Per-Seat Pricing: Examples & Companies

What is it

Per-Seat Pricing is a billing unit where the vendor charges a fixed fee per named user, regardless of how much each user consumes.

It is the unit SaaS was built on: the bill is seats × price, trivial to budget and to administer. 175 in-corpus companies use seats somewhere in their pricing — making it by far the most common billing unit in the corpus. But for AI products the seat is increasingly a floor rather than the whole bill, because AI cost and value scale with usage, not headcount.

The purest per-seat plays are seat-heavy productivity and vertical tools where usage variance between users is low: Superhuman at $25–$33/user/mo for email, Jasper at $59/seat/mo for marketing copy, and gated enterprise verticals like Harvey (legal, sales-quoted) and Glean (enterprise search). Consumer and developer products lean on the seat mainly to structure their team and enterprise tiers — Cursor Teams at $40/user, GitHub Copilot Business at $19/user, Notion AI Business at $20/seat.

The reason the seat still dominates is that most AI products are sold to teams by an admin — and headcount is the number that admin already budgets around. The reason it’s under strain is covered in How it works and Patterns observed: once an AI agent starts doing measurable work, “one price per human” stops describing the value.

How it works

The mechanic is simple — the bill is headcount times a rate. The design choices are about which seats count, what each seat includes, and how the tier ladder steps up:

A concrete worked example anchors the math. Take Tabnine’s Code Assistant Platform at $39/user/mo:

Unit math: A 25-developer team on Tabnine Code Assistant pays 25 seats × $39 = $975/mo (~$11,700/yr) — and that number does not move whether each developer accepts 50 completions a day or 5,000. Contrast Cursor Teams at $40/user, where the seat is a floor and a metered credit pool can push the real bill well above seats × $40 in a heavy month.

The structural test for whether a plan belongs on this page: if adding 1,000 AI actions per user does not change the bill, it’s per-seat. Once usage moves the bill, the seat has become the fixed half of a hybrid pricing model — which is why so many companies here also appear on the hybrid theme page. For the broader mechanics of choosing a meter, see the guide on choosing the right usage metric.

Companies using this

175 companies in the current corpus charge per seat somewhere in their lineup — the largest billing-unit cohort by far. They span seat-first productivity tools (Superhuman, Linear, Notion AI), vertical enterprise apps (Harvey, Glean, Writer, Gong), developer tools (Cursor, GitHub Copilot, Tabnine, Sourcegraph Cody), and creative/media platforms (Canva, Descript, Synthesia). The table below lists each company’s structure, whether it also meters usage, and its last verification date.

Patterns observed

Across the 175 seat-based companies, a handful of structural patterns repeat with striking consistency:

• The individual → team → enterprise ladder, with a ~2× team step. Almost every self-serve product uses the seat to structure a three- or four-rung ladder, and the jump from the individual plan to the team plan is frequently a clean doubling. Cursor Teams at $40/user is exactly 2× its $20 Pro plan; GitHub Copilot steps from Pro $10 to Business $19; Shortwave runs a full five-rung ladder from Free to $120/seat/mo. The markup funds shared admin (SSO, roles, audit) and reflects the buyer shifting from a personal card to an expense budget.

• The seat is now a wrapper for a usage allowance, not pure access. Very few AI companies sell a naked seat anymore. Notion AI folded its old standalone $8–$10/member AI add-on directly into its $10 Plus and $20 Business seats in May 2025; Glean pairs each Enterprise Flex seat with a pooled FlexCredit allowance; Canva attaches tiered “AI uses” to each seat and sells an AI Pass add-on at ~$100/user for 40× the allowance. The seat is the container; the allowance inside it is where AI cost hides.

• Annual billing is the headline price, and the discount is steep. Seat vendors quote the annual per-user rate first because it locks in commitment and improves cash flow. Otter.ai advertises Pro at $8.33/user annual versus $16.99 monthly (~51% off); Jasper Pro is $59/seat annual versus $69 monthly; Writer Starter is $29/user annual versus $39 monthly. The monthly price functions largely as an anchor.

