What to do when you have an idea but don't know what the next step is

Having a business idea and not knowing what to do next is far more common than it appears. This is precisely where many people remain stuck for weeks or months, not because they lack ability or motivation, but because an early idea allows too many possible directions.

You could research the market, speak to potential customers, define the business model, design a landing page, choose a name, look for co-founders, create a prototype or begin publishing content. Every one of these actions may make sense at some point. The problem is that, until you clarify what you need to learn, none of them has an obvious priority.

It is therefore easy to confuse movement with progress. You work on a name, compare tools, register a domain, prepare a presentation and describe the features the product will include. By the end of the week, you have produced many things but still do not know whether the problem matters enough, who wants to solve it or why they would choose your proposition.

When you do not know what to do next, you rarely lack work. You lack a structure for reducing uncertainty in the right order.

The first step is not building: it is defining what you need to learn

After having an idea, the first task should not be to turn it immediately into a product. It should be to understand which observation led to the idea, who it affects and which part of your interpretation remains an assumption.

This does not mean spending months on research before taking action. It means formulating a question specific enough for the next action to help you answer it.

If you still do not know who experiences the problem, creating a landing page is probably premature. If you know who experiences it but not whether they consider it a priority, interviews may provide more information. If the problem and segment are relatively clear but the proposition remains uncertain, a demonstration or manually delivered service may teach you more than another abstract conversation.

The appropriate next step depends on the current uncertainty. This is why copying another startup’s sequence rarely works: that company may already have known something that you still need to establish.

An idea is not yet a business hypothesis

An idea usually appears as a mixture of observation, intuition and solution. You may have noticed an inefficient process, experienced a frustration personally, discovered a technology capable of improving a task or imagined an experience that does not yet exist.

Any of these can provide a useful starting point, but they do not yet offer enough clarity to decide what to build. You must first separate several elements:

• The observation: what you have genuinely seen or experienced.
• The interpretation: why you believe it happens and which consequences it produces.
• The segment: who might experience the problem most intensely.
• The current alternative: what that person or company does today.
• The assumption: what would need to be true for an opportunity to exist.
• The imagined solution: how you think the problem could be solved, without treating it as a conclusion.

This separation prevents a common mistake: presenting something as a fact when it remains only one possible explanation. You may have observed that several people use spreadsheets to manage a process. That does not prove they want to replace them, that the process is important enough or that a separate tool would be more convenient.

The idea becomes manageable when you can identify what you know, what you are interpreting and what you need to test.

Write down the starting point without trying to sell it

Before researching or interviewing, describe the idea in a simple sentence. You do not need a perfect value proposition or commercial message. You need a formulation that exposes the main assumption clearly enough.

You can use a structure such as:

I believe [type of person or company] struggles to [task or outcome] when [context], and that the current alternatives create [cost or consequence].

For example:

• I believe freelancers managing several projects simultaneously struggle to coordinate deliveries, communication and payments because they use separate tools and lose information between them.
• I believe international students arriving in a city for the first time struggle to compare reliable accommodation before understanding its neighbourhoods and local conditions.
• I believe small service businesses lose repeat business because they do not know when or how to contact previous customers again.

These statements do not demonstrate that a business exists. They allow the idea to stop being a cloud of possibilities and become something that can be researched, discussed and contradicted.

Avoid placing the solution inside the initial formulation. “I want to create an AI application for freelancers” describes what you would like to build but not the problem that should justify it. If you begin with the solution, subsequent research may become an exercise in finding reasons to preserve it.

Narrow down who might need it first

Many ideas initially appear to serve “everybody”, “any small business”, “entrepreneurs” or “people who shop online”. This breadth can seem like an advantage, but it makes validation difficult.

Two people may recognise the same problem while experiencing it at entirely different levels of intensity. A self-employed person with one client does not manage work in the same way as a freelancer handling twelve active projects. A small company completing an occasional task will not assess a tool in the same way as one that repeats the process every week.

The initial objective is not to define the total market forever. It is to choose a group specific enough to learn from. Look for people who share a context, behaviour and recognisable consequence.

Instead of “entrepreneurs”, you might begin with first-time founders who are still at the idea stage. Instead of “small businesses”, you could examine small professional firms that prepare reports manually every week. Instead of “students”, you might focus on people moving to another country without a local network.

The more specific the initial segment, the easier it becomes to ask relevant questions, compare responses and identify patterns. You can expand or change the market later. At the beginning, you need enough definition to avoid mixing problems that only look similar from a distance.

Identify the uncertainty most capable of changing the idea

An early idea contains many questions: whether the problem exists, who experiences it, how frequently it occurs, which alternatives customers use, who makes the purchasing decision, how much they might pay, how you could reach them or whether the solution is technically feasible.

You do not need to answer every question at the same time. You need to identify the one most likely to change the project.

