How to design pricing tiers for product-led growth motions in 2027

Two ways to slice a PLG ladder: gate the feature or gate the volume

Every pricing conversation in a product-led company eventually collapses into one binary, and it is not "how much." It is *what stops the user*. Option A is the feature gate: the free tier ships a deliberately incomplete product, and the paid rung restores the missing organs — search, export, integrations, version history. Option B is the volume gate: the free tier ships the *complete* product with a hard ceiling on quantity — projects, issues, videos, seats, storage, events, runs.

The feature gate is seductive because it is easy to build and easy to explain to a board. You draw a matrix, you put checkmarks in columns, and the upgrade reason looks obvious on the page. The problem is that it breaks the thing PLG runs on. A product-led motion depends on a user reaching a genuine moment of value alone, without a demo, without an AE, without a champion translating the roadmap. If the free tier cannot complete one real workflow end to end, that moment never lands. The user does not upgrade to find out whether the product is good — they leave and try the next tab. Teardown work across self-serve companies has repeatedly found feature-gated free tiers converting materially worse than limit-gated ones, often by a factor of roughly two.

The volume gate inverts the psychology. The user succeeds first, builds habit, imports data, invites a colleague, wires an integration, and *then* hits a wall. At that point the upgrade is not a leap of faith — it is the removal of friction from something already working. Loom's cap on videos per person, Calendly's single-calendar limit, Linear's caps on issues and teams are all the same architecture wearing different clothes: full product, hard ceiling, obvious moment of pain.

There is a third pattern worth naming because it is adjacent and frequently confused with both: the *time* gate, a classic trial. Trials are not a bad instrument — card-required trials convert far better than any freemium tier, often several times better — but they solve a different problem. A trial is a sales tool for a considered purchase. Freemium is a distribution tool for a viral one. A trial with a credit card up front is closer to a self-serve SMB motion than to product-led growth, and companies that quietly swap freemium for a card-gated trial to fix conversion metrics usually find they fixed the ratio by shrinking the numerator and the denominator together. Top-of-funnel volume collapses, the conversion rate looks beautiful in the board deck, and net new self-serve revenue is flat.

The honest framing is that these are not exclusive. The mature 2027 shape is a *limit-gated free tier with an embedded time-boxed trial that triggers on cap-hit*. The user works free, hits the wall, and gets seven days of the paid rung right at the moment of maximum felt need. Conversion on that cap-triggered trial runs dramatically higher than a blanket trial offered at signup — the difference between roughly 40% and low double digits in published trial-conversion data. Same product, same price, radically different timing.

One caveat on the volume gate: it only works if the metric you meter is the metric the user experiences as value. Metering something invisible — API calls behind a UI, background sync jobs, internal object counts — produces a wall the user cannot anticipate and reads as a bug. Pick the noun the user already counts in their head.

Deciding which gate to pull, and where the rungs land

The decision is mechanical once you answer three questions in order. First: can one person get real value alone? If yes, freemium is viable and your free tier should be limit-gated. If no — if the product only makes sense with three colleagues in it — you do not have a freemium product, you have a team-trial product, and you should ship a no-card trial with in-product onboarding instead. Second: is there a *countable* unit of value the user already thinks in? Videos, issues, dashboards, meetings, contacts, documents, workflow runs. If yes, that is your cap. If no, you are back to feature gating by default and you should expect worse conversion and plan for a sales-assist layer to compensate. Third: does the product carry meaningful marginal cost per unit of usage — inference, storage, bandwidth, egress? If yes, you need a usage meter on the paid rungs regardless of what you do at the free tier.

Where the free ceiling actually lands is an empirical question, not a taste question, and there are three tests worth running before any change ships. The ceiling test: of users still active on day 14, what share have consumed at least half the free allowance? Somewhere in the 35-50% band is healthy. Below about 25% and your free tier is a product you are giving away rather than a funnel. Above about 70% and you are strangling people before habit forms — they churn instead of upgrading, and the two outcomes look identical in a conversion dashboard while being opposite in cause.

