Test a New Service Before It Costs a Lot: Pilot Customers and Pre-Sales

Why new services often fail too late

A new service usually begins with a good feeling. A customer asked for it, a competitor offers it, or the owner sees a gap. Then money goes in: training, tools, a dedicated page on the website. Only after months does it become clear whether enough customers buy and whether the price covers the effort. If not, the money is spent and the time used up.

This piece asks: how does a small business check a new service as early and as cheaply as possible for whether it will hold up? To answer it, we first clarify the terms and a mental model from research on entrepreneurial decision-making, then what studies show, weigh two views, and finally set out a test process with pilot customers and pre-sales, plus a rule for when to stop.

A test without a prior definition of failure is not a test but confirmation in installments.

Terms: assumption, pilot customer, pre-sale

An assumption is a statement that must be true for the service to succeed but is not yet proven. Typical assumptions are: enough customers have the problem, they will pay a certain price, and the business can deliver the service with a certain effort. The riskiest assumption is the one whose failure would bring down the whole service and about which least is known.

A pilot customer is a customer who receives the service in an early version, knowing that it is still being tried out. In return they usually get an advantage, such as a lower price or more support, and give feedback and allow the business to learn from the order. A pre-sale means that customers place a binding order or pay a deposit before the service is fully built.

Both forms have one thing in common: they ask the customer for something that costs them. That is where their value lies. A customer who says in a conversation that the idea is interesting loses nothing. A customer who signs a pilot contract loses money, time, or standing in their own company if the service is no good.

The mental model: testing assumptions like a scientist

Camuffo, Cordova, Gambardella, and Spina (2020) describe a scientific approach to entrepreneurial decisions. The entrepreneur states explicitly why the idea should work, derives testable assumptions from it, designs tests that can confirm or refute those assumptions, and decides according to the result. The counter-model is the usual approach: searching by feel, trying things out, and above all looking for confirmation.

The core of the model is a distinction from statistics. A false positive means considering an idea good when it is not. A false negative means rejecting a good idea. Because most new ideas do not hold up in the end, the first risk is the more common one. A clean test lowers it by making weak ideas visible earlier. For a business this means: the test should above all prevent money from flowing into a service that only looked like a success.

What studies show about systematic testing

Camuffo and colleagues (2020) tested their approach in a randomized experiment with 116 young Italian companies. All received ten training sessions on how to obtain market feedback, for example through customer interviews and simple first versions of the offer. Half additionally learned to formulate assumptions explicitly and test them rigorously. The companies were observed at 16 points over about one year.

The result: the scientifically trained companies achieved better results, switched to a different idea more often, and did not give up more often in the early phase than the control group. A replication with 759 companies in four experiments (Camuffo, Gambardella, Messinese, and others 2024) confirmed one finding in particular: trained founders ended weak ideas more often. They changed direction a few times, not never and not constantly.

Both studies concern start-ups, not established businesses trying out an additional service. The transfer is nevertheless plausible, because the situation is similar: a new service is a small start-up within the business, with an unknown market. The article why young companies fail describes why risk rises the more is new at once, and recommends reducing newness, for example through a pilot customer instead of a new market.

For business customers, a qualitative study by Alam (2021) adds to the picture. He followed several development projects for new services at a large financial services firm in India and interviewed, among others, 17 responsible managers. The study builds on the research finding that interaction with customers during development is key to the success of new services, and examines how personal conversations and social media work together for this. It is a case analysis and does not measure success in numbers.

The evidence ladder: how much each kind of feedback counts

Not all feedback proves the same amount. The evidence ladder is the tool of this guide. It ranks signals by what they cost the customer. The higher the rung, the less likely a false positive.

1. Rung 1: Agreement in conversation

   The customer likes the idea. It costs them nothing, so it proves little. Politeness and genuine interest are hard to tell apart.

2. Rung 2: Time invested

   The customer takes an hour, shows you their processes, or brings in colleagues. A first real signal.

3. Rung 3: Written intention

   A letter of intent or a quote the customer submits internally for approval. They put their reputation in their own company on the line.

4. Rung 4: Paid pilot order

   The customer pays, even if with a pilot advantage. Now demand and willingness to pay are proven, at least for them.

5. Rung 5: Follow-up order or pre-sale at full price

   The customer renews, reorders, or pays the regular price in advance. This is the strongest evidence before market entry.

Two views: test thoroughly or offer quickly

Offering quickly is supported by the fact that a small business has no research department. Testing every assumption separately loses months in which a competitor is already selling the service. Many successful services came about because an owner simply started and learned with the first customers. The analysis effectuation describes how experienced entrepreneurs under uncertainty start with the means at hand and commit only as much as they are prepared to lose.

Thorough testing is supported by the finding of Camuffo and colleagues: the structured approach did not make the companies slower in the sense of more tasks, but more precise. They ended weak ideas earlier and directed their resources to better ones. A pilot order is also not a laboratory experiment but already revenue.

I do not see a real contradiction here. The pilot with a few paying customers is quick offering and test at the same time. The difference from simply starting lies in two things: the written assumption recorded beforehand, and the criterion that determines when to stop. Both take an hour and, in the unfavorable case, save a year.

The pilot in six steps

1. Write down assumptions

   List five assumptions about demand, price, effort, repeat purchase, and capacity. Mark the riskiest.

2. Set criteria

   Write down beforehand at which result you continue, adjust, or stop, for example: at least three of ten customers approached sign a paid pilot, and effort is at most 20 percent above plan.

3. Select pilot customers

   Choose existing customers who have the problem, give open feedback, and are not among your most important. How to rank customers by importance is shown in the guide ABC customer analysis.

4. Sign a pilot contract

   Scope, duration, price, pilot advantage, obligation to give feedback, and permission to use the result as a reference. No free pilot: those who pay nothing prove nothing.

5. Measure, do not just ask

   Record hours, materials, questions, and problems for each pilot order. Post-costing shows whether the planned contribution margin holds.

6. Decide and record

   Compare the results with the criteria set beforehand and make one of the three decisions. Write down what you learned, even if you stop.

The pilot advantage should come with something in return, such as feedback, a reference, or willingness to show processes. This matches the rule from managing discounts of giving no discount without something in return. A pilot at half price also only proves willingness to pay half the price. Better is the full price with a different advantage, such as a right to withdraw or additional support.

Transfer and limits

A pilot with a few customers shows whether the service works for those customers. It does not show how large the market is overall. That question is answered by a separate calculation, as described in the guide estimating market size. Pilot customers are also often the most open-minded customers. What works with them may go more slowly with more cautious customers.

After a successful pilot, the service belongs in the regular review of the offering. The guide pruning your service portfolio recommends adding new services only with a review date after twelve months. For services that recur as contracts, building recurring revenue describes how to cost the first contracts.

Conclusion: test early and cheaply

How does a small business check a new service early and cheaply for whether it will hold up? It writes down the assumptions, tests the riskiest one first, decides beforehand when it will stop, and looks for evidence that costs the customer something: paid pilot orders and pre-sales instead of friendly agreement. Research on scientific approaches in start-ups shows that this approach does not slow things down but helps end weak ideas earlier and direct resources to viable ones.

1. This week: Write down five assumptions for your next planned service and mark the riskiest.
2. Next week: Set success and stop criteria in writing and approach ten suitable existing customers.
3. In three months: Evaluate the pilot orders against the criteria and decide: continue, adjust, or stop.

A service that fails in the pilot has cost little. The expensive service is the one that was never tested and therefore is never allowed to stop.