Model innovation: a constructed example of two-part pricing arithmetic

This is a constructed example. No company is described, no figure is real, and none of it is a benchmark. The value is in the arithmetic and in which arguments turned out to be answerable.

The situation: a firm charging a single per-unit price, growing, and losing money on its largest customers. The proposals in the room were a price rise, a minimum order, and a move to a two-part structure with a base fee plus a lower unit rate.

What to take away

What the records showed

Realized price against list was the first thing computed, and it had never been computed before. List was 100 a unit. Collected cash divided by units shipped came to 88, because of volume rebates, promotional credits, and freight the firm had agreed to absorb without recording it as a discount.

So the argument about raising the list price to 105 was an argument about a number the firm did not charge. That single calculation reframed the discussion, and it took an afternoon.

The three options, priced

The firm's unit cost was 60, so contribution at the realized 88 was 28 a unit rather than the 40 everyone had assumed.

The two-part option was modeled at three splits, all reaching the same total at 300 units: 0 base with 12 a unit, 1,200 base with 8 a unit, and 2,400 base with 4 a unit.

At 150 units those become 1,800, 2,400 and 3,000. At 600 units they become 7,200, 6,000 and 4,800. Same structure family, opposite effects at either end of the book. The choice between them is a decision about which customers you want.

The argument that could be settled

Whether the discount authority was the real problem. On a 40% margin, a 10% discount needs 33% more volume to hold gross profit and a 20% discount needs double. The firm's average discount was well inside its own authority limits and nobody had ever run that arithmetic in a negotiation.

Fixing the discount threshold was reversible next week and required no customer conversation at all. It was done first, alone, for exactly that reason. The layer discipline behind that sequencing is set out in pricing architecture.

The argument that could not be settled

Whether light users would accept a base fee. Nothing in the firm's records answered it, because the firm had never charged one. Proposals to model it from existing behavior were rejected on the ground that behavior under one structure is weak evidence about behavior under another.

So the base fee went to new customers only, with an end date on the trial and a written stop condition. That is slower than the firm wanted and it is the only honest route when the record cannot answer the question.

What the transition actually cost

Two structures held in billing at once for three quarters. Sales quoting both. Support answering questions about both. A period where finance could not produce a clean margin figure because the book was a mixture.

None of that appeared in any of the three options as modeled. It was budgeted separately once someone noticed, and it was larger than the first year's expected gain from the change. That is normal. Firms that fund a transition out of its expected return stop halfway and end up running two models with the benefits of neither.

The cost behavior underneath is worked out in unit economics, and financing a change before it pays is treated directly in MIT's entrepreneurial finance course.

What this case does not show

It does not show that two-part pricing is better, that 88 against a list of 100 is typical, or that the firm was right. It shows which questions the records could answer and which they could not, and that the distinction costs an afternoon before any restructuring begins.

What counts as a model change at all is set out in model innovation, and the classification that decides which family of change you are in is in business model types. Where the differences land between competing business buyers, the federal price discrimination provision is the reading to take to a lawyer before a tier structure is published.

Common questions

Why compute realized price first?

Because every other calculation sits on it. A firm arguing about a list price it does not collect is arguing about nothing, and this is the single most common version of that mistake.

Could the firm have run all three options at once?

It could, and then no result would have been attributable to any of them. Two changes in a quarter buys one year and no information.

Is a base fee always the answer for light users?

No. It is the answer when readiness costs you money whether or not the customer buys. Where your cost is genuinely all per-unit, a base fee is a charge you cannot justify when asked.

How long did the trial need?

A full sales cycle plus one renewal, which the firm had to be told twice. A structure change read at week six measures how fast the sales team adapted.