There is a persistent myth circulating in startup circles that your Minimum Viable Product must actually work. That it needs to be a real piece of software, something a user can log into, click around, and get value from before you’ve validated anything. It’s an understandable assumption, but it’s wrong often enough to warrant serious unpacking.
Let me be precise about what I mean, because this isn’t a license to ship garbage. The real question is not whether your MVP needs to work. It’s what it needs to do, for whom, and what you are actually trying to learn.
-
-
-
- An MVP’s job is to validate your riskiest assumption, not to be a polished product. Functionality is one way to do that, not the only way.
- A landing page tests demand. A concierge MVP tests the workflow. A functional build tests retention. Know what you’re testing before deciding what to build.
- Hardware/physical products, regulated industries (healthcare, fintech), and developer tools all require something real to put in front of buyers. Everything else is negotiable.
- Hardware/physical products, regulated industries (healthcare, fintech), and developer tools all require something real to put in front of buyers. Everything else is negotiable.
- Resist the urge to add features before validating the core one. Every extra week of building before validation is a week of runway you can’t recover.
- A team running 12 landing page experiments in a year has 12 data points. A team building a polished v1 has one. At the validation stage, the volume of learning beats the quality of the product.
- A waitlist is weaker than a pre-order. A pre-order is weaker than a first payment. Know in advance what result would give you genuine conviction to proceed.
-
-
Table of Contents
ToggleThe Confusion Starts with “Viable”
Eric Ries defined MVP as “the version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort.” Notice what that definition doesn’t say”functional.” It says learning. The minimum viable product is a learning vehicle, not a product milestone.
The word “viable” gets misread as “usable.” But viable means capable of producing the signal you need. A working prototype is one way to generate that signal. It’s not the only way.
This distinction matters enormously at the pre-seed and seed stages, where you are burning real cash to answer questions that could be answered with far less. If you build a functional product before you have validated that anyone wants it, you have just done the most expensive user research imaginable.
You can apply the MVP lean startup agile methodology to build faster, test early, and iterate based on real user feedback rather than assumptions.
When Non-Functional is the Right Call
Consider Dropbox. Drew Houston did not build a working product first. He made a three-minute demo video. The video showed a product that did not exist yet, demonstrating a workflow that would exist if someone built it. Overnight, the beta waitlist grew from 5,000 to 75,000 people. That’s not a product. That’s a landing page and a video. But it was the MVP.
This works because the question Dropbox needed to answer was: Do people want seamless file syncing badly enough to sign up for something? You don’t need a functioning sync engine to answer that question. You need a compelling enough articulation of the problem and solution to get a behavioral signal (an email address, a credit card, a scheduled call).
Zappos is the canonical example on the other end. Founder Nick Swinmurn didn’t build inventory management software or a warehouse. He took photos of shoes at local stores, put them on a website, and when people ordered them, he bought the shoes at retail and shipped them manually. The “product” worked for customers. They got their shoes but the backend was completely non-automated.
Was it functional? Technically yes. Was it scalable? Absolutely not. Did it answer the question it needed to answer? Perfectly: Will people buy shoes online without trying them on first?
The lesson from both cases is not “don’t build”. It’s “don’t over-build before you know what to build.
The Three Questions That Actually Matter
Before you decide whether your MVP needs to be functional, answer these:
What’s your riskiest assumption?
Every startup has a hierarchy of risks. Most founders worry about whether they can build the thing. That’s rarely the riskiest assumption. Far more often, the biggest risk is whether anyone will pay for it, whether the problem is painful enough to change behavior, or whether you can reach the customer at a viable acquisition cost.
A functional product tests your engineering execution. A landing page tests your value proposition. Know which one you need to test first.
What constitutes a meaningful signal?
A waitlist signup is weaker than a pre-order. A pre-order is weaker than a first payment. A first payment is weaker than a second payment. Define up front what signal would make you believe you have validated your hypothesis.
“1,000 email signups” means nothing if you have not defined what conversion you would need to see from those signups to proceed with conviction.
What’s the cost of being wrong?
If you are building a B2B enterprise tool where a single customer might pay $200K ARR, your MVP might need to be functional enough to survive a security review and procurement process. The cost of a non-functional demo is low; the cost of losing a deal because your “product” can’t hold up under real use is high.
The calculus is different from a consumer app, where you can run a landing page test for $500 in Facebook ads.
Understanding MVP cost early allows you to plan your budget efficiently and avoid overbuilding features that do not contribute to validation.

The Concierge Model: Functional But Not Scalable
There is a middle path worth naming explicitly that is the concierge MVP. This is where your product works, but humans entirely power it behind the curtain. Airbnb’s early team manually photographed apartments, wrote descriptions, and managed bookings themselves. The user experience was real; the automation was not.
