Sales-Led Pricing: Examples & Companies

What is it

Sales-Led Pricing is a sales motion where a human at the vendor must engage to close the deal — typically tied to custom quotes and committed contracts.

It is the top-of-market motion. Deals that need security review, custom terms, and volume discounts don’t fit a self-checkout, so a human seller scopes the account and returns a quote. In this corpus the motion is nearly universal at the enterprise tier: 294 of 338 in-corpus AI companies run a sales-led tier — but almost always on top of a self-serve base, not instead of it. The published self-serve price becomes the anchor, and the sales-led conversation negotiates up (more seats, dedicated capacity, private deployment) and down (committed-use discounts).

A minority go fully sales-led with no public number anywhere. Harvey sells its legal AI “exclusively through an enterprise, sales-led motion with no public rate card,” and Glean quotes per-user Enterprise Flex seats plus pooled FlexCredits with no posted price. These are the exceptions — seat-heavy enterprise-knowledge plays where the buyer already expects a procurement process. For the other ~280, sales-led is simply the enterprise rung of a ladder that starts with a published price, whether that’s Cursor at $20/mo Pro before its team and enterprise quotes, or Intercom at $29–$132/seat plus $0.99 per Fin resolution before custom Fin agreements.

The interesting question a sales-led model raises is what triggers the hand-off to a human — a feature gate (SSO, audit logs), a dollar threshold (seat count or usage volume), or a compliance requirement (data residency, private deployment). That trigger is the real design decision behind the motion.

How it works

A sales-led deal moves through a predictable pipeline. The vendor withholds a firm price until the account is scoped, then prices as a function of size and term.

The unit math is usually seat- or consumption-based with a discount applied against a commitment:

Unit math: Annual contract = (seats × negotiated seat_price) + (committed usage × unit_price) − commit_discount

The negotiated rate almost always references a public anchor. Writer publishes a self-serve Starter at $39/user/mo ($29 annual, 5-seat cap), then quotes its Enterprise platform combining Pro seats, unlimited free Lite seats, and usage/credit metering — the $39 anchor frames the enterprise conversation. On the consumption side, outcome vendors negotiate a per-result rate: Sierra prices per resolution with volume pricing for routine interactions, and Decagon quotes per-conversation or per-resolution on annual contracts. The lever in both cases is the commitment — the buyer trades an annual (or multi-year) commit for a lower unit rate.

Note the distinction from gating: sales-led is about who closes the deal (a human), while gated is about whether the price is public. A vendor can be sales-led with a fully published anchor (most of the corpus) or sales-led and gated (Harvey, Glean). Most sales-led vendors publish a self-serve price below the enterprise tier; only the fully-gated few hide it entirely.

Companies using this

294 companies in the current corpus run a sales-led tier: fully-gated enterprise verticals (Harvey, Glean), model and inference vendors with enterprise or committed offerings (Anthropic, OpenAI, Cohere, Together AI, Baseten, CoreWeave), outcome-based agents (Sierra, Decagon, Intercom’s Fin), and apps with an enterprise plan atop self-serve (ElevenLabs, Synthesia, Perplexity AI, Writer, Cursor). The table below lists each.

Patterns observed

The dominant pattern across these 294 companies is that sales-led is a tier, not a whole company. For the large majority it sits at the top of a ladder that starts self-serve, and the public price anchors the negotiation. Cursor exposes free Hobby, $20/mo Pro, $60 Pro Plus, and $200 Ultra before you ever reach a team or enterprise quote; Intercom posts $29–$132/seat plus $0.99 per Fin resolution before custom agreements. The published number does double duty — it lands the small buyer and frames the enterprise redline.

Committed-use discounts are the core lever that separates the enterprise tier from the self-serve one. Enterprises trade an annual (or multi-year) commitment for a lower unit rate, and the inference clouds make this most explicit: Together AI, Baseten, and CoreWeave all pair on-demand rates with reserved-capacity commitments quoted through sales. For seat-based apps like Writer, the same logic applies to the $39/user Starter versus the sales-quoted Enterprise platform — commit for scale, pay less per seat.

The trigger that moves a buyer up the ladder is almost always security and compliance rather than raw volume. SSO, SCIM, audit logs, and data-handling terms are the up-tier gate — see enterprise SaaS pricing. A team that would happily stay on a self-serve plan gets pulled into a sales-led contract the moment its security team requires SSO enforcement or a signed DPA. Outcome-based agents are a distinct sub-pattern: Sierra and Decagon are sales-led not because of a feature gate but because a per-resolution price only makes sense once the vendor has scoped the customer’s conversation volume and quality bar.

Full gating, finally, is rare and vertical. Only a handful — led by Harvey and Glean — hide pricing entirely, and both are seat-heavy enterprise-knowledge plays where the buyer expects procurement. For the other ~280, hiding the number would cost more funnel than it would protect margin.

Counterexamples & variants

The clean counterexample is the pure self-serve vendor with no sales motion at all. Many developer APIs let even large customers scale on prepaid credits without ever signing a contract — they trade negotiated commitments for zero sales cost and bet that volume makes up for the lost enterprise premium. On a page about sales-led pricing, these are the companies that deliberately refuse to add a human to the loop, and they populate the sibling self-serve pricing theme.

The notable variant within sales-led is fully gated sales-led — no public price anywhere, exemplified by Harvey (legal AI sold per seat via sales-led quotes, no public rate card) and Glean (per-user Enterprise Flex seats plus pooled FlexCredits, quote-based). Full gating maximises pricing power and discretion but raises buyer friction, so it is viable only where the buyer already expects a procurement process. It misfires when applied to a product that a $20 buyer could have self-served: gating the whole funnel behind a sales team throws away the bottom of the market to protect a margin the vendor could have captured with a published anchor plus an enterprise tier.

A subtler failure mode is the outcome-priced sales-led deal that can’t be benchmarked. Because Sierra and Decagon price per resolution or per conversation with no public rate, a buyer has nothing to anchor against and must negotiate blind — the same discretion that helps the vendor hurts the buyer’s ability to evaluate the quote. The fix, covered below, is to benchmark against a vendor that does publish a comparable rate.

What this means for buyers vs vendors

For buyers

Use any public self-serve price as your anchor and negotiate from it — for most vendors on this page there is a published number (Cursor’s $20/mo Pro, Writer’s $39/user Starter, Intercom’s $0.99 per resolution) that frames the enterprise quote. Committed-use discounts are standard, so a multi-year commitment should buy a lower unit rate; get the per-unit price and the true-up terms in writing. For fully-gated vendors like Harvey or Glean, benchmark against a public-priced alternative to keep the quote honest, and model the seat- or usage-based total with the pricing calculator before you sit down. Our guide to choosing the right usage metric helps you pressure-test whether the vendor’s meter (seats, resolutions, tokens) actually maps to the value you’ll get.

For vendors

Reserve sales-led for deals that genuinely need it — security review, custom terms, large commitments — and keep a self-serve path below it so you’re not gating $20 buyers behind a sales team. Full gating maximises discretion but costs you the bottom of the funnel; most AI vendors are better served by self-serve plus a sales-led enterprise tier, which is exactly what ~280 of the 294 companies here do. Anchor the enterprise conversation on a published price so buyers negotiate against a number rather than a void, and make the up-tier trigger (SSO, data residency, committed volume) explicit. For the contract mechanics — commits, true-ups, and billing cadence — see our guides on usage-based pricing models and usage invoicing & billing cycles.

Explore this theme in the knowledge graph →