Stop paying for AI seatsnobody uses.
Tool-level audits tell you which apps to keep. They miss the dead seats inside the apps you keep. This scorecard rates every individual license on real utilization, gives each a KEEP / DOWNGRADE / RECLAIM verdict, and reports your fleet waste rate — so the unused capacity stops hiding behind the average.
Seat sprawl is where AI budgets quietly blow out.
Per-seat AI tool costs have climbed multiples in a year. The bill grows even when usage doesn't.
Fleet-wide 'utilization' hides the seats at zero. Tool-level audits keep the tool and never look inside it.
Seats get provisioned for a project and never pulled. The friction to add is low; the friction to reclaim is a conversation nobody owns.
Re-rate a seat and watch the verdict move.
Tap any 0–3 cell to re-rate a signal and watch the verdict move. Worked example as of 2026-06-25.
| Seat | Active-days | Feature | Last-used | Output | Tier-to-usage | Util | Verdict |
|---|---|---|---|---|---|---|---|
A. Okafor S-001 · Premium | 100 | KEEP nothing to fix | |||||
B. Lindqvist S-002 · Premium | 77 | DOWNGRADE over-provisioned | |||||
C. Ramirez S-003 · Standard | 67 | DOWNGRADE fix: Tier-to-usage fit | |||||
D. Whitfield S-004 · Premium | 33 | RECLAIM fix: Tier-to-usage fit | |||||
E. Nakamura S-005 · Standard | 18 | RECLAIM fix: Active-days share | |||||
F. Adeyemi S-006 · Premium | 0 | RECLAIM dead-seat gate | |||||
G. Petrov S-007 · Standard | 80 | KEEP fix: Tier-to-usage fit | |||||
H. Donnelly S-008 · Premium | 10 | RECLAIM fix: Feature depth |
Same math as the workbook: weighted utilization (signals sum to 100), the dead-seat gate forces RECLAIM when a seat is both inactive and cold, and a 70+ seat still lands DOWNGRADE when it's on a tier it doesn't use. A decision aid for software-spend right-sizing — not financial advice.
How the gate works, in one image
The five weighted utilization signals and the over-provisioned gate that caps a heavily-used seat at DOWNGRADE when it's on a tier it doesn't need — the same math the demo runs, as a diagram you can share or embed anywhere.
View & embed the full diagramEight seats in. A fleet posture out.
This is the shipped example, scored by the rule the workbook and the demo both run. Note the two seats that make the design earn its keep: S-002 scores 77 — well above the KEEP floor — yet lands DOWNGRADE because it's on a premium tier it doesn't use. S-006 scores 0 and the dead-seat gate fires.
AI Seat Right-Sizing & Waste-Rate Scorecard (as-of 2026-06-25) ================================================================ Fleet: 8 seats | waste rate 50% -> OVER-PROVISIONED KEEP 2 DOWNGRADE 2 RECLAIM 4 ---------------------------------------------------------------- S-001 A. Okafor util 100 -> KEEP S-002 B. Lindqvist util 77 -> DOWNGRADE [over-provisioned] S-003 C. Ramirez util 67 -> DOWNGRADE S-004 D. Whitfield util 33 -> RECLAIM S-005 E. Nakamura util 18 -> RECLAIM S-006 F. Adeyemi util 0 -> RECLAIM [GATE: dead seat] S-007 G. Petrov util 80 -> KEEP S-008 H. Donnelly util 10 -> RECLAIM
Three rules keep the verdict honest.
Every license is rated on five weighted signals that sum to 100. Utilization is a number you can defend line by line.
Both inactive and cold? RECLAIM, regardless of the other three signals. A strong-on-paper seat can't hide an unused one.
A 70+ seat on a tier it doesn't use is DOWNGRADE, not KEEP. That's a plan move — the cheapest dollars to recover.
A seat-level scalpel, not a tool-level axe.
- A per-seat verdict engine for one tool's license roster.
- A fleet waste-rate readout that trends over time.
- A reclaim runbook with the actual downgrade and reclaim scripts.
- A tool-level keep/cut audit — that's the Subscription Auditor's job, one level up.
- A live usage integration — you rate the signals from your admin export.
- Financial advice. It informs a spend decision; it doesn't make it for you.
A decision aid for right-sizing software spend — not financial or accounting advice. Confirm each seat with its owner and finance before changing or cancelling a license.
The person who signs off on the renewal.
Build the full AI cost & governance stack.
Model and route token spend before it lands on a seat.
ViewAudit what every agent and connector can reach across your stack.
ViewStand up the governance scaffolding spend decisions report into.
ViewEverything else you'd ask before buying.
Different altitude. A subscription auditor works at the tool level — keep, right-size, consolidate, or cut each app in your stack. The AI Seat Right-Sizing & Waste-Rate Scorecard works one level down, inside the tools you keep: it rates every individual license on real utilization and returns KEEP, DOWNGRADE, or RECLAIM per seat, then a fleet waste rate. The audit tells you which apps to keep; this tells you which seats inside them you're paying for and nobody uses.
It's the structural override that keeps the verdict honest. A seat that is both inactive (zero active days) and cold (never recently used) is RECLAIM no matter how it scores on the other three signals — feature depth, output volume, or tier fit can't rescue a license nobody opens. Both gate-eligible signals at the floor means you're paying for nothing, so the score is overruled. In the worked example, seat S-006 scores 0 and the gate fires.
Because active isn't the same as right-sized. A seat can score above the KEEP floor on usage and still be on a premium tier it doesn't actually use — high activity, low tier-to-usage fit. That's a DOWNGRADE, not a KEEP: keep the person working, move them to the cheaper plan. In the worked example, seat S-002 scores 77 — well above the floor — yet lands DOWNGRADE because its tier fit is zero. Those are the cheapest dollars you'll recover all quarter.
It's the share of seats on a roster that come back RECLAIM, rolled up to one headline — LEAN, TRIM, or OVER-PROVISIONED. It's the number a fleet-wide 'average utilization' hides, because the average quietly absorbs the seats sitting at zero. The worked example's 8-seat roster runs a 50% waste rate, which reads OVER-PROVISIONED. Re-run it each renewal cycle to trend the rate down.
No. You rate each seat's five signals from your admin or usage export — it never connects to a provider, reads live usage, or changes a license, and nothing is uploaded. The verdict comes entirely from the numbers you enter, scored the same way every time. The included Reclaim Runbook gives you the actual downgrade and cancellation scripts to act on the results.
One .xlsx Seat Scorecard with the per-seat verdict engine and the fleet waste-rate dashboard, plus a Facilitator Playbook and a Reclaim Runbook (both .docx) and a worked example. It opens in Excel, Google Sheets, or Numbers — deterministic and offline. It's a decision aid for right-sizing software spend, not financial or accounting advice; confirm each seat with its owner and finance before changing or cancelling a license.
Find the dead seats.
Recover the spend.
One purchase, lifetime access, 12 months of updates. $79, once.
A decision aid for right-sizing software spend — not financial or accounting advice. Confirm each seat with its owner and finance before changing or cancelling a license.
Sold by RedHub AI LLC · Secured by Stripe · redhub.ai