Model mismatch cost more than list pricing, because named user seats ran 40 to 65 percent utilized while consumption programs overran their first year by 15 to 35 percent
Named user, consumption, or mixed. The licensing model fits your usage curve or it quietly doubles the cost per active user.
Prepared by Redress Compliance · August 19, 2026 · Engineering and design software estates. 15 to 25 engagements benchmarked, 2024 to 2025.
Executive summary
Named user seats in CAD estates ran 40 to 65 percent utilized, leaving a third or more of subscription spend idle and recoverable without any vendor conversation.
Consumption programs overran budget by 15 to 35 percent in their first year wherever usage governance was absent at the start rather than added after the overage.
Estates running two or more of these vendors carried 20 to 30 percent overlapping capability they paid for twice, usually left behind by acquisitions.
Cost per active user is the comparable number, not list price. A 60 percent utilized estate costs two thirds more per active user than the rate card suggests.
How do the three licensing models differ?
One sells named users with tokens for occasional use, one bills consumption against an enterprise commitment, and one mixes perpetual, subscription and token schemes by product line.
| Factor | Named user model | Consumption model | Mixed model |
|---|---|---|---|
| Primary basis | Named user subscription | Consumption against commitment | Perpetual, subscription and tokens |
| Occasional users | Token pools | Counted per use | Token pools by product |
| Cost predictability | High, per seat | Low without governance | Medium, varies by line |
| Where it stings | Idle seats | Peak usage true ups | Maintenance on legacy perpetual |
Which model fits which usage curve
Named user wins for daily heavy users, and consumption wins for large populations of occasional users. Perpetual plus maintenance only wins where versions can be frozen for years.
Map the usage histogram before debating price. The curve settles the model, and the model settles more money than the rate does.
The same team prices three different ways
A single 500 person engineering organization produces three materially different numbers under the three models. That spread is the decision, and the discount conversation sits well downstream of it. The vendor catalogs are published on the Autodesk product pages, the Bentley software pages and the Hexagon product pages.
What does each model really cost per active user?
Cost per active user is the comparable number, and idle seats and unmanaged consumption distort it in opposite directions.
Three different divisions, one comparable figure
- Named user: divide subscription spend by users active in the last 90 days, then reclaim or convert the rest.
- Consumption: divide program spend by distinct active users and check the peak hours driving overage.
- Mixed: add maintenance on shelfware perpetual licenses before comparing anything.
Converting an idle named seat to a token pool is the usual first move, and the break even math sits in the token pricing guide.
What 60 percent utilization actually means
An estate running at 60 percent utilization costs two thirds more per active user than the rate card suggests. No discount on that rate card closes the gap, because the gap is in the denominator.
The engineering software audit defense guide
The compliance triggers, the usage evidence that holds, and the negotiation sequence for engineering software estates.
Get the brief →What 15 to 25 engineering software estates showed
Across roughly 15 to 25 engineering and design software engagements Fredrik Filipsson benchmarked between 2024 and 2025, model mismatch cost more than list pricing. Three patterns recur.
- Named user seats in CAD estates ran 40 to 65 percent utilized, leaving a third or more of subscription spend idle.
- Consumption programs overran budget by 15 to 35 percent in their first year when usage governance was absent.
- Estates running two or more of these vendors carried 20 to 30 percent overlapping capability they paid for twice.
The right licensing model for your usage curve beats a discount on the wrong one, and the gap between the two is larger than any rate negotiation reaches.
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- Entitled vs deployed vs active seats compared, priced at your actual contract terms
- A ranked savings queue with dollar values, not license counts
Why do multi vendor estates overspend?
Because the three portfolios overlap. Plant design, reality capture and structural analysis exist in more than one stack, and acquisitions leave teams on parallel tools.
The overlap is budget, not architecture
The 20 to 30 percent overlap benchmarked across these estates is spend that consolidation recovers directly. It is rarely a technical decision and almost always an inherited one.
