Enterprise SaaS Pricing: Examples & Companies

What is it

Enterprise SaaS Pricing is pricing plans designed for large organisations — typically custom-quoted, with SSO, SCIM, audit logs, and committed-use discounts.

It is the segment where SaaS margin actually lives. The list price on the marketing site is a starting anchor at best and irrelevant at worst; the real deal is negotiated per account, shaped by seat count, usage volume, contract length, and how much the buyer’s budget and value can bear. In the UsagePricing Blueprint, 280 in-corpus companies sell an enterprise plan — making it the single most common customer segment across the entire corpus, ahead of mid-market and SMB.

The interesting question is never “what’s the price” — it’s what forces a customer up to the enterprise tier. Almost universally the line is drawn at compliance and identity: Linear gates SAML and SCIM behind its custom Enterprise plan while listing Basic at $10 and Business at $16 per user/month; Notion AI reserves SCIM provisioning, audit logs, and zero data retention with LLM providers for Enterprise, one tier above its $20/member Business plan; and Superhuman puts SSO/SCIM and a dedicated CSM one step above its listed $33/user/month Business tier.

At the top of the market the plan disappears from public view entirely. Harvey — one of the fastest-scaling vertical-AI companies on record, past $100M ARR — publishes no rate card, no checkout, and no free tier; the only path to a price is a demo booking, and third-party estimates put it at low-to-mid four figures per seat per month with ~20-seat minimums and annual commitments. Glean, Sierra, and Decagon are equally gated. That opacity is a feature, not an oversight: it lets vendors price-discriminate across accounts without anchoring the whole market to one number.

How it works

An enterprise plan is built from four levers that a sales rep composes into a single quoted figure. The published price (when there is one) reflects only the first; the other three are where the deal is actually made.

The base unit is where enterprise pricing has diversified most. A pure per-seat floor still dominates — Gong charges per-user licenses plus a platform fee scaled to supported users, and Clari sells per-seat subscriptions tailored to the use case — but usage and outcome layers are now routine. Intercom charges $29–$132 per seat across its three platform tiers plus $0.99 every time its Fin AI agent closes a conversation. GitHub Copilot pairs a $10–$39 per-seat subscription with a GitHub AI Credits pool (1 credit = $0.01). And some enterprise deals drop the seat entirely: Sierra and Decagon price per resolution or per conversation, aligning vendor revenue to the value delivered.

Deal math: For a gated seat-plus-usage vendor, the quoted annual figure ≈ (committed seats × negotiated seat rate) + (expected usage × metered rate) − volume discount. For Glean, third-party reconstruction of a 200-seat account lands near ~$50/user base + ~$15/user AI add-on + ~$2K–5K/mo FlexCredit overage — none of which appears on the marketing site.

Where a vendor does publish a floor, it usually functions as an anchor rather than a real price: Pinecone lists a $500/mo Enterprise minimum plus usage (with a 99.95% uptime SLA and CMEK), and Snowflake Cortex sets a per-credit rate by edition — $2.00 Standard, $3.00 Enterprise, $4.00 Business Critical — that scales the entire consumption bill. For a deeper treatment of how these floors and commits interact with metered usage, see the guide to entitlements and usage grants and the usage invoicing and billing cycles guide.

Companies using this

Enterprise is the corpus’s largest segment: 280 in-corpus companies ship a plan aimed at large organisations, spanning vertical-AI leaders like Harvey and Glean, horizontal platforms like Writer and Notion AI, and outcome-priced agents like Sierra and Decagon. The table below is sortable and filterable — note the Free tier and Last verified columns to separate the fully-gated vendors from those with a self-serve on-ramp beneath the enterprise line.

Patterns observed

Across the 280 enterprise-segment companies, a handful of patterns recur strongly enough to treat as the shape of enterprise SaaS pricing in 2026:

• The upgrade trigger is compliance, not capacity. The feature that forces a buyer to Enterprise is almost never “more of the product” — it’s SSO/SAML, SCIM, audit logs, RBAC, and certifications, the gates drawn by Linear, Superhuman, and Notion AI above. Buyers pay the enterprise premium for security posture and procurement-readiness, not for extra features.

