A vendor offers to remove the meter. The pitch is a four-times return and freedom from counting. Here is how the buyer should actually evaluate it, and why the answer usually turns on a number nobody in the room has looked up.
NovaTech runs a product analytics platform, UX-Hero, across its customer and internal applications on a per-MAU contract. Usage has outgrown the meter: they are at 1.55M monthly active users against 1.2M contracted, paying overage on the difference. UX-Hero proposes replacing the meter entirely with an unlimited enterprise agreement, supported by a commissioned study claiming a multi-million-dollar return.
Two companies and one platform, all fictional. The analysis is the point.
The question in the room is usually "is the unlimited price good?" That question has no answer, because the ELA is a fixed cost and the metered structure is a variable one. Comparing them at a single point in time compares nothing.
The real question is where the two cost curves cross, and whether you have any credible basis for forecasting past that point. Everything else is decoration.
In this case the crossover sits at 2.05M MAU, about 32% growth from today. At the growth rate NovaTech can actually defend, the two structures land within a rounding error of each other over three years. Which is the most useful possible finding, because it means price does not decide this. It moves the decision to the arguments neither side has quantified.
Start with the overage rate slider. It is the input that most often has not been looked up, and it moves the answer more than anything else on the page. Then try the suppressed-demand slider: on the base case it takes surprisingly little unlocked usage to flip a no into a yes.
Fictional company and vendor · Illustrative figures · Best experienced on desktop
Commissioned ROI studies are not dishonest, but they are built to be quoted, not audited. Three things reliably need correcting.
Soft savings are not cash. A recovered hour is a saving only if the hour comes out of a budget. If nobody removes headcount or spend, the benefit is a nicer day, not a number the CFO can bank. That is not cynicism, it is the difference between a P&L and a feeling.
Linear scaling breaks. Taking a per-employee benefit from a composite organization and multiplying by the ratio of workforces assumes the benefit is uniform across every role and every site. It never is. The further the scale factor, the more the estimate is doing work the evidence cannot support.
Avoided overage is the exception. It is contractual, verifiable, and already in your invoices. In the decomposition it takes a zero haircut, and it is usually the only line that survives intact. It is also, notably, the smallest number in the vendor's deck.
It assumes a single blended consumption unit. Real contracts often meter several populations separately at different rates and different growth trajectories, and those should be modeled independently before being summed.
It treats growth as smooth. Adoption is lumpy, and a single large deployment can cross the break-even in a quarter.
It does not price optionality directly. There is genuine value in not having to forecast usage, in removing the internal friction of true-up conversations, and in letting teams instrument without asking finance. That value is real and this model does not capture it, which is an argument for paying somewhat above the break-even rather than exactly at it.
And it says nothing about renewal leverage. Unlimited resets your baseline. The next negotiation starts from the ELA price, not the metered one, which is a cost that shows up in year four and never appears in a three-year model.
Most vendor evaluations end with a recommendation. Better ones end with the conditions under which the recommendation would change. This model is built to produce the second kind: three assumptions carry the whole case, and the tool makes it obvious which one you should go look up before the next meeting.