Notion at 10,000 Seats: The Glean-Displacement Math No CFO Runs

Notion lists at $10/seat. At 10,000 seats that's $1.2M/yr — large enough that procurement notices, small enough that nobody runs the displacement math. Most CFOs treat Notion the way they treat Slack: a per-seat tax everyone pays, renegotiate the headline price every two years, move on. That instinct is wrong when an enterprise search agent is on the table, because Notion is one of the few documentation tools where the AI-compression argument runs in both directions — and the break-even isn't where finance assumes it is.

The $1.2M number is real, and the catalog says Glean takes 30% of it

Start with the gross. Notion's enterprise catalog rate is $10/seat/mo. Ten thousand seats × $10 × 12 months = $1.2M annual contract value. That's the number on the invoice, before any renegotiation.

Now the displacement claim. The SeatCompress catalog carries Glean → Notion at 30% compression. That number isn't a vendor brochure pull — it's a sourced, capped derivation. Glean is classed as a vertical_replacement agent for knowledge-search workloads, which has a 0.65 hard ceiling. A 30% claim sits well below that ceiling, which means the catalog is taking the conservative interpretation: Glean displaces 30% of the active user base that would otherwise have logged into Notion to search for something, not 30% of every seat the org provisions. The methodology is documented in why we source every compression percentage.

Thirty percent on a $1.2M line is $360K of theoretical annual gross savings. Theoretical. The realistic year-one number is lower for reasons we'll get to, and that gap is where most CFOs sign the wrong contract.

What Glean actually costs at enterprise scale

Glean's catalog pricing is $6,000 per 100-seat block above the enterprise threshold, which works out to $60/user/mo at small bands and $45/user/mo above the 100-seat tier. Glean also carries an explicit $50,000 setup cost in the catalog — a real implementation fee, not a default placeholder.

For a 10,000-employee enterprise running Notion across the whole company, the relevant question isn't "how many Notion seats do we have?" — it's "how many seats should Glean actually be deployed to?" If you license Glean for every Notion seat, you're paying $45/user × 10,000 × 12 = $5.4M/yr plus $50K setup. That's not an AI strategy. That's lighting money on fire to replace a $1.2M tool.

The honest deployment question is: how many active Notion users would Glean genuinely serve? Active-seat ratios on documentation tools at enterprise scale typically run 70-85%. Call it 8,000 active users out of 10,000 provisioned. Glean at 8,000 seats × $45 × 12 = $4.32M/yr. Still wildly negative if Notion is the only target.

The math only starts working when Glean is also compressing the rest of the knowledge stack — Confluence at 40%, Coda at 25%, Dropbox Business at 15%, Box at 15%. Glean isn't a Notion-killer; it's a knowledge-graph rollup that earns its keep across four or five documentation surfaces simultaneously. The full whole-stack Glean economics post runs the multi-tool math at the company-wide level; the rest of this post deliberately isolates the Notion line so the renegotiation framing is clean.

The break-even calculation no CFO runs

Here's the math that should anchor the conversation. We're isolating the Notion line: what's the smallest active-seat count where Glean's slice of cost (allocated pro-rata to its targeted tools) is covered by the Notion savings alone?

At a per-user agent like Glean, deployment size scales with the maximum active seat count across its targeted tools. The MAX-overlap rule in the engine means we don't double-count: if Glean compresses both Notion and Confluence in the same stack, the deployment size is max(activeNotion, activeConfluence), not the sum.

For a pure-Notion test case — strip Confluence and Coda out, pretend you only own Notion — the break-even threshold sits where:

activeSeats × $10 × 0.30 × 12  =  activeSeats × $45 × 12  +  $50,000
       Notion savings              Glean cost     setup

Solve for activeSeats: the per-seat math is hopeless. Glean costs $540/user/yr at the enterprise rate; Notion compression yields $36/user/yr (30% of $120). That's a negative $504/user gap on every seat. There is no Notion-only break-even — Glean costs 15× what it saves on the Notion line in isolation.