The workflow completion test: can a free user finish one complete job, start to finish, without hitting a wall? Not a demo of a job — the actual job, saved, shared, exported. If not, expect conversion to sit in the low single digits permanently regardless of how you price the rungs above.

The multiplayer test: does the first invite require money? The pattern that works across Loom, Notion, Figma, and Linear is to let free users invite viewers freely but require the paid rung for comment and edit. Viral distribution stays free; collaboration is the toll booth. Upgrade-event analysis in PLG teardowns consistently puts "second person needs to edit" among the largest single triggers, on the order of a third of all upgrades.

A note on what these tests do *not* tell you. None of them price the rung. They tell you where the wall goes. Rung price comes from the value of the workflow to the buyer's role, benchmarked against the nearest substitute — which in 2027 is increasingly a general-purpose AI assistant plus a spreadsheet, not a direct competitor. That substitute has reset the floor for a lot of light-weight tooling and is the quiet reason many $8-$12 tiers have stopped converting.

The numbers behind each rung

Free. Expect a freemium free-to-paid conversion somewhere around a mid-single-digit median across the market, with top-quartile companies landing roughly 8-12% and the bottom quartile down at 1-2%. Visitor-to-paid tells a cleaner story about motion choice: freemium in the low teens, no-card trials somewhat higher, card-required trials dramatically higher still — the tradeoff being that each step up the conversion ladder shrinks the top of the funnel. Do not benchmark your freemium rate against a card-trial company's rate and conclude you are broken.

Pro / Team, $15-$25 per seat per month. This is the collaboration rung and it should be priced to be expensable without approval — the number a manager pays on a corporate card without opening a procurement ticket. Anchors in the market sit around the mid-teens per seat: Figma's professional tier, Loom's business tier, Notion's paid team tiers, Linear's standard tier all cluster in the $10-$18 range depending on billing period. Annual prepay discounts of roughly 17-20% (the "two months free" convention) are close to universal and are worth more than they cost — they convert monthly churn risk into a twelve-month revenue-recognition window and materially improve CAC payback math.

What belongs on this rung: unlimited the-thing-free-capped, shared workspaces and libraries, the common integrations (chat, cloud storage, source control), a meaningful version-history window in the 30-90 day range, and email support with a next-business-day expectation. What must stay locked: SSO, SCIM provisioning, audit logs, granular admin roles, API rate limits above a modest ceiling, and AI credits beyond a starter pool. Leaking SSO down to the Pro rung is the single most common self-inflicted wound in PLG pricing — it removes the only forcing function that pulls a 200-person customer into a conversation with a human.

Business, $25-$45 per seat per month plus a meter. The seat component pays for the org-readiness features. The meter pays for the marginal cost. Vercel's per-seat plus bandwidth-and-compute structure, PostHog's per-event pricing, and Notion's credit add-on layered on its business tier are all the same architecture: a predictable floor plus a variable ceiling. Target gross margin on the metered component in the 65-75% band after cost of goods, and review that ratio quarterly, because the underlying model costs move faster than your pricing page does. Passthrough markups on inference in the 1.3-1.5x range are common and defensible; below 1.2x you are running a payments business with none of the volume.

The reason the meter is non-negotiable when marginal cost exists is concentration. Usage in AI-augmented products is savagely long-tailed — a small share of accounts drives the overwhelming majority of inference spend. On a pure seat price, that top few percent silently consumes a large fraction of gross margin, and because they are usually your happiest, most-referenced customers, nobody wants to be the person who raises it in a QBR.

Enterprise, quoted. No public number. A platform floor in the $50K-$150K annual range depending on category, always including SSO with SAML and SCIM, audit-log export to a SIEM, a data processing agreement with a published sub-processor list, a named CSM, an uptime SLA with service credits, data residency options, and a security review the vendor actually staffs. The floor is not greed — it is the price at which a human-touch motion pays for itself once you load in solutions engineering, security questionnaires, procurement cycles, and legal redlines.