Consolidate what you do not need, keep what gives you leverage
Consolidate the overlap you genuinely do not need and keep one alternative alive where switching is credible. Those are two different questions and answering them as one is how estates end up either paying twice or paying lock in.
How aggressive are these vendors on true ups?
Differently, and the difference decides what evidence you need. One runs an active compliance program; the others enforce through usage reconciliation that surfaces as an invoice.
Three postures, three defenses
- Named user vendor: watch installer telemetry and legacy serial use, because old network licenses are the usual claim base. The program is described on the compliance page.
- Consumption vendor: the quarterly usage report is the audit, so dispute peak counting rules and shared machine effects in writing.
- Mixed vendor: entitlement sprawl across acquired product lines makes baseline reconstruction the defense.
A true up dispute is an evidence fight over counting rules
Concurrent peaks, shared workstations and idle session timeouts move the count materially. The contract language on each is negotiable before signature and not after the invoice, which is covered in the audit defense guide.
Where the common advice on engineering software is wrong
The standard advice is to standardize the whole engineering estate on one vendor to maximize discount leverage. We disagree.
The consolidated vendor prices lock in, not loyalty
Across the engagements benchmarked in 2024 and 2025, single vendor estates paid higher unit rates at renewal than estates that kept a credible second stack in a defined niche.
The buyer side move is to consolidate the overlap you genuinely do not need and keep one alternative alive where switching is real. The agreement structure that supports that sits in the enterprise agreement negotiation guide.
What the estates measured, 2024 to 2025
Two cuts of the benchmark file, and neither is a rate.
Measured against users active in the last 90 days, leaving a third or more of subscription spend idle across the estates.
Where two or more of the three portfolios ran in parallel, usually left behind by an acquisition rather than chosen.
Both are recoverable inside the estate. Neither requires the vendor to agree to anything first.
Your first five moves
- Build a twelve month usage histogram per product before any renewal conversation, because the model decision needs the curve and the discount conversation does not settle it.
- Reclaim or convert idle named user seats and right size the token pools to the histogram, since utilization ran 40 to 65 percent across the benchmarks.
- Cap consumption true ups with a banded commitment and rollover of unused volume, because first year programs overran 15 to 35 percent without governance.
- Price the overlapping capability across vendors and put the consolidation case in the room, which is where the 20 to 30 percent duplicate spend lives.
- Lock multi year unit rates with a reduction right rather than a price hold. The spend health check builds the histogram before the quote arrives.
Frequently asked questions
What are the three models?
Named user subscription with tokens for occasional use, consumption billed against an enterprise commitment, and a mix of perpetual, subscription and token schemes by product line.
Which model fits which usage?
Named user for daily heavy users, consumption for large populations of occasional users, and perpetual with maintenance only where versions can be frozen for years.
How utilized are CAD seats?
Between 40 and 65 percent across the benchmarked estates, measured against users active in the last 90 days. A third or more of subscription spend sat idle.
What does low utilization actually cost?
An estate at 60 percent utilization costs two thirds more per active user than the rate card suggests, and no discount on that rate card closes the gap.
How far do consumption programs overrun?
By 15 to 35 percent in their first year wherever usage governance was absent at the start rather than bolted on after the first overage invoice.
Why do multi vendor estates overspend?
Because the portfolios overlap. Plant design, reality capture and structural analysis exist in more than one stack, and acquisitions leave teams on parallel tools.
How much capability is duplicated?
Between 20 and 30 percent in estates running two or more of the three vendors. That is spend consolidation recovers directly.
How do the compliance postures differ?
One vendor runs an active compliance program with a dedicated team, while the others enforce through usage reconciliation that arrives as a true up invoice.
What decides a true up dispute?
The counting rules. Concurrent peaks, shared workstations and idle session timeouts move the count materially, and that language is negotiable before signature only.
Should the estate standardize on one vendor?
Not entirely. Single vendor estates paid higher unit rates at renewal than estates keeping a credible second stack, because the consolidated vendor prices lock in rather than loyalty.