You can prioritise each uncertainty according to three criteria:

• Impact: how much the idea would change if the assumption were false.
• Available evidence: what you genuinely know and where the information came from.
• Cost of learning: the effort required to obtain a more reliable signal.

If you do not know whether the problem matters, discussing price is still secondary. If the problem is well documented but you do not know who authorises the purchase, that question may deserve priority. If the proposition depends on an uncertain technical capability, you may need a feasibility test before promising the outcome.

The first test should address a high-impact uncertainty for which you still have little evidence. This choice turns “I do not know what to do” into an operational question: “I need to learn whether this segment considers the problem important enough to seek a solution.”

Research the context before asking customers for their time

Speaking to real people is essential, but you do not need to ask every question during an interview. Research the available solutions, the language used to describe the problem, the tools adopted by the segment, changes affecting the industry and any public information that can help you understand the context.

This work prevents you from spending a conversation on questions you could have answered independently. It also helps you recognise what the interviewee is describing and explore the details that do not appear in reports or product pages.

Artificial intelligence can help organise this stage, locate categories of competitors, suggest search terms and turn an initial description into a list of assumptions. You should still verify any information that will influence an important decision. A fluent answer does not guarantee that a figure is current, a competitor remains active or a conclusion genuinely represents the segment.

The purpose of research is not to complete validation from your desk. It is to enter the market with better questions.

Speak to real people to understand their behaviour

Once you have an initial hypothesis and a specific segment, you can conduct useful conversations. The objective is not to present the idea and request approval, but to understand how the problem appears within a real situation.

Ask about previous experiences: when it last occurred, what triggered the situation, what the person did, which tools they used, how much time or money it consumed and what happened when they failed to resolve it. Answers about facts tend to provide more information than predictions about what someone might do when presented with a hypothetical solution.

Questions such as “would you use a platform that simplified this process?” invite a polite answer with no consequences. “How did you solve this problem the last time it occurred?” requires the person to describe a behaviour. This difference matters because, as we explain in validation is not asking whether people like your idea.

You also need to speak to people who genuinely belong to the segment. Friends, colleagues and general contacts may help improve an explanation, but their opinions do not replace the experience of someone who lives within the context, has the problem and could make a decision related to it.

Look for patterns and costs, not one enthusiastic comment

A positive conversation may be encouraging, but it does not validate the idea. One negative response does not invalidate it either. The first interviews help you identify patterns and understand differences between profiles.

Observe whether several people describe the problem without being prompted, whether they have attempted to solve it, which alternatives they use and which costs they already accept. The cost may appear as money, time, errors, manual work, delays, risk or lost opportunities.

You must also look for contrary signals. Everybody may recognise the inconvenience while nobody regards it as a priority. The problem may be frequent but already solved well enough by a tool included in another system. The person experiencing it may not have the authority to purchase a solution.

Contradictions do not damage validation. They are precisely what allows you to refine the segment, modify the hypothesis or discover that the opportunity differs from what you originally imagined.

Turn what you learn into a small experiment

After the first conversations, you do not need to build the complete product immediately. Choose a test proportionate to the next uncertainty.

• If you do not know whether the message is understood, test several formulations with people from the segment.
• If you need to measure initial interest, a landing page directed at a specific audience may provide a signal.
• If you want to explore willingness to pay, present an offer with a real price and conditions.
• If you need to establish whether the outcome creates value, deliver the service manually to a small number of customers.
• If the question concerns usage, create a prototype that allows you to observe how people complete a task.
• If the uncertainty is technical, build a limited feasibility test rather than the full product.

The experiment should state which hypothesis it tests, which behaviour you will observe and which decision will follow from the result. “Create a landing page” is not an objective by itself. It may test whether a message attracts qualified enquiries, but it will not automatically demonstrate that the product retains customers or that the business can become profitable.

Before running the test, define which result would represent a favourable signal, which would weaken the hypothesis and which would make the experiment inconclusive. This reduces the temptation to interpret every number as a victory.

When it makes sense to build an MVP

“Do not build yet” does not mean remaining indefinitely in interviews and research. It means that building should answer a question that can no longer be resolved through a simpler method.

An MVP begins to make sense when you have sufficient evidence about the problem and segment, understand the current alternatives and need to observe the use of a solution in order to keep learning. You should also define which behaviour the MVP is intended to measure.

If you can test the value through a manual service, demonstration or prototype, you may not yet need to develop software. If the hypothesis depends on people using a feature repeatedly, integrating the solution into their process or achieving a result independently, you probably need to build something.

The minimum viable product should not be a smaller version of everything you can imagine. It should be the smallest solution capable of producing the evidence required for the next decision.

Tasks that often create a false sense of progress

Developing the brand before understanding the proposition

Choosing a name, designing a logo and registering profiles feels concrete and satisfying. If you still do not know which problem you will solve or for whom, however, the brand is being constructed around a definition that will probably change.

You do not have to wait until the company is mature, but you should prevent visual identity from replacing validation work.

