How to build a freemium pricing strategy

A freemium pricing strategy works when three things are true: your product has a huge total addressable market, your marginal cost per free user is near zero, and there's a clear upgrade trigger that hits within weeks of activation. Build a free tier that delivers a complete narrow workflow, one paid tier that unlocks the obvious next step, and a contact-us tier for buyers who need procurement. If any of those three foundations is missing, charge from day one instead.

When freemium actually works

Freemium is not a default. It's a specific go-to-market motion that fits a specific shape of company.

The shape: huge TAM (you need millions of free users to find your tens of thousands of paid ones), low marginal cost per free user (storage, compute, and support don't blow up your margin), a viral coefficient (free users bring more free users without paid acquisition), and an integration moat (the product gets harder to leave the longer you use it).

Notion fits all four. So does Figma. So does Cal.com. So does PostHog. None of these companies sell freemium because it's trendy. They sell freemium because their product economics force that shape.

The math: if your free tier costs you $0.20/month per user in infrastructure and your conversion rate is 3% to a $10/month plan, your blended ARPU across all signups is $0.30 minus $0.20, or $0.10/user/month. That works if you can acquire users at $1 or less. If your free tier costs you $4/month in infra (think AI inference, video hosting, large databases), the same 3% conversion math collapses.

When freemium fails

Freemium fails predictably. The pattern shows up across hundreds of dead startups.

High per-user infra cost. AI products burn $1 to $20 of inference per active free user per month. Unless your conversion sits above 8% and your paid plan exceeds $50/month, you bleed cash on every signup. Most AI startups should be on 14-day free trials, not freemium.

B2B without a bottoms-up motion. Freemium needs individual users to onboard themselves and drag their teams in. If your buyer is a CFO and your user is the same CFO, there's nobody to evangelize internally. Sell directly. Skip freemium.

No clear upgrade trigger. If a user can run their entire workflow on the free tier forever, they will. The trigger has to be inevitable: hitting a usage cap, needing to add a teammate, needing a feature that only matters once they're committed. Notion's trigger is "I want to invite my team." Figma's is the same. Cal.com's is "I need a team scheduling page." Without a trigger, you're a charity.

Real conversion benchmarks

Founders quote freemium conversion rates from blog posts that haven't been updated since 2018. Here are the current numbers, validated against ChartMogul's 2026 study of 200 SaaS products and First Page Sage's 2026 benchmarks.

Motion	Bad	Good	Great
B2C freemium	<1%	2-3%	4-5%
B2B PLG freemium (self-serve)	<3%	5-8%	10-15%
Free trial (opt-in, no CC)	<5%	8-10%	15%+
Free trial (opt-out, CC required)	<20%	25-35%	40%+
Sales-assisted freemium (B2B)	<3%	5-7%	10-15%

A few uncomfortable truths buried in those numbers.

The median free-to-paid conversion across all products in 2026 is 8%, but that median masks bimodal distribution. Half of products convert under 3%. The other half convert above 12%. There is almost no middle. This means freemium is binary: it either works for your product shape or it doesn't.

Companies that implement role-based feature gating (different limits for individual contributors vs admins vs viewers) lifted conversion to 5.1% on average, adding roughly $890K in incremental ARR. Most founders gate the wrong things, which is the next section.

Pricing-page structure: free, paid, contact-us

The three-tier pricing page is the single most copied pattern in SaaS, and most founders implement it wrong. Here's the structure that works.

Free tier. One user. Complete access to one core workflow. Hard limit on a single dimension that scales with value (projects, seats, API calls, automations, document count). The limit must be high enough that a real user can complete their job-to-be-done end-to-end at least once. Otherwise the free tier is a demo, not a freemium product.

Paid tier. One price, one tier. $10 to $50/month for B2C, $20 to $50/seat/month for B2B PLG. This tier removes the hard limit and adds team features (multiple users, shared workspaces, basic admin controls). Self-serve checkout. Stripe Billing handles it in 20 lines of code.

Contact-us tier. Enterprise. SSO, SCIM, audit logs, custom contract terms, security review, dedicated support. No price listed. This tier exists for two reasons: procurement needs a human to negotiate with, and you need a way to charge $30K annual contracts without scaring off the $20/month buyer.