Discount governance. Write the matrix down before you need it: single-digit discounts at AE authority, up to roughly 20% at VP Sales, 20-30% requiring CRO and Deal Desk sign-off, anything above that escalating to finance. Companies operating without a written matrix routinely see list-to-net erosion in the low-to-mid twenties percent; those with one hold it closer to the low teens. That gap is pure margin and it costs nothing but a document and the discipline to enforce it.

Adjacent motions the ladder has to survive

A pricing ladder does not live alone in a spreadsheet — it collides with four neighboring systems, and most failed repricings fail at one of these seams rather than at the price itself.

Sales-assist handoff. The moment you add a Business rung with a usage meter, you have created a queue of accounts that are too big for self-serve and too small for an AE to chase on instinct. That queue needs a product-qualified-lead definition, and the definition has to be behavioral rather than firmographic: seats added in the last 30 days, cap-hit events, admin-role invitations, SSO documentation page views, billing-page visits from a non-billing user. Route those to a human; leave everything else alone. The failure mode is an AE reaching out to a 3-person team that was going to convert on its own, which converts a free acquisition into an expensive one.

Finance and revenue recognition. Hybrid pricing complicates the close. Seat revenue recognizes ratably; metered revenue recognizes as consumed; prepaid credits create a deferred-revenue balance with a breakage policy someone has to decide. If credits expire, say so on the pricing page in plain language, and pick an expiry that is generous enough not to feel punitive — annual is common. Finance also needs an unbilled-usage view, because with a soft-overage policy you are extending credit to customers between invoice cycles whether you think of it that way or not.

Support economics. Every rung implies a support load. Free users generate tickets at a nontrivial rate and generate no revenue, which is fine as a marketing cost as long as it is *measured* as one. Community forums, searchable documentation, and in-product guidance are the levers. The rule of thumb worth holding: if free-tier support cost per active free user exceeds a couple of dollars a month, either the product is confusing or the free tier is too broad.

Procurement. On the buyer side, the last two years have hardened the renewal conversation considerably. Buyer-initiated downgrade requests have climbed sharply, average renewal discounts have crept up, and any procurement lead at a mid-size company now opens with a request for usage data before discussing seats. This is actually an argument *for* the hybrid model rather than against it: a vendor who can show a per-seat utilization report and a consumption trend enters that conversation with evidence. A vendor selling a flat seat price with no telemetry enters it with a story, and stories lose to spreadsheets.

There is a fifth seam that gets less attention: partner and reseller pricing. If any portion of revenue flows through marketplaces or resellers, the metered component has to be expressible in their billing systems, which are frequently seat-shaped and hostile to consumption. Cloud marketplaces have improved on this, but private-offer plumbing for metered products still requires deliberate work rather than a checkbox.

Sequencing the migration without breaking the base

Repricing a live product-led business is the highest-variance operation a revenue org runs, because you are changing the price of something thousands of people already bought while they watch. Sequence it over roughly a quarter.

First month — audit. Pull eighteen months of billing data and cohort it by acquisition month. For each cohort, compute free-cap hit-rate by day 14, time to second collaborator, average revenue per account by rung, net revenue retention by rung, and the distribution of usage against marginal cost. That last one is the most important and the most frequently skipped: rank every paid account by cost-to-serve and look at the top 5%. If the shape is a cliff rather than a slope, you have a meter-shaped problem and no amount of seat-price tuning will fix it. Convene the owners in the same room — the revenue lead, the growth product lead, RevOps, deal desk, finance, and marketing — because a pricing change that RevOps does not instrument is a pricing change that cannot be measured, and one that marketing does not have copy for is a pricing change that leaks through a support queue.

Second month — design and instrument. Draft the ladder. Wire the PQL events into the customer data platform and route them to whatever sales-assist tooling you use. Deal desk authors the discount matrix and the enterprise SKU list. Finance models three scenarios — base, an upside where free conversion improves by two points, and a downside where it drops by one — and stress-tests each against cash collection timing, not just booked revenue. Legal drafts the grandfather clause. Do this work *before* touching the pricing page, because the most expensive version of this project is the one where the page changes on Monday and the billing system learns about it on Thursday.

