Minimum Marketable Product

For a long stretch the phrase minimum viable product (MVP) covered two jobs with different requirements. One is building the smallest thing that shows whether anyone wants the idea. The other is charging money for that thing. The lists differ, the name didn't, and so the same product got called finished by one side of the room and nowhere near ready by the other.

Pagedeck had that argument. Sofia was a week from signing a paid enterprise contract and asked about invoice formats and the refund policy; engineering answered that it was an MVP and those came later. For Sofia they could not come later. Charging money means an invoice goes out, and it means a written answer to what a customer gets back after quitting halfway through a term. Neither side was wrong. Both were describing the same build against different checklists. A product that was enough to learn from was not enough to sell.

Minimum marketable product (MMP) names the gap between those two. It keeps the instinct to build less, but it measures the minimum with a different ruler — not what the product can teach you, but whether a transaction runs end to end once you put a price on it.

Move the baseline from what can we learn to can we charge for this, then let the new question decide what stays in scope. This is not an argument for more features. It splits the remaining work into things a transaction cannot happen without and things a transaction happens fine without, and fills in only the first group.

Five pieces usually have to be there.

1. A way to take money. Either wire up a payment gateway, the service that passes card charges through to the bank, or, for enterprise buyers, set up a process that sends an invoice and confirms the transfer. Retrying the charge when a card expires belongs to this piece too.
2. A reason the number is what it is. Pricing decides how much and the billing model decides what the amount is based on; without both there is no figure to put on the screen. If it is a subscription charging a fixed amount each cycle, the next charge date has to be visible as well.
3. A way to undo it. A refund policy for people who paid by mistake, written down in advance — the full amount within so many days, or the amount minus whatever they used.
4. The promises, in writing. Terms of service set the conditions for using the product. Nobody misses them while the thing is free, but money creates disputes, and this is the document a dispute gets settled against.
5. Somewhere to ask. A person who answers "my card was declined" and a time by which they answer. With no channel, refund requests go to the card issuer instead.

None of the five teaches you anything. While you are serving ten people by hand to check demand, there is no need for billing or terms. Their absence at the MVP stage is a decision rather than an oversight, and the MMP is the threshold where you decide to stop making it. That makes it a decision about timing rather than a list of features to finish.

It comes up in the meeting right after the hypothesis is settled and the product is expected to produce revenue. Once product-market fit — the state where what you built matches real demand in a specific market — starts to show, the next question is when you can start charging. Re-sorting whatever is left on the roadmap into items that block a sale and items that do not is the work of defining an MMP.

It shows up again when free users get moved onto a paid plan. At Pagedeck, which starts people on a free plan and hopes teams upgrade, the day the checkout screen opens is the day the MMP is done. If a charge fails on that day, users doubt the company rather than the feature.

Selling to companies makes the list longer. Contracts, tax handling, a named account owner and security questionnaires matter to none of the individuals paying by card. That is how a product ends up sellable to individuals and unsellable to enterprises at the same time. How high the threshold sits depends on who the buyer is.

The reverse case is just as clear. Aiming at an MMP before you know whether anyone wants the thing spends two months attaching billing and terms to something nobody asked for. Confirmation comes first.

"The goal this quarter is the MMP — billing wired up and terms published."

A declaration that the quarter is about becoming sellable, not about adding features.

"By MVP standards this is done. By MMP standards it isn't."

Enough to learn from, still missing something the team needs in order to charge.

"Localization doesn't belong in the MMP. Our first customers are all domestic."

A proposal to push an item outside the threshold because it doesn't block a sale.

When Pagedeck opened its first paid plan, Mia put the remaining items on the wall in two columns. The left column held everything that had to exist before anyone could press the pay button, and the right column held everything that was merely nice to have. Payment method entry, receipts, refund handling and the terms consent screen went left. A billing history chart and usage reports went right.

Grace wanted the two right-hand items in the same release. Mia set the date against the left column alone, because the money still arrives without the right column and stops arriving if one item on the left is missing.

The most common reading is that an MMP is the next version of an MVP. The names rhyme and one comes after the other in time, so it sounds like a bigger MVP. What differs is the ruler rather than the size: an MVP measures the minimum against learning, an MMP measures it against a transaction. A minimum lovable product (MLP) uses a third ruler, holding out for enough polish that people actually like the thing. Lined up by size instead, the three turn into "add a few more features and you have an MMP," and items that never appear on a feature list — failed-charge handling, terms — stay missing while the launch date gets set.

The word minimum also tends to fall out. Being told to sell makes people picture what products that already sell well have, and a competitor's feature list ends up on the desk. Then MMP means build everything. The test is the first customer, not the competitor: anything the first ten paying customers do not need in order to pay goes later. Miss that distinction and revenue arrives six months late, by which time the demand you confirmed has cooled.