Article24 Feb 20257 min read
Your MVP is a thin slice, not a bare floor
The classic MVP pyramid gets cut the wrong way. Shipping only the functional layer gets you something that works and nobody wants. Slice vertically instead - a narrow product that is functional, reliable, usable and genuinely nice to use, all at once.
Gareth turns tangled business problems into clear delivery plans, keeping strategy, engineering and clients pulling in the same direction.
Stack the things a product has to be and you get a familiar pyramid. Functional at the base, because it has to do the thing. Reliable above it, because it has to keep doing the thing. Usable above that, because people have to be able to work out how. And emotional design at the top, because the difference between tolerated and loved lives there. As a hierarchy it is hard to argue with, and it has shaped a great deal of good thinking about user needs.
The trouble starts when someone asks what the minimum viable version is. Look at that pyramid with a deadline in your hand and the answer seems obvious: take the bottom layer. Ship functional, leave reliable for hardening, leave usable for a later design pass, and treat emotional design as the thing you'll do if there's budget at the end. It is a tidy reading of the diagram and it produces software nobody wants to use.
Minimum viable product
Horizontal minimum, vertical minimum
Cut the pyramid horizontally and you get the full breadth of the product at the lowest possible quality: every feature present, nothing pleasant, nothing dependable. Cut it vertically and you get a narrow product at full quality: fewer features, but the ones that are there work, hold up, make sense and feel considered. Both are smaller than the finished thing. Only one of them is something a person would choose to use twice.
This correction isn't ours - it is usually credited to Jussi Pasanen's redrawing of the pyramid, building on Aarron Walter's hierarchy of user needs - but it is one of the most useful pictures in product work, because it reframes the question. "What is the least we can build?" invites the horizontal cut. "What is the smallest complete experience we can build?" invites the vertical one, and those two questions produce entirely different roadmaps from the same backlog.
Viable was never a synonym for barely working.
Why the horizontal cut keeps winning anyway
It wins because it is easier to plan and easier to sell. Breadth is legible: a list of features maps neatly onto a schedule, and a stakeholder can count the things they asked for. Depth is harder to point at, because "reliable" and "considered" don't decompose into tickets that look impressive in a status report, even though they are the difference between adoption and a launch nobody returns to.
It also wins because the deferred layers sound deferrable. Reliability sounds like something you add once traffic justifies it. Usability sounds like something a designer tidies later. Emotional design sounds like polish, and polish sounds optional. Each of those is a reasonable-sounding sentence, and each one is describing rework, because retrofitting reliability into a system that was never shaped for it is a rebuild wearing a smaller name.
What a thin slice looks like in practice
The slice is chosen, not trimmed. You pick one journey that matters most - usually the one that proves the value proposition fastest - and you build that journey properly through every layer. Everything else is absent rather than degraded, which is the crucial distinction: an absent feature is a decision, whereas a degraded one is a liability you have shipped and will have to defend.
- Functional: the one journey does what it promises, end to end.
- Reliable: it holds up under real conditions and fails visibly rather than silently.
- Usable: someone unfamiliar can complete it without being talked through it.
- Emotional: it feels like it was made on purpose, by people who cared about it.
Held to that standard, the first release is genuinely narrow, and that narrowness is the point - it is what pays for the depth. A launch with one polished journey earns the right to a second one; a launch with nine shaky journeys spends its credibility on the first day and then asks for more time to fix what it already showed people.
The question worth asking instead
When scope has to come down, and it always does, the useful question isn't which layer to drop. It's which journey to keep. Cutting features is a product conversation you can have honestly with anyone; cutting quality is a promise you make quietly and pay for publicly, usually at the exact moment the thing finally has users.
Minimum, viable, product. All three words are doing work, and viable is the one that gets ignored. A bare functional floor is minimum and it is a product, but it was never viable - and a thin, complete slice through all four layers is the smallest thing that actually is.
