Method 33 of 100 · Stage 4: Build a no-code prototype

Concierge Test

Deliver the service by hand to your first 5 to 10 paying customers and log every step before you automate anything.

What is a Concierge Test?

A concierge test (also called a concierge minimum viable product, a minimum viable product being the smallest version that delivers the result) means you deliver the result your future product promises completely by hand, to a handful of real customers who KNOW a person is doing the work. You are the hotel concierge: you take each request personally, sort it out and come back with the answer.

The goal is to learn the real workflow before you write a line of code. You find out which steps customers care about, which ones they ignore and how long each one takes. Eric Ries tells a well-known version in his book The Lean Startup: the founders of Food on the Table, a meal-planning service, started by serving their first customer in person, building recipes and shopping lists by hand before they had any software.

When should you use it?

It fits best when:

  • Your product is a service or software that does a job people would also pay a person for: planning, matching, bookkeeping, sourcing, scheduling, research.
  • You have finished problem interviews and a few people said “I would pay for that”.
  • You are unsure what the finished product should do, which steps matter and in what order.

Skip it for physical products you have to manufacture, and for products whose whole value is instant speed at scale (a search engine, a real-time translation app). If customers would hate knowing a human is involved, look at the Wizard of Oz test instead.

How do you do it, step by step?

  1. Write the promise and a price. One line: what the customer gets, by when, for how much. “Every Monday you get a one-page summary of last month’s spending, by WhatsApp, for €40 per month.” Charge from day one, even a small amount. Free customers are polite; paying customers complain, and complaints teach you.

  2. Recruit 5 to 10 customers by hand. Go back to the people from your interviews first. Message: “You told me [problem] last month. I’m doing this by hand for 10 people before I build anything. It’s €40 a month, you can stop any time. Want one of the spots?” Our rule of thumb: you need to ask 20 to 40 people to fill 5 to 10 spots.

  3. Set up a process log before the first delivery. A Google Sheet with one row per step per customer: date, customer, step, minutes spent, tool used, what went wrong, what the customer said. Start a timer (Toggl Track or your phone) every time you work. This sheet becomes your product specification later.

  4. Deliver, then ask three questions. After each delivery, ask: “What did you do with it?”, “What did you skip?” and “What did you have to do yourself afterwards?” The third question reveals the step your product is missing.

  5. Change one thing per round. If customers ignore half of what you send, cut it next week, and when they keep asking for the same extra, add it. Write each change in the log so you can link it to what happened next.

  6. Sort every step into three piles after 10 deliveries. Anything identical every time goes in automate. Work that needs judgment goes in keep human for now, and whatever nobody noticed lands in drop. Expect the third pile to be bigger than you think.

  7. Do the money maths. Hours per customer times your hourly rate, compared with the price. If one customer takes 3 hours and pays €40, software has to remove most of those hours. Your unit economics start here.

What does a worked example look like?

Say Hana wants to build software that turns a small café’s receipts into a monthly money summary. Before building anything, she runs a concierge test with 6 cafés in Antwerp at €40 per month.

Each café owner photographs receipts and sends them to Hana’s WhatsApp Business number. Hana types them into a Google Sheet, sorts them into categories and sends a two-page PDF at the end of the month. Her process log shows the first month took 2 hours 10 minutes per café, mostly reading crumpled receipts.

The three questions change everything. Four of the six owners never opened the PDF. What they did read was her short WhatsApp message: “Milk was 31% of your costs this month, up from 24%.” Two owners asked her to warn them mid-month when a supplier raised prices. Hana drops the PDF, keeps the short message and adds a mid-month alert. Her future product shrinks from “bookkeeping software” to “a weekly cost alert for cafés”, which is far cheaper to build and easier to sell.

Which free tools help?

Tool What you use it for Free plan
Google Sheets The process log and the work itself Free with a Google account
WhatsApp Business Receiving requests and sending results Free app
Toggl Track Timing every step you do by hand Yes, for individuals
Stripe Payment Links Taking a monthly payment without a website No monthly fee, you pay per transaction
Calendly Booking short check-in calls with customers Yes, with limits

What mistakes should you avoid?

  • Doing it for free. A free service tests whether people like gifts. Charge something, even a symbolic amount.
  • Not timing yourself. Without minutes per step you cannot tell what to automate first or whether the price works.
  • Over-delivering to impress. If you add extras nobody asked for, you learn what you like doing, and the product drifts away from the customer.
  • Automating after customer two. Wait for at least 5 customers and 10 deliveries. The workflow keeps changing until then.
  • Treating it as your business forever. A concierge test has an end date. Set it on day one, say 4 weeks.

How do you know it worked?

Our rule of thumb for a go: at least 5 paying customers, at least 3 of them renew or ask for the next round without a nudge, and your log shows the steps that would become software. You also want one surprise, a step you planned that nobody needed or a request you never expected.

No-go: customers stop after the first delivery, or they only stay because the price is tiny. Pivot: they stay, but for a different part of the service than you planned. Follow that part.

What should you do next?

When the workflow settles, put a simple front on it with a form-based service, or hide the manual work behind an interface with a Wizard of Oz test. For business customers, turn the concierge round into a paid pilot with clear success criteria. The startup game for women by Fe/male Switch walks you through a concierge round step by step with an AI co-founder.

FAQ

What is the difference between a concierge test and a Wizard of Oz test?

In a concierge test the customer knows a person does the work and talks to you directly. A Wizard of Oz test hands the customer what looks like a finished product while a person works behind the screen. Concierge teaches you the workflow, Wizard of Oz tests whether people use the product itself.

How many customers do you need for a concierge test?

Five to ten is enough for most services. Fewer than five and one unusual customer dominates your conclusions. More than ten and you are running a small agency, which leaves no time to learn.

Should I charge for a concierge test?

Yes. Even €10 changes how people behave: they show up, they tell you when something is wrong and they decide whether to renew. Their renewals are your strongest signal that the product deserves to exist.

Updated 26 October 2026 · Written by the Fe/male Switch team

Methods that go well with this one

Method 34

Wizard of Oz Test

Show customers a working product interface while a person quietly does the work behind it, to test real use before you build.

No-code prototypeIntermediate2 to 3 weeks
Method 39

Form-Based Service

Sell a service through one intake form, deliver the result by hand by email and automate the dull steps only once orders repeat.

No-code prototypeBeginner1 day to set up, 2 to 4 weeks to run
Method 21

Problem Interviews

Ask people about the last time they hit the problem, never about your idea, to learn whether the problem is real and costly.

Customer discoveryBeginner45 minutes per interview, 2 to 3 weeks for 10 to 15
Method 47

Paid Pilot

Run a short, paid trial with a business customer, with scope, success criteria and the path to a full contract agreed before you start.

Price and pre-sellAdvanced30 to 90 days per pilot
Method 98

Unit Economics (CAC and LTV)

Calculate what a customer costs to win and what they bring in over time, per channel, to see which growth efforts pay for themselves.

Measure and improveIntermediate2 to 3 hours

Go deeper