This is the calculation that exposes single-tool ROI as the wrong frame. Glean only pays for itself when it sits on top of a multi-tool documentation stack and the savings pool aggregates. That's also where the catalog's pro-rata cost allocator — the per-tool best-agent picker — earns its keep: it splits Glean's flat cost across Confluence + Notion + Coda + Dropbox + Box by share of gross savings, so the Notion line shows its honest contribution.

Notion AI defends the base — partially

Here's where it gets interesting. The catalog also carries Notion's own first-party agent, Notion AI Business, at $10/user/mo (above the 100-seat enterprise tier: $8/user). Notion AI compresses Confluence by 40% and Coda by 30% — but not Notion itself, because there's no AI-vs-AI compression. An AI assistant inside a documentation tool doesn't displace the tool's own seats; that's a category error the catalog rejects on principle.

What Notion AI does do is change the renegotiation conversation. The bundled cost of Notion plus its AI layer lands around $20/seat at enterprise scale — but that bundle gives you 40% compression on Confluence if you happen to be running Confluence in parallel. For an enterprise running both — and many do, because engineering tends to live in Confluence while everyone else lives in Notion — the Notion AI line item starts to look like a Confluence-killer rather than a Notion enhancement.

This is the inversion that breaks the standard playbook. The CFO who walks into the Notion renegotiation thinking "drop my seat price 15%" is fighting the wrong battle. The real lever is: should I be on Notion at all, or should I consolidate to Confluence + Notion AI? The Notion AI compression on Confluence (40%) is higher than Glean's compression on Notion (30%). The vendor with the lower list price (Confluence at $5.16/seat) plus the AI layer becomes the consolidation target.

What Perplexity Comet Enterprise contributes

Perplexity Comet Enterprise sits in the catalog at $40/user/mo with a 5% compression mapping on Notion. Five percent. That's the floor the catalog is willing to assert for a horizontal AI search tool that happens to surface Notion content alongside everything else.

This number is deliberately tiny. Comet is classed as horizontal_ai — the 0.15 category cap applies — and the 5% claim represents the small slice of Notion seat-time that genuinely gets displaced when knowledge workers shift to a general-purpose answer engine. Anyone selling you a higher number is selling fiction. At 5% on a $1.2M base, Comet's Notion contribution is $60K/yr gross, against $40 × 8,000 active × 12 = $3.84M/yr in Comet cost. Comet's economics live or die on its breadth across the entire SaaS stack, not on any single line — which is the same lesson Glean teaches, but with worse single-tool math.

Worked example: a 12,000-employee SaaS company

Take a 12,000-employee enterprise software company. Knowledge management stack:

• Notion at 10,800 seats × $10 = $108,000/mo = $1.296M/yr (90% provisioning)
• Confluence at 4,200 seats × $5.16 = $21,672/mo = $260K/yr (engineering only)
• Coda at 800 seats × $10 = $8,000/mo = $96K/yr (data + product ops)
• Dropbox Business at 12,000 seats × $20 = $240,000/mo = $2.88M/yr (org-wide files)

Total documentation + storage layer: ~$4.53M/yr. Active-seat ratios: Notion 75% (8,100 active), Confluence 85% (3,570 active), Coda 60% (480 active), Dropbox 60% (7,200 active).

Scenario A: Glean across the whole stack. Glean deployment size = max active across targeted tools = 8,100 (Notion-driven). Cost: 8,100 × $45 × 12 = $4.374M/yr + $50K setup = $4.424M/yr. Gross compression savings, applying MAX-overlap and the active-seat-only rule:

• Notion: 8,100 × $10 × 0.30 × 12 = $291,600
• Confluence: 3,570 × $5.16 × 0.40 × 12 = $88,422
• Coda: 480 × $10 × 0.25 × 12 = $14,400
• Dropbox: 7,200 × $20 × 0.15 × 12 = $259,200

Annual gross savings = $653,622. Glean cost = $4.424M. Net: −$3.77M/yr. Catastrophically unprofitable. Glean's all-stack deployment doesn't pencil because the Dropbox seat count drags the cost base into the millions while the per-seat compression on file storage barely moves the needle.

Scenario B: Glean scoped to documentation only (skip Dropbox). Deployment size = max(8,100 Notion, 3,570 Confluence, 480 Coda) = 8,100 — still Notion-pinned. Cost is identical: $4.424M/yr. Same problem. The Notion seat count is the cost anchor regardless of which other tools you include.

