How to combine bottom-up and top-down problem-solution finding to create genius level new companies

Two ways to solve a problem

You can solve problems from two different angles:

Bottom-up

Start with novel technical concepts, then brainstorm what kind of value they can generate and finally try to find a market that wants those values.

Example: “Wow pre-and-post-training works like this. I wonder what would happen if every company had their own pre-trained model on their own data?”. Then investigate what kind of value (online research, AI brainstorming, talk to people you know who works at companies you suspect might benefit from this tech) companies could get from that to find some sort of need it would solve.

Top-down

Find a problem that many companies have, then come up with a solution that can be deployed across all of them.

Learn tech to enable top-down problem-solving

Some people are incredible with technology but never find a market. They start with solutions no one asked for, and then wonder why customers don’t understand them. That’s the bottom-up trap.

Others talk to customers all day but can only offer generalized advice or copy existing solutions. They have no unique approach because they don’t understand what the tech can actually do. That is the opposite trap.

The right approach is a combination:

Use top-down to find problems. Use bottom-up to solve them.

Problems without understanding the tech lead to generic offerings. Tech without real problems leads to “cool but useless” inventions. You have to do both. That’s what separates founders from cloners, talkers, and inventors:

• Cloners copy things that make money but don’t shift markets.
• Talkers impress people but rarely produce anything valuable.
• Inventors create interesting things no one needs.

Top-down to identify valuable problems. Bottom-up to build novel ways to solve them. That’s where real value is created. The rare people who can do both are the ones who build category-defining companies.

What that means in practice

Problem-digging

Average founders focus on what customers say. Great founders understand the actual job the customer is trying to get done and aim directly at that. They ignore the literal request and go after the real purpose behind it.

Solution-finding

This is the ability to understand how something works at a fundamental level. Not memorizing a manual, but understanding why it’s that way and predicting what will happen if you change it.

How to find a problem

You find problems through some version of listening, observing, and digging beneath the surface. What customers say is rarely the real problem. You must understand:

• What they are trying to achieve
• Why they need it
• What prevents them from achieving it
• The context around those constraints

Practical ways to do this:

• Interview companies until you hear a problem repeatedly. Then dig deeper to understand the pattern behind it.
• Do online research to discover problems.
  • Pick industries or roles you like, and find where those people discuss work and frustrations. (If you don’t know where that is, ask ChatGPT.)
  • Search job posts on Upwork. When you see a recurring task, dig into similar posts to understand the underlying context and why it keeps appearing.*

* If you think about it, the meta-problem here is “How do I find a business idea?” I’ve already proposed several approaches. I could almost build a product for this. I.e. an AI that helps people find business ideas by doing the online research steps I’ve described above.

A problem is worth solving when it is painful, recurring, and important enough that people are willing to pay for a solution. A genius-level problem is when you identify the root cause, not the symptom. You aim at the “parent” problem, circumstances creating the problem, root causes, or even the problems the person who hired the person saying the problem has, and the reason they hired the person.

The solution criteria

If you want to build a scalable company, the solution must:

• Apply across different contexts, or
• Target companies similar enough that the same narrow solution makes sense, or
• Be customizable by the customer without requiring more effort than the value it generates

Example: CRM systems solve general problems like tracking customers and forecasting sales, but can be used in many different contexts. The issue is that they often require heavy customization before they create real value, which leads many companies to not use them properly. (I owned a CRM company. This is the core problem with traditional CRMs. Let’s see if the new AI CRM category finally solves this.)

If you find a general problem paired with a general-fit solution, you have the entry point for a scalable business.

What an agency or consultancy is

With this framework, an agency or consultancy is a problem fit without a solution fit. Each solution must be built from scratch for each customer.

Consultancies are companies that have the first (problem fit), but not the second (solution fit). Hence, they must design unique solutions for each customer.

But doing unique solutions for a particular customer can often generate much more value than a generic solution – hence consultancies can scale too. The first requirement is simply that the value per customer is higher than the cost to build it. The second requirement is that they can train more employees to make the process of implementing custom solutions more effective and reach bigger scale.

What to do now to build a business

Search for problems

• Interview people in industries or roles you like. Ask what they do every day and why. Dig into the why until you understand the real job they are trying to accomplish.
• Join groups where these people hang out and read what they talk about.
• Use Upwork and look for recurring job posts. Patterns emerge quickly if you read enough.

Once you find a problem worth investigating, talk to more people about that same problem. You’re looking for repetition and context.

Generate leads for problem interviews

• Build a landing page. Focus on the problem, not your imagined solution.
• Run ads or reach out directly and ask people if they have this problem.
• Or simply cold call a bunch of people and ask if they have this problem – then interview them.

When you start seeing patterns, you can begin exploring the solution.

Explore the solution

If you are technical, you may already see how the solution works. Test core assumptions privately. If you’re not technical, you need to learn the tech or you will end up being a talker or a copycat.

You should find a solution that can solve the problem across the context(s) that you have discovered in your problem research.

And you need to know that the solution you are thinking of will actually work. You can do this by building mini-prototypes of the specific parts of the solution you are less sure about, just to prove that it will work the way you have imagined.

When you know that the solution can work, you can move onto the next stage.

Validate the problem-solution fit

• Update the landing page so it covers both the problem and your solution concept.
• Run ads or do cold outreach again. This time try to sell the solution!

If people are willing to buy what you propose, you have early signs of product-market fit. Now you can build.

That’s it. That’s the process. Five steps to build a product company. Simple in theory, difficult in practice, but it works.