Third month — launch and hold. Run the new ladder against a slice of new traffic — 10% is a reasonable starting point — for at least two full weeks so weekly seasonality washes out. Measure visitor-to-free, free-to-paid, paid-tier mix shift, average revenue per account, and CAC payback. Roll to everyone only if blended ARPA improves by a clear double-digit percentage *and* visitor-to-paid has not degraded more than a few points. If ARPA is up but top-of-funnel is down 15%, you have not repriced — you have quietly become a different company with a smaller market.

Grandfathering is not optional. Give existing paid customers twelve months on their current terms and tell them the date in writing on day one. The retention difference between grandfathering and force-migrating is large enough to swamp the incremental revenue from migrating early — companies that grandfather generally hold net revenue retention near or above 108%, while force-migrations have been observed dropping it into the mid-90s. You can win the pricing change and lose the year.

Overage policy. Three choices, in order of preference. Soft overage — bill at list with warning emails at 80%, 100%, and 120% of the allowance — is best for paid tiers because it never interrupts work. A hard cap with an upgrade prompt is right for the free tier, where the interruption *is* the conversion mechanic. Throttling to a degraded rate is the worst of the three: it makes the product feel broken without ever presenting a clear decision, and it drives churn among exactly the heavy users you want to upsell.

Tooling. Billing and metering is the one place not to build in-house — the general-purpose billing platforms cover seat-shaped products well, and the newer usage-metering vendors exist specifically because sub-second metering against AI workloads broke the older architectures. Pair that with a customer data platform for event capture, a reverse-ETL path into the CRM, and an experimentation tool for the traffic split. Comp tooling matters too, because a hybrid model breaks quota logic: pay on seat expansion and consumption growth separately, or reps will optimize for the one that is easier to close and let the meter drift.

Comp implications, briefly. A product-led assist rep should carry no quota on free-to-paid — self-serve owns that, and paying a human for it just taxes an automated motion. Quota belongs on Pro-to-Business and Business-to-Enterprise crossings. Headcount scales predictably: a RevOps lead and a deal-desk analyst around the $10M mark, a dedicated PLG operations owner for PQL routing around $25M, and a standalone pricing-strategy role reporting to the revenue leader past $50M. That last hire is usually made a year later than it should be.

Related questions

Should the free tier ever require a credit card?

No. A card requirement converts freemium into a trial and collapses top-of-funnel volume, which is the entire reason to run a product-led motion. Require the card at the paid rung, and use a cap-triggered in-product trial to capture intent at the moment of friction instead.

How do you price when AI inference cost is unpredictable?

Decouple the meter from the dollar with credits or units, so you can adjust the credit-to-cost ratio without renegotiating contracts. Review margin on metered revenue quarterly, target 65-75% after cost of goods, and disclose the conversion rate plainly on the pricing page.

What breaks first when SSO leaks into a lower tier?

The enterprise motion. SSO is the natural forcing function that brings IT and procurement into the conversation at a few hundred seats. Once it is available for $18 a seat, large accounts self-serve indefinitely and never trigger a security review, a DPA, or a floor-priced contract.

Is per-seat pricing dead in 2027?

No, but it is now the minority default. Pure per-seat still fits products where the unit of value is genuinely a human collaborating — design tools, meeting tools. Anywhere marginal cost scales with usage rather than headcount, seat-only pricing quietly transfers margin to your heaviest accounts.

How long should a pricing test run before rolling out?

At least two full weeks on 10% of new traffic, so weekly seasonality and payday cycles wash out. Watch blended ARPA, visitor-to-paid, and tier mix together — a single improving metric almost always hides a compensating decline somewhere else in the funnel.

FAQ

What is the most important thing to gate in a free tier?