Building to relieve uncertainty

Building creates a sense of control because it produces something visible. The problem is that it can turn a market question into months of technical work without answering it.

If the main uncertainty is whether anybody considers the problem important, adding features does not reduce it. It merely increases the investment made before receiving the answer.

Asking only people close to you

People close to you generally want to help and may offer useful ideas. They also have stronger incentives to be polite and may not share the customer’s context.

Validation requires conversations with people who can describe genuine behaviours, constraints and decisions, not merely assess whether the proposition sounds attractive.

Generating a complete plan before obtaining evidence

Artificial intelligence tools can produce a buyer persona, Business Model Canvas, content strategy and financial projections in one afternoon. These documents may support thinking, but they do not provide four sources of evidence when all of them derive from the same initial assumptions.

A project is not better validated because it has more documents. It is better prepared when those documents clearly distinguish assumptions, evidence and pending decisions.

Opening every question simultaneously

Attempting to define the product, price, channel, brand, acquisition strategy and technology at the same time creates a large volume of work that can invalidate itself. A change of segment may alter the proposition, price, sales process and required features.

Sequence matters. Begin with the uncertainties capable of changing the rest of the project.

A simple plan for the first seven days

You do not need to resolve the complete business during the first week. Use those days to turn a vague idea into an initial learning cycle:

1. Describe the origin of the idea: what you observed, experienced or discovered and why it appeared relevant.
2. Separate facts and assumptions: mark which information has been verified and which statements remain interpretations.
3. Choose an initial segment: identify who may experience the problem most frequently or intensely.
4. Write the main hypothesis: describe the problem, context and consequence you expect to find.
5. Research the alternatives: review what people currently do and which solutions already exist.
6. Prepare the conversations: design questions about previous experiences and behaviour.
7. Arrange the first interviews: contact specific profiles and record what you learn without attempting to sell yet.

By the end of the week, you may not have a product, brand or definitive answer. You should have something more useful: a better-formulated hypothesis, a more specific segment, several identified assumptions and a next experiment based on what you learned.

What to do when you have several ideas and none stands out

When several ideas are available, choosing only the most original or exciting one may lead you towards the hardest option to test. During the early stage, compare both the opportunity and your ability to learn about it.

You can assess each idea according to:

• How clearly you can describe the problem.
• Your access to people within the segment.
• The apparent frequency and severity of the need.
• Existing behaviours or costs that already demonstrate interest.
• The ease of conducting a small and realistic test.
• Your knowledge or particular advantage within the context.
• The cost of building what is needed to obtain the next evidence.

The idea with the largest theoretical market does not always deserve to come first. It can be more useful to begin with the one that allows you to obtain reliable evidence sooner. An idea that is easier to investigate, test and modify can produce learning at a lower cost than an apparently brilliant opportunity you cannot access.

How to recognise progress before launching

During this stage, progress is not measured only in published pages, developed code or followers acquired. Reducing an important uncertainty is also progress.

You are moving forward when you can describe the problem more precisely, discard a segment that originally appeared promising, discover that the buyer is not the person you imagined or identify an alternative that changes your proposition. Even a negative signal may save several months of development.

You should also notice that the questions become more specific. You move from “is this a good idea?” to “does this type of business experience the problem every week and already dedicate resources to solving it?” The second question allows you to design a test; the first only invites an opinion.

The objective is not to achieve complete certainty before acting. It is to obtain enough clarity for each new action to have a purpose and generate useful information.

From a vague idea to a sequence of decisions

Foundeia treats the beginning of a business as a connected sequence of decisions. The definition of the problem affects the segment; the segment changes the proposition; the proposition determines what needs to be validated; and the evidence should influence the product, price and business model.

When these decisions are made in isolation, it is easy to build a plan full of answers that do not fit together. You can select a customer with a limited budget, design an expensive sales process and propose a solution requiring complex implementation. Each decision may appear reasonable separately and become unviable when combined.

Foundeia therefore uses methodology and artificial intelligence to preserve context, formulate hypotheses, identify inconsistencies and convert information into next steps. The tool supports thinking and organisation; the market supplies the evidence required for decisions.

The purpose is not to accelerate every possible activity. It is to identify which decision belongs now and prevent enthusiasm for building, or the sheer volume of tasks, from hiding the uncertainty that is actually directing the project.

Clarity needs to be built before the product

Having an idea and not knowing what to do next does not mean you are lost. It means that you have not yet turned an intuition into a sequence of questions that can be answered.

Begin by describing the problem, choosing a segment and separating what you know from what you assume. Then identify the uncertainty most capable of changing the idea, research the context, speak to real people and design a small test that produces observable behaviour.

You do not need to decide at the beginning what the complete company will become. You need to establish what must be true for the idea to deserve further investment and find the least expensive way to test it.

The correct next step is not necessarily the one that generates the most visible work. It is the one that reduces the most important uncertainty and allows you to make a better decision than the previous one.

Frequently asked questions