Doing It by Hand First
A concierge test delivers the promised outcome manually for a few real customers before any product is built. · 11 min
There is a build even cheaper than the smallest product: no product at all. Instead of making a machine that delivers the result, you deliver the result yourself, by hand, for a few real customers. You become the thing you plan to sell. It does not scale, and it is not meant to. Its job is to put the real promised outcome in front of real people for the least possible cost — so you learn whether the outcome is worth paying for, and exactly what delivering it takes, before you build anything.
Guess before you learn
You think people will pay for a service that plans their week of workouts. Before building an app, what would count as a concierge test?
A concierge test means you become the product for a few real customers: you deliver the promised result by hand, watch what they do with it, and learn whether it is worth paying for — all before writing a line of code. The landing page only measures interest. The rough app is still a product build. Friends' opinions are neither delivery nor payment.
9–12
3–5
A concierge is a person at a hotel who does things for you by hand. A concierge test works the same way: instead of building a product, you do the whole job yourself for a few real customers. You find out whether the result is something they actually want — before you spend anything building it.
6–8
A concierge test delivers the exact outcome you plan to sell, but by hand, for a few real customers. There is no product yet — you are the product. You do the work manually, watch closely, and learn whether people value the result enough to come back. It costs your time, not a build.
This does two things at once. You find out whether the outcome is worth paying for, and you learn exactly how to deliver it — which is what you would otherwise have guessed at, wrongly, by building first.
9–12
A concierge test is the manual delivery of your value proposition to real customers before any automation exists. You handle each customer individually, doing behind the scenes whatever a finished product would eventually do. Because nothing is built, the test is cheap to start and quick to change; because the outcome is real, the customer's response is genuine evidence, not a reaction to a description.
K–2
Pretend you sell lunches. Before you buy a lunch machine, make five lunches yourself, by hand. Give them to five real people. Watch if they eat them and ask for more. Your hands taught you before any machine did.
If they finish it and want another, that is a real yes. You learned it with your hands, five lunches, and no machine at all.
Undergrad
The concierge minimum viable product, named in the Lean Startup literature, trades scale for learning. Rather than build software to serve many, you serve a handful by hand, absorbing the operational reality a specification would hide. Two lessons emerge: whether customers value the delivered outcome, and what delivering it actually requires — the second often reshaping the product you eventually build.
Its limit is deliberate. Manual delivery does not scale, and it is not meant to. The point is to reach the moment of truth — a real customer receiving the real outcome — for the least cost and the shortest delay, and to change freely while the numbers are still small.
Postgrad
A concierge test isolates demand-side risk from build-side risk. By delivering the outcome through human labor, you hold the product's existence at zero and vary only whether the outcome commands value and repeat behavior. You also convert tacit operational knowledge into explicit knowledge: the manual runs surface the edge cases and hand-offs that a premature specification would have encoded wrongly.
The economics invert those of a product. A product carries high fixed build cost and low marginal cost; the concierge run carries near-zero fixed cost and high marginal cost per customer. That structure is correct for small numbers and ruinous at scale — which is precisely why it belongs where the numbers are still small.
concierge test
Delivering the promised outcome by hand, to a few real customers, before any product exists — so you learn whether the outcome is worth paying for, and how to deliver it.
The rule that keeps a concierge test honest is that the customer must be real and, ideally, paying. Serving friends for free teaches you little, because they may be paying out of kindness rather than real need. Serving strangers who hand over money teaches you almost everything, because people part with money only when the outcome is worth it to them. Keep the number small — five or so — so you can watch each person closely and notice what a larger, automated launch would have buried.
Why is this true?
Why deliver by hand instead of building even a small product first?
Because hand delivery is the cheapest way to put a real outcome in front of real customers and watch what they do. You reach the moment of truth for almost nothing, you keep the freedom to change course, and you discover what delivering the outcome actually takes — which is exactly what a first product would have guessed at and gotten wrong.
Delivering by hand is cheap to start but costly for each customer, because your time does not scale. A product is the opposite: costly to build, then cheap for each customer served. So a concierge test wins while the customers are few, and a real product wins once they are many. The plate below shows where the lines cross — the point past which doing it by hand stops making sense.
Design a concierge test for a resume-review service — the steps fade as you master them
A marked-up resume with three concrete fixes, back within 24 hours.
You read each resume yourself and type the notes in a document.
Five paying customers, no more, so you can watch each one closely.
At least three of five come back or refer a friend.
Doing it by hand tells you two things a description never could: whether the outcome is worth paying for, and what delivering it actually takes. Both will surprise you. Next, you will set these cheap tests turning in a loop — build a little, measure what happens, and let the result choose what you do next.
Note
Want to run the numbers behind the crossover point yourself? Managerial Accounting — an accession in Commerce — covers fixed and variable cost in one short unit.
Practice — new ink and old, interleaved
1.Why is a problem that happens daily usually worth more than one that happens once a year?
2.Your town has 5,000 households; about 1 in 10 own a dog; about half of those might pay for a walking service. Roughly how many reachable paying customers is that?
3.During a customer-discovery interview, what should you avoid doing?
4.Name one sign, based on behavior, that a problem is real enough to build on.
5.You have two signals: ten people say they 'would definitely buy,' and one person already pays a rival $40 a month for a worse version. Which do you trust more?
6.At $240 per customer per year, what is the yearly revenue from 300 reachable customers, in dollars?
7.In one sentence, why do you ask about the past instead of the future?
8.What should a minimum viable product be built to test?
9.You want to test a meal-delivery idea cheaply. Which is a concierge test?