This approach is underused because it feels like “cheating”. Founders want to build software, not do manual work. But the concierge model gives you something invaluable: you deeply understand the workflow you’ll eventually automate, and you can validate your unit economics before you spend money on infrastructure.
If you are building a document processing tool and aren’t sure customers will pay $500/month, you can start by processing documents manually at that price point. If customers churn, you’ve lost a few customer-hours.
If customers stay and love it, you’ve validated your pricing and the core workflow with zero engineering.
When You Genuinely Need It to Function
There are contexts where a non-functional MVP is the wrong choice, and you should push back on anyone who suggests otherwise.
Hardware and deep tech are the obvious ones. If you are building a medical device, a physical product, or anything that requires the actual artifact to exist to be evaluated, a video won’t cut it. You will need to build something real, which is why hardware startups have longer runways and higher seed rounds.
Regulated industries are another. If your startup operates in healthcare, fintech, or insurance, your MVP may need to meet compliance standards just to put it in front of the right buyers. HIPAA compliance is not optional at the demo stage if you’re demoing to hospital procurement.
The third case is developer tools. If you are building an API, a CLI, or any product where your customer is a technical buyer, they will want to use it before they believe it. A landing page would not convince a backend engineer. They want to run the thing, hit errors, read your documentation, and form their own opinion. For developer-facing products, functional is the minimum.
The Trap of Over-Engineering the MVP
The reason this matters isn’t philosophical. It’s financial and temporal. The average seed-stage startup has 18–24 months of runway. Every month spent building a functional product before validating core assumptions is a month you can’t get back. Worse, it creates sunk-cost pressure. Once you have built something, it’s psychologically harder to throw it away, even when the signal says you should.
There’s also the problem of market feedback. A landing page experiment takes two weeks. A functional product takes six months.
The startup that runs 12 landing page experiments per year has 12 data points on what its customers want. The startup that spent that year building a polished v1 has one. Iteration speed at the validation stage is a genuine competitive advantage.
A strong approach to startup validation and MVP helps you confirm real demand before investing time and resources into full product development.
What “Minimum” Should Actually Mean
The word “minimum” is the one that the founders internalize the least. There is always a feature that feels essential, always a rough edge that feels embarrassing to show, always one more week of polish that would make the demo smoother. Minimum does not mean minimum quality, it means minimum scope. It’s the smallest thing that can produce a meaningful learning signal.
That might be a functional product. It might be a Wizard of Oz prototype where everything works, but nothing is automated. It might be a landing page with a waitlist. It might be ten customer discovery conversations and a pricing test done via email.
The point is to match the instrument to the measurement. Don’t use a mass spectrometer when a litmus test will do. And don’t build a full product when a well-crafted landing page will tell you everything you need to know to decide whether to build the full product.
The best founders I’ve watched operate treat the MVP stage as a scientific process with hypothesis, experiment, signal, decision. The product comes later. The learning comes first.
Final Thoughts
The question founders should be asking isn’t “does my MVP need to be functional?” It’s “what’s the cheapest, fastest way to prove my riskiest assumption is wrong?”
Sometimes that’s a landing page. Sometimes it’s a manual order fulfillment. Sometimes it genuinely is a working product. The answer changes depending on what you’re trying to learn, who you’re trying to learn it from, and what signal would actually give you conviction to keep going.
What doesn’t change is the underlying logic: every week you spend building before validating is a week of runway you’re betting on an assumption you haven’t tested. That’s not building, but that’s gambling with other people’s money, or your own.
The founders who get this right treat the MVP stage like a scientist, not an engineer. The engineer asks “Can I build it?” The scientist asks, “Should I?” Build the experiment first. Build the product when the experiment tells you to.
Dropbox didn’t need a sync engine to know people wanted one. Zappos didn’t need a warehouse to know people would buy shoes online. What they needed was the discipline to ask the smallest possible question before writing the first line of code.
That discipline, choosing the right instrument for the measurement, resisting the urge to over-build before you’ve over-learned, is the actual skill. Functionality is just one tool in the kit.
Use it when it’s the right tool. Not before.
FAQs
What is a functional MVP?
Can an MVP be a landing page with no product behind it?
What is a concierge MVP, and when should you use it?
Do investors fund non-functional MVPs?
When does an MVP actually need to be functional?
- Hardware or physical products: the artifact must exist to be evaluated.
- Regulated industries like healthcare or fintech, where compliance requirements kick in at the demo stage.
- Developer tools: technical buyers want to run the product themselves before believing it exists.
- In all other cases, the level of functionality should match what's needed to test your specific hypothesis.




