The Concierge MVP
Learn exactly what to build by becoming the product yourself. A Concierge MVP delivers the value by hand — openly — for a few real customers, so you see what they actually need before writing any code. Here's how to run one, what to capture, and when to use it.
A classic pretotyping method · Built at Google, taught at Stanford
Banked
$68M+
Fund the best, save on the rest.
What is a Concierge MVP?
A Concierge MVP — sometimes called a concierge test or a manual MVP — is a pretotyping method where you deliver the value of a product manually and openly — as a high-touch, human service — before building any technology. Unlike the hidden human of a Wizard of Oz test, the customer knows you’re doing it by hand. By becoming the product yourself for a handful of customers, you watch every hesitation and workaround and learn exactly what people need. It produces the richest qualitative insight of any pretotyping technique — before you commit a line of code. It’s a natural next step when you’re working out how to validate a business idea, and pairs well with a fake door test to gauge demand before you learn what to build.
$5M
Food on the Table raised after starting by planning one customer's meals by hand
Food on the Table
5–10
customers is enough to learn deeply from a concierge test
Pretotyping playbook
90%
of new products fail — usually by solving the wrong problem
Harvard Business School / Christensen
How to run a Concierge MVP in 6 steps
Deliver the value by hand, watch closely, and let what people actually do — not what they say — tell you what to build. See all 10 pretotyping methods →
Pick one real customer
Start with a single person in your target market — or a tiny handful. Depth beats breadth here; you're learning, not scaling.
Deliver the value by hand, openly
Do for them, manually, what the product would eventually do. Be upfront that it's a hands-on service, not a finished tool.
Sit with them and watch
Observe every step in person or on a call. See where they hesitate, what they skip, and the workarounds they reach for.
Record what they need — not what they say
Capture the real decisions, constraints, and edge cases. Stated preferences mislead; observed behaviour is the evidence.
Find the patterns across customers
Repeat with a few more people and look for what recurs. The shared needs are what your eventual product must nail.
Decide what to automate first
Use the insight to prioritise the riskiest, highest-value piece to build — or to walk away before wasting engineering effort.
Both are manual — only one is visible
A Concierge MVP openly delivers the value by hand to discover what customers need. A Wizard of Oz MVP hides the human to test whether the solution is valued. Pick the one that matches the assumption you're testing. See the Wizard of Oz method →
| Dimension | Concierge MVP | Wizard of Oz MVP |
|---|---|---|
| Do customers know it's manual? | Yes — the high-touch service is explicit | No — the work appears fully automated |
| Best for testing | Deep problem discovery — what customers actually need | Whether the solution / output is valued |
| Data you get | Rich qualitative insight from watching people up close | Behavioural signal at a small, automated-feeling scale |
| What you build | Often nothing — you deliver the value in person | A polished interface; humans fake the engine |
| Reach for it when | You don't yet understand the customer or workflow | The risk is in expensive tech (AI, ML, automation) |
Concierge MVPs, answered
What is a Concierge MVP?
A Concierge MVP is a pretotyping method where you deliver a product's value manually and openly — as a high-touch human service — before building any technology. You become the product for a handful of customers, watching exactly how they behave. It produces the richest qualitative insight of any validation technique because you witness every hesitation and workaround first-hand.
What's the difference between a Concierge MVP and a Wizard of Oz MVP?
Both rely on humans doing the work by hand, but visibility is the difference. In a Concierge MVP the customer knows the service is manual and high-touch — it's for deep problem discovery. In a Wizard of Oz MVP the human is hidden and users think it's automated — it tests whether the solution itself is valued. Concierge teaches you what to build; Wizard of Oz tests whether a built version would land.
How many customers do you need for a Concierge MVP?
Usually just 5–10 — sometimes you start with one. The goal is depth, not statistical significance. You're trying to understand the workflow, constraints, and real needs so well that you know exactly what to build, so a small number of closely-observed customers is far more useful than a large, shallow sample.
What's the difference between a Concierge MVP and an MVP?
A Concierge MVP is a manual, non-scalable service you run to learn what customers need — there's no product yet. An MVP (minimum viable product) is a real, working, minimal product people use and pay for. Concierge comes first: it's how you discover what the MVP should actually do before you build it.
Is a Concierge MVP the same as a manual MVP?
Yes — "concierge test" and "manual MVP" are common names for the same idea: delivering a product's value by hand before building it. The one distinction worth keeping is visibility. A Concierge MVP is openly manual — the customer knows it's a high-touch service — whereas a Wizard of Oz MVP hides the manual work behind an automated-looking interface.
What are real examples of Concierge MVPs?
Food on the Table founder Manuel Rosso started by visiting one customer's home, reviewing her preferences and manually planning her weekly meals — the company later raised $5M. Before automating, Wealthfront's founders manually managed portfolios for a few clients and discovered customers cared most about tax efficiency, which shaped their entire product. Both learned by being the product first.
Want help becoming the product before you build it?
Twenty minutes with Leslie Barry. We look at one idea on your roadmap, design a concierge test, decide what to capture, and tell you whether pretotyping is the right next step. No pitch.