higher, you priced too low for too long.
Notion, Linear, Figma, and HubSpot have all raised prices in the past 24 months using this playbook with no meaningful churn spike. The companies that did see spikes — Adobe Creative Cloud, Unity's runtime fee, Heroku killing free — broke at least two of these rules.
Common mistakes
The same five pricing mistakes appear in nearly every founder consultation. They are predictable, expensive, and usually fixable in a quarter.
FAQ
How do I price a brand-new SaaS with no customers?
Pick a value-based anchor and start 30% above your gut. Three tiers — roughly $19, $49, $99/month — covers most early B2B SaaS. Exact numbers don't matter; having a published price you can test does. Twenty prospect calls with a real price beats six months of pricing analysis with no price.
When should I switch from subscription to usage-based?
When the top 10% of accounts consume 50%+ of your costs, flat pricing can't solve it. The fix isn't a full migration — it's hybrid. Keep the subscription as a base with a usage allowance, then meter overages. Stripe, Twilio, and Snowflake all started usage-only and moved to base-plus-overage as enterprise buyers demanded predictability.
How long should a free trial be?
Long enough for the user to hit the "aha" moment twice. For most B2B SaaS, 14 days. For weekly workflows (newsletters, reporting, payroll), 30 days because the core action only fires once a week. Trials past 30 days don't convert better — they just delay the decision and increase the chance the user forgets they signed up.
Are discounts okay or do they cheapen the brand?
Annual discounts (15–20%) are fine and expected. Black Friday flash sales on a $300/month B2B SaaS are not — they train every buyer to demand a discount. Reserve discounts for annual upgrades, multi-year contracts, and competitive displacement on enterprise. Never run sitewide flash sales on subscription software.
Do I need annual plans on day one?
No. Ship monthly first, prove customers stick for 90 days, then add annual. Offering annual before you know your retention curve means collecting cash for a year of service 40% of those customers will never use — technically deferred revenue, practically a refund obligation. Add annual once monthly logo retention is above 90%.
How do I price AI features in 2026?
Three options. Credit packs — a monthly credit allowance that resets each cycle (Notion AI, Cursor, Linear). Metered overages — flat allowance in the base tier, per-token above (HubSpot AI, ClickUp Brain). Or a separate AI add-on SKU on top of the base product (Microsoft Copilot, Salesforce Einstein). Avoid "unlimited AI for $20/month" — it works for marketing and bankrupts unit economics within six months.
The Bottom Line
Pricing is the highest-leverage decision in SaaS and the one most founders treat with the least rigor. Pick a value-based anchor, ship three tiers plus enterprise, layer usage on top of the parts that vary, raise prices every twelve months with a clean grandfather, and listen to the silence when prospects don't push back — because if nobody is calling your price expensive, you priced it for the wrong customer. The companies winning in 2026 didn't find a clever pricing trick. They just stopped underpricing.
- Price reflects value to the customer, not cost to deliver. If your product saves a team $50K, charging $20 isn't humble — it's a misalignment.
- Three tiers plus a custom Enterprise tier is the 2026 default. Anything more is decision fatigue.
- Usage-based pricing is no longer infrastructure-only. Hybrid base+overage is the dominant model for AI-native SaaS.
- Reverse trials beat both pure freemium and pure free trials on conversion. Default to reverse trial unless your product is viral or collaborative.
- Annual billing fixes cash flow, churn, and LTV in a single move. Add it after 60 days of monthly retention proof.
- Enterprise pricing is custom by definition. SSO, SOC 2, and DPA are the entry tickets, not optional add-ons.
- Raise prices yearly with 60-day notice and 12-month grandfathering. Churn from a clean price hike is under 2%.
- The biggest pricing mistake is doing nothing. Inflation alone erodes 5–10% of pricing power per year of inaction.
Building a SaaS or a product business and need a place to host your pricing, signup, and growth funnel without spinning up a website? Spin up a UniLink page in two minutes — link your Stripe, embed your pricing tiers, drop in social proof, and ship a working sales page before you finish reading this article.