Does that mean we should avoid consolidating?

No. One relationship, one invoice, one security review and a single place to ask questions are genuinely valuable, and consolidation for those reasons is sound. The point is to decide on those grounds rather than treating the discount as the whole argument.

Why does leaving cost so much more than expected?

Because you lose the rate on everything that stays, not only on what moves. A workload that would be cheaper elsewhere can still be more expensive to move once the effect on the rest of the account is included in the arithmetic.

How do we protect ourselves?

Work out what leaving would cost while nothing is wrong. Doing it once, calmly, turns an invisible constraint into a number you can weigh, and it is much harder to calculate honestly in the middle of a disagreement with the supplier.

Glossary

Volume discount

A lower unit price at higher usage, which rewards putting everything with one supplier and makes moving any of it more expensive.

In plain terms

Paying less per unit the more you use. The saving is real. What comes with it is that every additional thing you move to the same supplier makes it dearer to move anything away later, because the price you are enjoying depends on the total staying where it is.

Why it matters

Because the saving is quoted and the constraint is not. Consolidation is often the right decision, and it is a different decision from the one being presented, which is why it deserves to be made deliberately rather than accepted as an obvious economy.

How it works

The mechanism is a discount that grows with total usage, so the cheapest arrangement is always the most concentrated one. That is a genuine saving and it is also an incentive pointed at exactly one behaviour.

It ratchets. Each workload moved to the supplier lowers the unit price across everything already there, which makes the next consolidation more attractive and the next departure more expensive, and neither of those effects is stated anywhere.

The exit cost is the discount you lose on everything else, not just on the part you move. Taking one workload elsewhere can drop the whole account below a threshold, so a decision about a small piece is priced as though it were a decision about all of it.

That is why the leaving calculation looks so unfavourable when somebody finally runs it. Nothing has gone wrong: the arrangement was working as designed the entire time, and the design was to make this exact calculation come out this way.

Consolidation buys real things beyond the price. One relationship, one invoice, one security review and one place to ask questions are all worth something, and an organisation that consolidates for those reasons has made a sound decision that the discount happens to reward.

The protection is knowing your number before you need it. Working out what leaving would cost, once, while nothing is wrong, converts an invisible constraint into a figure you can weigh against the flexibility you are giving up.

What the discount is paying for

The reason to be clear-eyed rather than suspicious is that consolidation is frequently correct. Running four suppliers means four security reviews, four sets of terms, four invoices to reconcile and four relationships to maintain, and the overhead of that is real work absorbed by people who could be doing something else. An organisation that consolidates for those reasons and happens to receive a discount has made a good decision. The failure is narrower and specific: consolidating because the price is better, without noticing that the price is better precisely because concentration is what is being bought. The test that separates the two takes an hour. Write down what it would cost to move your largest single workload elsewhere, including the effect on the rate for everything that stays. If that number is one you would still be willing to pay, the arrangement is fine. If it is a number that would end the conversation, you have already made a decision about the next three years without having discussed it.

Seen in the wild

Moving a second workload to the same provider because the blended rate improves.
OpenRouter
A cheaper model elsewhere that stops being cheaper once the discount is lost.
Hugging Face
Automation spend consolidated for one invoice and one review rather than for price.
Make

Common misconceptions

People assume

It is simply a better price.

In fact

It is a better price with a condition attached: that the volume stays where it is. That condition is a constraint on future decisions, and it is the part of the offer that never appears in the quotation.

People assume

Moving one workload only costs what that workload costs.

In fact

It can drop the whole account into a worse band, so the price of moving a small thing is calculated against everything that stays. That is why an apparently minor decision returns a surprisingly large number.

Questions

Does that mean we should avoid consolidating?: No. One relationship, one invoice, one security review and a single place to ask questions are genuinely valuable, and consolidation for those reasons is sound. The point is to decide on those grounds rather than treating the discount as the whole argument.
Why does leaving cost so much more than expected?: Because you lose the rate on everything that stays, not only on what moves. A workload that would be cheaper elsewhere can still be more expensive to move once the effect on the rest of the account is included in the arithmetic.
How do we protect ourselves?: Work out what leaving would cost while nothing is wrong. Doing it once, calmly, turns an invisible constraint into a number you can weigh, and it is much harder to calculate honestly in the middle of a disagreement with the supplier.