• Seat minimums and inverted team pricing engineer land-and-expand. Fathom prices its 2-seat Team plan ($19/user) below its solo Premium plan ($20/user) — the second seat is cheaper than going solo, a deliberate expansion nudge. Glean reportedly enforces a ~100-seat minimum, using the floor to keep the motion enterprise-only.

• Transparency correlates with self-serve. The seat-based vendors that publish an exact number (Superhuman $25/$33, Tabnine $39/$59, Linear $10/$16) are self-serve; the ones that hide it (Harvey, Glean, Jasper Business) run sales-led motions where the seat price is a negotiation input, not a listed fact.

Counterexamples & variants

The most instructive counterexamples are companies that start from a seat and then watch the meter eat it. Descript is the sharpest: its Business plan is $50/seat/mo, but the company’s own analysis notes that a “$250/month” 5-seat team can reach ~$600 in a heavy month once media-hour and AI-credit top-ups are added — “the consumption layer can rival or exceed the seat layer.” That is the exact failure mode of per-seat framing: the seat is the number on the invoice line, but not the number that determines the bill.

Intercom is the clearest example of a seat model being displaced from within. It pairs a $29–$132/seat platform fee with $0.99 per Fin AI resolution, so on a high-volume support team the per-resolution line can dwarf the seat line — the seat becomes the floor rather than the ceiling of value capture. It appears here because seats remain a real billing unit, but its center of gravity has moved to per-resolution outcome pricing.

The gated-enterprise variant is where the published seat vanishes entirely. Harvey does not list prices; third-party estimates range from ~$1,200 to over $1,750 per lawyer per month, plus a separate platform fee. Glean similarly refuses to confirm a per-user number (trackers estimate ~$45–$50/user plus a ~$15/user AI add-on). These are still per-seat by structure, but the seat is a quote, not a price tag — a reminder that “per-seat” describes the unit, not the transparency.

Finally, a growing set of AI-native products treat the seat as vestigial. Copy.ai bundles 5 seats into its $29/mo Chat plan but prices its real GTM tiers ($1,000–$3,000/mo) on workflow credits, not seats — the seat count is a rounding detail next to the credit meter. When usage variance between users is high and an agent does the work, the seat stops being the right unit at all.

What this means for buyers vs vendors

For buyers

Per-seat is the easiest model to budget — headcount × price is a number a CFO can approve in one line — but that simplicity hides three traps. First, provisioned-seat billing means you pay for idle users; audit actual active usage before renewal and true-down where the contract allows. Second, “AI included in the seat” is rarely unlimited — check whether the seat carries a credit or “AI uses” allowance (as with Notion AI and Canva) and what overage costs, because that is where a predictable seat bill becomes a variable one, as Descript customers discovered. Third, on sales-led seat deals (Harvey, Glean, Gong), the per-seat price and the platform fee are separate negotiation levers — get both in writing, and benchmark against the usage-invoicing and billing-cycle norms before signing an annual commit.

For vendors

The seat fits when usage variance across your users is low, the buyer is an admin who budgets by headcount, and your marginal cost per user is roughly flat — classic for productivity and collaboration tools. It stops fitting the moment AI does measurable work per action, because a flat seat over-charges your light users (who churn) and under-charges your power users (leaving money on the table). The corpus pattern is unambiguous: nearly every AI-first company here keeps the seat as a floor and adds a usage or outcome layer on top, converging on a hybrid pricing model. If you launch seat-only, design the ladder deliberately (a ~2× team step, an annual anchor, an optional AI allowance you can meter later) so you have somewhere to grow the account. For the underlying mechanics of layering a meter onto a seat, the guide on choosing the right usage metric and the Cursor pricing calculator both model the seat-plus-usage math directly.

Explore this theme in the knowledge graph →