Seat-Based Pricing: Examples & Companies

What is it

The math is the part everyone already understands: budget equals headcount times the per-user rate. A team of 10 on a $30/user/month plan pays $300/month, full stop. That predictability is exactly why seats were the SaaS default for two decades — finance can forecast the line item from the org chart, and procurement can reason about it without a usage model. In the corpus this remains the dominant pattern for vertical enterprise AI, where each user is a professional doing roughly the same job at roughly the same intensity.

The model assumes consumption is uniform enough across users that the vendor can absorb the variance internally. That assumption holds well for tools sold to a defined professional role. Harvey prices its legal AI platform purely on seats — a sales-led, gated per-user quote with no usage meter — because a lawyer’s workload doesn’t swing 100x day to day. Spellbook does the same for AI contract review inside Word, Numeric for month-end-close automation in accounting, and Freed and Heidi Health for ambient clinical scribing, each billing per clinician.

Where the model gets interesting in 2026 is at the edges. Frontier labs that built per-token API businesses — OpenAI, Anthropic, and Perplexity AI — all added per-seat team plans on top, because the buyer of a team subscription wants a headcount-based bill, not a token meter. And horizontal AI tools that started seat-only increasingly bolt on a usage component to stop subsidising power users, which is the boundary between this theme and seat-plus-usage pricing.

How it works

Worked example. Take Numeric’s published tier at $30/user/month. An eight-person finance team pays 8 × $30 = $240/month, or $2,880/year before any annual discount. Add a ninth analyst and the bill rises by exactly $30 — linear, predictable, and entirely decoupled from how many month-end closes the team actually runs. That decoupling is the model’s whole promise and its central weakness: the bill tracks who has access, not what gets done.

Contrast that with Spellbook’s segment packages, which start around $149–$240 per user per month but are quote-only and frequently sold in 10-seat blocks. A 10-seat firm could land anywhere from roughly $1,500 to $2,400 per month depending on the negotiated rate — the seat is still the unit, but the price per seat is a negotiated variable, not a rate card. Use the pricing calculator to model seat counts against tier breakpoints for the companies that publish their rates.

Companies using this

Patterns observed

The pure seat is increasingly a base layer, not the whole model. Many of the strongest examples list seat-based alongside freemium, hybrid, or seat-plus-usage. Glean pairs per-user Enterprise Flex seats with pooled FlexCredits; Codeium and Qodo layer credit and request limits on top of their seat tiers; Gladly and Kustomer charge per-Hero or per-seat packages and then add a per-conversation AI outcome charge. The seat anchors the base of the contract — who has access — while a usage layer captures the variable cost. This is the live edge with the credit-currency-abstraction trend: credits meter what you do, seats meter who does it, and the two coexist rather than compete.

Freemium-plus-seat is the PLG on-ramp. Fathom, tl;dv, and Heidi Health run a free-forever per-user tier that converts to paid seats as teams grow. The seat is doing double duty here: it’s the value metric and the natural expansion lever, because adding a teammate is the most legible upgrade a self-serve buyer can make. See the introduction to usage-based pricing for how seat expansion compares to consumption expansion as a growth engine.

Seats just reclaimed the lead as the modal billing unit. At the 158-company corpus, seats appear in 78 billing-unit arrays versus credits’ 70 — reversing months of credit dominance. The shift isn’t a change in any one company’s pricing; it’s the dataset broadening into seat-heavy vertical and horizontal SaaS. The takeaway: credits won the usage layer, but seats remain the most common unit overall because nearly every team-oriented product still bills the base on headcount.

Counterexamples & variants

The customer-support variant: seat plus outcome. Gladly and Kustomer show a distinct hybrid where the human-agent seat persists but the AI work is billed by outcome — per assisted or engaged conversation — not per seat. Here the seat prices the human in the loop while the outcome charge prices the AI’s labour, a structure that only makes sense because AI conversation volume is decoupled from agent headcount. This is the same logic the outbound-seat-plus-credits trend documents in sales-tech: a seat base with a consumption layer bolted on.

The opacity variant: a seat you can’t see. Spellbook and Harvey charge per seat but publish no rate. The model is still seat-based — the unit is the user — but the price is a sales-led negotiation, often sold in fixed seat blocks. This protects pricing power in high-ACV legal AI, where a published rate would anchor every subsequent negotiation downward. The trade-off is friction: a buyer can’t self-qualify, and the lack of an anchor price slows the top of the funnel.

What this means for buyers vs vendors

For buyers

Seat-based pricing is the easiest model to forecast and the easiest to over-buy. Budget is headcount times rate, so the line item is predictable — but provisioned seats drift above active seats over time, and you pay for both. Audit utilisation before each renewal: if half your Glean or Jasper seats are dormant, you’re funding access nobody uses. When a vendor bundles usage inside the seat (as Surfer SEO does with document allotments), check the cap — a “seat” with a low usage ceiling is a metered plan in disguise, and your real cost depends on consumption, not headcount. For quote-only vendors like Spellbook, expect to negotiate the per-seat rate and the minimum block size; there is no rate card to defend against, so come with comparables. The usage-invoicing and billing-cycles guide covers how seat true-ups and mid-term adds get billed.

For vendors

The seat is the cleanest value metric you can pick when consumption is uniform — it maps to budget, it’s legible to procurement, and seat count is the most natural expansion lever in a PLG motion (Fathom and tl;dv live on this). But the moment your power users consume 10–100x the light ones, a flat seat either erodes margin or prices out the casual buyer. The corpus answer is near-unanimous: keep the seat as the base and add a usage layer — credits (Glean, Codeium), requests (Qodo), or per-outcome charges (Gladly, Kustomer). That preserves the predictable base bill while recovering variable cost from heavy users. If you sell into high-ACV verticals, the Harvey and Spellbook playbook — seat-based but quote-only — protects pricing power at the cost of self-serve velocity; only choose it if your deals are large enough to justify a sales-led top of funnel.