Notion runs this exact structure. So does Linear. So does Figma. So does PostHog (with a twist: usage-based rather than per-seat for the paid tier, which is a better fit for analytics tools). The structure works because it maps to three distinct buyer types: the individual user, the small team, and the enterprise.

If you build a build vs buy framework before launching, you'll see that the pricing page itself is one of the highest-leverage build decisions you make. A good build vs buy framework for founders will surface whether your tier structure can support the unit economics you need.

Anti-patterns that kill conversion

Five mistakes show up in roughly 80% of freemium pricing pages we audit.

Giving away too much. If a solo founder can run their business on your free tier forever, they will. The free tier should solve one user's problem, not a team's. Slack got this right by capping message history. Figma gets it right by capping pages on the free tier.

Paywalling essentials that should be free. API access, basic exports, mobile apps, email notifications. Charging for these signals you don't trust your product. The user assumes the paid tier is also missing the basics. Trust evaporates.

Tier names that confuse. "Pro" vs "Business" vs "Premium" vs "Plus" vs "Team" vs "Enterprise" vs "Scale". Pick three names. Make them obvious: Free, Pro (or Team), Enterprise. Don't invent a new vocabulary.

Hiding the price. If your paid tier has a price, show it. "Contact for pricing" on a $30/month tier kills conversion by 60-80% versus a clearly listed price. Reserve "contact us" for the actual enterprise tier where it's expected.

Per-seat billing on a usage product. If your product is fundamentally about volume (logs, events, storage, API calls), per-seat pricing punishes your best customers. Use usage-based or hybrid pricing. PostHog and Vercel both moved to usage-based after starting with per-seat and saw retention jump.

Real examples worth studying

Five companies that got freemium right, with the specific mechanic each one nailed.

Notion. Free tier is unlimited blocks for one user, capped at 7-day version history and small file uploads. Upgrade trigger is inviting a teammate. Conversion lift came from making team workspaces a core part of the product, not an upsell.

Figma. Free tier is unlimited personal files, capped at 3 collaborative files. Upgrade trigger is the fourth project. The genius move was making files free to view forever (huge viral coefficient) but capping editing seats.

Linear. Free tier is 250 issues for unlimited members. The cap forces growing teams to upgrade exactly when they're starting to feel the value. Linear is also the cleanest pricing page in SaaS: three tiers, clear prices, no marketing copy.

PostHog. Free tier is 1M events/month plus 5K session recordings, with full feature access. Upgrade is purely usage-based. The bet: developers who exceed 1M events are running real production traffic and have a budget. Conversion at scale is in the 12-15% range.

Cal.com. Free tier is unlimited personal scheduling, paid tier ($15/month) adds team scheduling and integrations. Open-source with self-host option. Cal.com proves freemium works even in a category (scheduling) that Calendly already won, because their differentiation is openness and developer-first integrations.

What they have in common: each company can articulate the upgrade trigger in one sentence. If you can't do that for your product, your pricing strategy isn't done yet.

Building the strategy: a four-week plan

Here's the sequence we recommend for founders who've validated demand and are ready to ship pricing.

Week 1: validate the upgrade trigger. Talk to 15 customers (paid or trial) and ask: "What was the moment you decided to pay?" If you don't get a consistent answer in 5 words or fewer, your trigger isn't real yet. Fix the product before the pricing page.

Week 2: define limits. Pick one dimension to cap on the free tier (projects, seats, events, integrations, automations). The cap should be hit by 60-70% of active users within 30 days of signup. Anything looser and your conversion stalls. Anything tighter and your free tier feels like a trial.

Week 3: ship the page. Three tiers: Free, Pro, Enterprise. Stripe Billing for self-serve. A /pricing route that loads in under 200ms. Use Vercel or Netlify hosting and a static page if you can. Pricing pages are the highest-converting page on most SaaS sites. Don't over-engineer them.

Week 4: instrument and iterate. Track free-to-paid conversion weekly, segmented by acquisition channel. Track time-to-upgrade. Track cohort LTV by acquisition month. PostHog or Mixpanel handles this in a day. If your week-4 numbers don't roughly match the benchmarks above, the issue is almost always the upgrade trigger, not the price.