Scenario C: Glean scoped to Confluence + Coda only, exclude Notion from the target list. Deployment size = max(3,570, 480) = 3,570. Cost: 3,570 × $45 × 12 = $1.928M/yr + $50K = $1.978M/yr. Savings: $88,422 + $14,400 = $102,822. Net: −$1.875M/yr. Still unprofitable, but the loss is smaller and Notion is preserved as a standalone line.

Scenario D: don't deploy Glean. Renegotiate Notion instead. Apply the year-one renegotiation realization factor (0.5) to a target seat reduction of 25% (i.e., consolidate the 25% of inactive Notion seats out of contract at renewal): theoretical savings = 10,800 × 0.25 × $10 × 12 = $324K. Year-one realistic = $162K. Net positive, zero setup cost, zero deployment risk.

The board-meeting version: at this stack composition, Glean costs more than it saves at every deployment scope. The right move is Scenario D — renegotiate Notion down to active-seat count at renewal, and revisit Glean only when the company's knowledge stack consolidates into a smaller, more Glean-friendly footprint. For the renegotiation script, see our Salesforce renegotiation playbook — the structural moves are the same.

When does Glean actually pay for itself?

The honest answer: when the targeted-tool savings base exceeds Glean's per-seat enterprise cost on the dominant tool's active-seat count, plus the $50K setup amortized over year one.

Practically, that means Glean breaks even when:

• The customer is running 4-6 high-compression knowledge tools simultaneously (Confluence, Notion, Coda, Box, Dropbox, plus internal wikis)
• The active-seat counts on those tools are roughly aligned (no single tool with 10,000 seats pinning the deployment size while the others sit at 500)
• At least two tools in the stack carry compression mappings above 30% — Confluence (40%) is the usual anchor

For the synthetic 12,000-employee company above, Glean wouldn't pay until Notion was either removed entirely (collapsing into Confluence) or shrunk to roughly the active-Confluence headcount via the renegotiation lever. That's a two-year plan, not a deploy-this-quarter decision.

What the CFO does Monday morning

Three moves, in order:

Run the Notion seat audit before the renewal. Pull provisioned vs. active seat counts from the IdP — the spread is the renegotiation lever. At 10,800 seats with 75% activity, you have 2,700 inactive licenses worth $324K/yr in gross contract value. Even 50% realization on a clean renegotiation puts $162K back. This is the calculation why unused seats is the wrong metric in 2026 — but at this scale, the unused number is still the simplest argument.

Refuse the Glean pitch on Notion alone. When the Glean rep walks you through their slide deck, the Notion line will get top billing. Don't engage with the single-tool ROI claim. Ask for the multi-tool deployment math with your specific active-seat counts loaded, your dominant tool clearly identified, and the $50K setup amortized over year one. If their model doesn't show a deployment scope where year-one net is positive, that's the answer — and it's the answer they don't want to give you in the room.

Consider Notion AI Business as a Confluence-displacement play. This is the lever most CFOs miss. If you're running both Notion and Confluence, the $10/seat Notion AI add-on compresses Confluence by 40% — which on a $260K Confluence line is $104K/yr theoretical, $42K year-one realistic. That's a higher-confidence move than the Glean rollout because it's a single-vendor bundle decision with no integration cost and no setup fee.

The bottom line: Notion at scale is a renegotiation target, not a displacement target, until your knowledge stack consolidates. The Glean math doesn't pencil for the same reason most flat-fee enterprise AI deployments don't pencil at 5K-50K employee scale — the deployment-size denominator is bigger than the compression-savings numerator on any single tool. To run the math on your own stack, the SeatCompress calculator takes the per-tool active-seat counts and computes the break-even point for each candidate agent against your specific renewal dates.

The CFOs who get this right next quarter are the ones who treat Notion the way they should have treated Slack five years ago: a per-seat tax with a renegotiation cycle, not a transformation opportunity. The transformation, when it comes, will be vendor consolidation — not an AI overlay on a stack the company hasn't earned the right to keep.