Quoted token pools exceeded measured burn by 20 to 40 percent on first draft, because the vendor estimate assumes a utilization real estates rarely reach

The agreement trades named seats for a token pool, so the only number that matters is how many tokens your teams actually burn in a year.

Prepared by Redress Compliance · August 18, 2026 · Autodesk agreement negotiations. 15 to 25 negotiations reviewed, 2024 to 2025.

A pooled token balance rather than a seat count. You commit to an annual token spend and the vendor grants access to most of the product catalogue, with the structure set out on the enterprise agreement page.

Tokens are consumed per product session, and each product carries a different rate. A heavy product run daily burns far more than an occasional viewer, which is where the population split starts to matter.

Three frequency bands do the work

• Daily power users are usually cheaper on a named subscription than on tokens.
• Weekly or monthly users are often cheaper on the pool.
• Viewers and reviewers are almost always cheaper on flexible tokens.

Pull twelve months of access logs from your own administration console, map each user to a frequency band, then apply the token rate per product against the published pricing reference. The result is measured burn, and it is the only number worth negotiating from.

Vendor estimates assume full utilization on every band. Real estates rarely reach it anywhere except the daily population, which is precisely the population the pool prices worst.

Across roughly 15 to 25 Autodesk enterprise agreement negotiations reviewed between 2024 and 2025, the vendor token estimate ran high in almost every case. Three patterns recur.

• Quoted token pools exceeded measured burn by 20 to 40 percent on first draft.
• Annual uplift clauses started at 8 to 12 percent and dropped to 3 to 5 percent once challenged.
• Low frequency users on the pooled agreement cost 2 to 3 times more than the same users on flexible tokens.

In writing, at signature, ideally at or below general inflation. Left open it compounds against a committed spend that was already 20 to 40 percent too large.

• Ask for the uplift in basis points rather than a vague percentage.
• Tie any increase to a published index rather than to vendor discretion.
• Add a downward true up if measured burn falls below the pool.

Treat anything above 5 percent as negotiable

The vendor has signalled steady list increases through its corporate news channel, which is exactly why the clause needs an index rather than a promise.

Timing, term length, and a credible flexible alternative for the light users. The vendor wants multi year commitment and predictable revenue. You want a right sized pool and a capped uplift.

The light users cap the pool size

A modelled flexible or named subscription alternative for the occasional population caps how large a pool can be pushed. Put it on the table with numbers attached, since those users cost 2 to 3 times more inside the pool. The comparison sits in the flexible token guide.

Never let the true up clause auto expand the pool without a review on your side. That clause is how a 20 to 40 percent overquote becomes permanent, and the perpetual comparison in the subscription cost analysis shows what the alternative baseline looks like.

Two cuts of the engagement file describe the two halves of the bill.

Both are measurement problems rather than pricing ones. The logs already exist, which makes this one of the cheapest corrections available on any renewal.

How does the pooled agreement price?

You buy a token pool rather than a seat count, committing to an annual token spend in exchange for access to most of the catalogue. Tokens are consumed per product session at different rates per product.

How far above real usage is the first quote?

Between 20 and 40 percent above measured burn on first draft, because the vendor estimate assumes a full utilization that real estates reach only in the daily population.

How do you measure real burn?

Twelve months of access logs from your own administration console, with every user mapped to a frequency band and the token rate applied per product. That number is what to negotiate from.

Which users are cheapest on the pool?

Weekly and monthly users, who burn 40 to 60 percent of the assumed tokens. Daily users are usually cheaper on a named subscription and occasional users on flexible tokens.

How much do occasional users cost inside the pool?

Two to three times more than the same users on flexible tokens. They burn 10 to 25 percent of the assumed rate while the pool charges for full utilization.

What does the uplift open at?

Between 8 and 12 percent, and it settled at 3 to 5 percent once challenged in the negotiations reviewed. It is the single biggest cost driver if it is left open.

How should the uplift be drafted?

In basis points rather than a vague percentage, tied to a published index rather than vendor discretion, with a downward true up if measured burn falls below the pool.

How much shelfware is normal?

Between 20 and 40 percent of the pool in the engagements reviewed, which is the same overquote arriving as a bill instead of as a proposal.

Can flexible tokens be used as leverage?

Yes. A modelled flexible or named alternative for the light users caps how large a pool can be pushed, provided it arrives with numbers rather than as a mention.

What is the most dangerous clause?

The true up that auto expands the pool without a buyer side review. That is the mechanism by which a first draft overquote becomes a permanent baseline.