Three seat counts, and why they are never the same
Every per-seat contract has three populations and they diverge quietly. Purchased is what the contract bills: a fixed number, agreed in advance, usually at the previous renewal. Assigned is what the admin console has handed to a named person. Active is who actually did something in the last month. Purchased is almost always the largest, and the gap between it and active is shelfware.
The gaps have different causes and different fixes. Purchased minus assigned is a procurement problem: seats bought for growth or a project that never arrived. Assigned minus active is an offboarding problem: leavers whose accounts were never reclaimed, people who tried the tool once, teams that switched to something else. The second gap is the easier win because every seat in it has a named owner you can ask about.
Utilisation puts a single number on it: active divided by purchased. That framing matters because the alternative — active divided by assigned — flatters you by hiding the seats nobody was even given. What you pay for is purchased, so that is the denominator.
The financial translation is the effective cost per active user: total contract value divided by the people who use it. At 67.2% utilisation, a $180 seat costs $267.86 per real user. That is the figure to quote when someone asks whether the tool is worth its price, because it is the price you are actually paying.
From utilisation to a seat count you can negotiate
Utilisation is the simple part: U = A / P. Shelfware follows as (P − A) × c, the idle seats multiplied by what a seat costs. Both are descriptive; neither tells you what to buy.
The number to negotiate with is the right-sized seat count, ⌈A × (1 + h)⌉. It takes the people actually using the tool and adds headroom for hiring and onboarding, then rounds up because a partial seat does not exist and rounding down leaves someone unlicensed. Headroom is a policy choice: 5% for a stable team, 10–15% where hiring is steady, more where a rollout is in progress.
Saving at renewal is (P − right-sized) × c, and it can be negative. If your utilisation is high, adding headroom pushes the right-sized count above what you hold and the calculator says so plainly — the correct action there is to budget for more seats, not to celebrate a saving. Any tool that reports only positive savings is hiding half its answers.
Waste over the remaining term, shelfware × months ÷ 12, is the money already committed. It is not recoverable through negotiation in most contracts, and it is the argument for reclaiming seats early rather than waiting: seats you free now can absorb new hires instead of triggering a mid-term purchase.
Worked example: 250 seats, 168 people using them
You hold 250 seats at $180 a year. The admin console shows 230 assigned and 168 distinct users active in the last 30 days. You want 10% headroom, and the contract has 7 months left.
- Utilisation. 168 ÷ 250 = 67.2%.
- Assignment rate. 230 ÷ 250 = 92.0%, so 20 seats were never handed to anyone at all.
- Idle seats. 250 − 168 = 82 seats paid for and unused, of which 62 are assigned to a named person and 20 are unassigned.
- Annual shelfware. 82 × $180 = $14,760 a year.
- Right-sized count. 168 × 1.10 = 184.8, rounded up to 185 seats.
- Saving at renewal. (250 − 185) × $180 = 65 × $180 = $11,700 a year.
- Waste over the remaining term. $14,760 × 7 ÷ 12 = $8,610 already committed before you can change anything.
- Effective cost per active user. 250 × $180 ÷ 168 = $45,000 ÷ 168 = $267.86 against a $180 list seat.
Two conclusions come straight out. The renewal saving is $11,700, not the $14,760 of shelfware, because 10% headroom keeps 17 spare seats deliberately. And the $8,610 of remaining-term waste is not recoverable by negotiation, which is the case for reclaiming the 62 assigned-but-inactive seats now and using them for the next two quarters of hiring rather than buying more.
What utilisation is good, and what to do at each level
Above 90% the contract is well sized and the risk has flipped: you may be one hiring wave from an overage charge, which is usually priced above your negotiated rate. Check the headroom output rather than the saving, and consider buying a small block at renewal while you still have leverage.
Between 75% and 90% is normal and healthy for a tool with routine turnover. Reclaim assigned-but-inactive seats, but there is no case for a difficult renewal conversation. Between 60% and 75% is where the money is: a fifth to a third of the contract is idle, the saving is material, and the vendor knows the usage data as well as you do.
Below 60% the question stops being how many seats to buy and becomes whether to keep the tool. An effective cost per active user well above the list price usually means either a failed rollout or a product that lost to an alternative already in the estate. That is a consolidation question rather than a licence-count question.
Whichever band you are in, measure over a sensible window. Thirty days is the standard, but a tool used monthly at close, quarterly for planning, or annually for compliance will look dead at 30 days and be entirely legitimate. Match the window to the workflow, and say which window you used whenever you quote a utilisation figure.
Shelfware and right-sizing at each utilisation level
| Utilisation | Active seats | Annual shelfware | Right-sized seats | Seats returned at renewal |
|---|---|---|---|---|
| 100% | 100 | $0 | 110 | −10 |
| 95% | 95 | $900 | 105 | −5 |
| 90% | 90 | $1,800 | 99 | 1 |
| 80% | 80 | $3,600 | 88 | 12 |
| 70% | 70 | $5,400 | 77 | 23 |
| 60% | 60 | $7,200 | 66 | 34 |
| 50% | 50 | $9,000 | 55 | 45 |
| 40% | 40 | $10,800 | 44 | 56 |
The negative values in the last column are real answers: at 95% utilisation with 10% headroom you need to buy five more seats, not return any. The break-even sits just under 91% utilisation, where the right-sized count first drops below the seats held.
Mistakes that make a utilisation number wrong
- Counting logins instead of distinct users. One person logging in forty times is one active seat. Take a distinct-user count from the admin report, not an event count.
- Using a 30-day window for a monthly-cadence tool. Financial close, board reporting and compliance tools are used a few days a year and are not shelfware. Match the window to the workflow.
- Measuring assigned rather than purchased as the denominator. It hides seats that were bought and never handed out, which is often the largest single block.
- Forgetting that some licences are contractually named. Where a seat is tied to an individual rather than a concurrent slot, reclaiming and reassigning may need a contract change, not just an admin action.
- Right-sizing to zero headroom. A contract with no spare seats generates emergency purchases at list price, which usually costs more than the seats you handed back.
- Assuming the saving is recoverable mid-term. Most agreements permit growth but not reduction until renewal, so the remaining-term waste is committed spend, not a target.
Where the numbers come from
Seats purchased comes from the order form, not the console — consoles sometimes display an entitlement that includes free or sandbox seats. Seats assigned comes from the admin user list. Active seats comes from the vendor's own usage report where one exists, and from your identity provider's sign-in logs where it does not; single sign-on logs are the most reliable cross-vendor source because they are yours rather than the vendor's. Where a tool exposes an API, pull all three monthly and keep the series — utilisation on the day of a renewal negotiation is far less persuasive than twelve months of it.
Turning the number into a renewal outcome
Take utilisation into the renewal early. Most vendors require 30 to 90 days' notice to change seat counts, and a right-sizing request made inside that window simply does not apply. Diary the notice date rather than the renewal date, and bring the twelve-month utilisation trend rather than a single reading.
Expect the counter-offer to be a discount rather than a seat reduction, because vendors defend seat count above price. A discount on 250 seats can beat list price on 185 seats or it can be worse — compare the two on total contract value, and check what the uplift terms do to each over the next three years before agreeing.
Where utilisation is low because the tool overlaps another one, the licence conversation is the wrong conversation. Price the alternative properly with the consolidation calculator, and include implementation and retraining in the total cost of ownership before switching, since a cheaper licence attached to a costly migration is not a saving. Where usage is genuinely growing, model both pricing shapes with the per-seat versus usage calculator — a tool you are outgrowing on seats may be cheaper on consumption.