• The higher the deal, the less the price is public. There is a clean gradient from listed to gated: prosumer-adjacent tools (Superhuman, Linear) publish full ladders; platform vendors publish a floor and hide the rest (Sourcegraph Cody, Pinecone); and top-of-market vertical AI (Harvey, Glean, Sierra, Decagon) publishes nothing at all.

• Usage and outcome layers are colonising the seat. Pure per-seat enterprise pricing is giving way to hybrids: Writer layers usage/agent metering onto Pro seats, and Glean bolts pooled FlexCredits onto Enterprise Flex seats. The seat is becoming the access fee, with value captured through metered AI on top.

• Commit + annual term is the default, not the exception. Enterprise deals are structured around minimum spend, seat floors, and multi-year discounts. Sierra’s annual commitment with dedicated SLAs and Snowflake Cortex’s edition-based credit pricing encode lock-in into the commercial model, not just the price.

Counterexamples & variants

The tidy “custom-quoted, contact-sales” archetype breaks in several instructive ways.

The published-floor enterprise plan. Not every enterprise vendor hides its number. Sourcegraph Cody retired its standalone tiers in favour of a single Enterprise plan marketed openly at “$16K starting,” with per-seat and final pricing quoted via Contact sales. Pinecone publishes a $500/mo Enterprise minimum alongside usage rates (~$24–27/M read units). Here the list price is a real anchor that shapes the negotiation rather than replacing it — a deliberate hybrid of transparency and gating.

The seatless enterprise deal. The per-seat assumption fails entirely for outcome-priced agents. Sierra — the flagship of outcome-based pricing — charges per resolution (“a job well done”) with no charge when the agent doesn’t resolve, quoted only by sales. Decagon prices per conversation or per resolution. Intercom’s Fin agent charges $0.99 per resolution with no seat minimum. For these vendors the “enterprise plan” is a commercial envelope of compliance and SLAs wrapped around a usage meter, not a seat count.

The consumption-primitive that has no “enterprise plan” at all. Infrastructure vendors like Snowflake Cortex and Vapi don’t really sell an enterprise tier — they sell metered consumption and let the enterprise features (data residency, SOC 2/PCI, dedicated support, committed volume) attach as a Scale contract. Vapi’s Build plan is pure usage at $0.05/min plus at-cost provider pass-through; the enterprise deal is a quoted annual commit layered on the same meter. The “enterprise segment” here is a contract shape, not a price on a card.

The prosumer-priced enterprise line. At the bottom of the market the enterprise premium is small and visible. Superhuman’s enterprise-grade controls sit just one tier above a $33/user Business plan, and Linear’s Enterprise is a light custom quote over a $16/user Business plan. When the buyer is close to the individual user, enterprise pricing stays honest — there’s a real published ladder underneath it.

What this means for buyers vs vendors

For buyers

Treat the list price (if any) as the ceiling of your anchor, not the floor of your cost. The real levers are commit size, seat floor, and term — and every one is negotiable, especially the multi-year discount. Ask three questions early: what forces us onto the enterprise tier (usually SSO/SCIM/audit logs — confirm you actually need them), what’s metered on top of seats (a per-resolution fee or credit-pool overage can dwarf the seat cost), and what’s the seat minimum (a 20-seat floor makes small pilots expensive). If a vendor publishes a floor, use it to benchmark; if they publish nothing, get a written usage estimate before signing, because the gated model exists precisely so the vendor can charge what your budget bears.

For vendors

Gate the enterprise plan at compliance and identity, not at core capability — buyers accept paying for SSO, SCIM, audit logs, and SOC 2 far more readily than for feature paywalls, and it keeps your self-serve funnel intact beneath the line. Publish a floor only if you want an anchor: a visible “$16K starting” or “$500/mo minimum” filters unqualified leads and speeds the sales cycle, while a fully gated model (Harvey, Sierra) maximises price-discrimination at the cost of organic discovery. Increasingly the winning structure is a seat access fee plus a metered value layer — model the blend with the pricing calculator and read the introduction to usage-based pricing before you decide whether the enterprise upsell should be more seats, more usage, or an outcome.

Explore this theme in the knowledge graph →