Gate volume of the single unit a solo user already counts in their head — videos, issues, meetings, documents, projects. The free tier should deliver the complete product against a hard ceiling, not an incomplete product with organs removed. Limit-gated free tiers have consistently converted better than feature-gated ones in published teardown work, because the user reaches genuine value before hitting the wall, and the upgrade removes friction from something already working rather than purchasing a promise.

What free-to-paid conversion rate should a PLG company target?

Median freemium conversion sits in the mid-single digits; top-quartile companies land roughly 8-12% and bottom-quartile companies sit at 1-2%. The strongest predictor is whether the free ceiling maps to a real workflow bottleneck within about two weeks of activation. Benchmark against companies running the same motion — comparing a freemium rate to a card-required-trial rate produces false alarm every time.

How should the Business tier differ from Pro?

Business adds the org-readiness layer — SSO, SCIM provisioning, audit logs, granular admin roles, higher API limits — at $25-$45 per seat, and layers a usage meter on top for anything with real marginal cost: inference, storage, bandwidth, events. Pro is priced to be expensable on a corporate card. Business is priced to be justified in a budget line, which is a different conversation with a different buyer.

When should an account move to a quoted Enterprise contract?

When the deal crosses the platform floor, typically $50K-$150K annually depending on category, or when the buyer's requirements exceed what self-serve terms can satisfy — custom data residency, a negotiated DPA, a named security review, an uptime SLA with credits, multi-year invoiced billing. The floor exists because a human-touch motion costs real money in solutions engineering, security questionnaires, and legal redlines.

Who owns the pricing ladder inside the company?

The revenue leader, the growth product leader, and RevOps co-own the ladder as a system; deal desk owns the Business-to-Enterprise crossover and the discount matrix. Finance owns the margin model on the metered component. The failure mode is single ownership — a ladder designed only by product ignores discounting reality, and one designed only by sales quietly leaks enterprise features downward to close this quarter.

What happens to existing customers when the ladder changes?

Grandfather them for twelve months on current terms and communicate the end date in writing on day one. The retention gap between grandfathering and force-migrating is wide enough to outweigh the incremental revenue from migrating early — force-migrations have been observed pushing net revenue retention down into the mid-90s while grandfathered cohorts hold above 108%.

Sources

• OpenView Partners — Product Benchmarks — https://openviewpartners.com/product-benchmarks
• Kyle Poyar, Growth Unhinged — Your Guide to PLG Benchmarks — https://www.growthunhinged.com/p/your-guide-to-plg-benchmarks
• Stripe — A Guide to PLG Pricing Models — https://stripe.com/resources/more/plg-pricing-models
• Pocus — The Ultimate Guide to Pricing and Packaging for PLG — https://www.pocus.com/blog/the-ultimate-guide-to-pricing-and-packaging-for-plg
• Bessemer Venture Partners — State of the Cloud — https://www.bvp.com/atlas
• a16z — Enterprise and SaaS research — https://a16z.com/enterprise/
• ChartMogul — SaaS Benchmarks and Retention Reports — https://chartmogul.com/reports/
• Metronome — Usage-Based Billing Resources — https://metronome.com/blog
• Vendr — SaaS Buying and Procurement Research — https://www.vendr.com/blog
• Reforge — Growth and Activation Research — https://www.reforge.com/blog

Related on PULSE

• [Revenue Architecture for Product-Led Growth B2B SaaS in 2027 — The Complete Operator Guide](/knowledge/ra0029)
• [How to build a forecast roll-up across multiple selling motions in 2027](/knowledge/ra0300)
• [Product-Led Sales Assist Handoff Design in 2027](/knowledge/ra0411)
• [Building RevOps for B2B Data Providers: API Access Tiers, Usage Caps, and Data Licensing](/knowledge/ra0588)
• [Customer Segmentation Tiers for SaaS in 2027](/knowledge/ra0222)
• [Revenue Architecture for Vertical SaaS for Salons + Spas in 2027 (Segment Tiers, Comp, NRR)](/knowledge/ra0222)