This is a lot of execution work, especially for non-technical founders. Building a Stripe-billed pricing page, a usage-tracking pipeline, and a feature-gating system is straightforward but unforgiving. One bug in the gate and either everyone's on the free tier forever or paying customers get locked out.

If you're doing this without a dedicated engineer, every Cadence engineer is AI-native by default and ships pricing-page work end-to-end. A mid-tier engineer at $1,000/week typically ships a Stripe-billed three-tier pricing page with feature gating in 5-7 days. A senior at $1,500/week handles the same scope plus usage-based billing and edge cases (proration, plan changes, dunning) in the same window.

Common founder mistakes when launching freemium

A few patterns we see repeatedly when reviewing pricing strategies.

• Launching freemium before product-market fit. Freemium amplifies whatever's true about your product. If your product has weak retention, freemium gives you more weak retention. Validate paid first. Switch to freemium when you've proven the product is sticky. The same logic applies to your B2B SaaS validation experiments before you commit to a model.
• Optimizing pricing instead of activation. Most freemium products with bad conversion have an activation problem, not a pricing problem. If 70% of free signups never reach the "aha" moment, no pricing change saves you. Fix activation first.
• Adding a fourth tier. "Starter" between Free and Pro almost always hurts. It splits your conversion funnel, adds support complexity, and confuses buyers. Three tiers. Stop.
• Hiring a senior to design pricing when a mid handles it. Pricing-page implementation is well-understood territory. A mid engineer with Stripe experience ships the same code a senior would, in the same time, for half the cost.
• Skipping the MVP scope discipline. Pricing without an MVP is a research project. Pricing your MVP correctly the first time is rarely possible; pricing it badly the first time is recoverable. Ship something. Iterate.

If you want a structured way to evaluate the build/buy/book decision on pricing infrastructure specifically, decide your next move with the Build/Buy/Book tool. It runs the trade-off math in 90 seconds.

The bigger picture

Freemium pricing is a narrow tool. It works for products with massive TAMs, low marginal costs, and clear upgrade triggers. It fails predictably for products without those traits. Most B2B SaaS companies should start with a 14-day free trial and add freemium later, only if the unit economics support it.

The mistake founders make most often is treating freemium as a marketing decision instead of a product-economics decision. Slack didn't choose freemium because they wanted virality; they chose it because their per-user marginal cost was near zero and their network effects were obvious. Reverse-engineer that math for your own product before you commit.

Building the pricing page is one of the lower-risk, higher-leverage engineering projects on a founder's roadmap. If you're ready to ship one and don't have an engineer, book your first Cadence engineer with a 48-hour free trial. Weekly billing, no notice period, replace any week.

FAQ

What's a good freemium-to-paid conversion rate?

For B2C freemium, 2-3% is good and 4-5% is great. For B2B product-led growth freemium, 5-8% is good and 10-15% is great. The 2026 median across all SaaS products is 8%, but the distribution is bimodal: products either convert above 8% or below 3%, with little in the middle.

Should I do freemium or a free trial?

Free trials convert better in absolute terms (median 8.9% for opt-in, 31.4% for opt-out with credit card required). Freemium wins when your TAM is huge and your marginal cost per free user is near zero. If your infrastructure cost per active free user exceeds $1/month, default to a 14-day free trial instead.

How many tiers should my pricing page have?

Three. Free, one paid tier (Pro or Team), and an enterprise contact-us tier. Adding a fourth tier almost always hurts conversion by splitting the funnel and confusing buyers. Notion, Linear, Figma, and PostHog all run three-tier structures.

What's the most common reason freemium fails?

No clear upgrade trigger. If a user can run their entire workflow on the free tier indefinitely, they will. The trigger has to be inevitable within 30 to 60 days: a usage cap, a teammate invitation, a feature that only matters once you're committed. Without a trigger, freemium is just a charity.

How long does it take to build a freemium pricing page?

A mid-tier engineer with Stripe experience ships a three-tier pricing page with self-serve checkout, feature gating, and basic usage tracking in 5 to 7 days. Add another week for usage-based billing or proration logic. A senior engineer at $1,500/week handles the full stack including edge cases in the same 5 to 7 days.