Chapter 2 · Part 1: Foundations

Finding Problems Worth Solving

“People don’t want to buy a quarter-inch drill. They want a quarter-inch hole!”

— Theodore Levitt, quoting Leo McGinneva

The scheduling problem that wasn’t

Brindlepath was founded to fix scheduling. Its founders had watched patients leave the hospital with a stack of follow-up appointments they could not keep: the cardiology visit set for a Tuesday when the patient’s daughter, the only driver in the family, worked; the physical therapy slot across town with no bus route. Readmissions followed. The founders built a clever tool that let care managers book follow-ups around a patient’s real-life constraints.

Hospitals liked the demo. Very few bought it.

When Maya joined, she spent her first three weeks shadowing care managers at two customer sites and one hospital that had said no. What she found was that scheduling was a symptom. The deeper problem was that after discharge, nobody clearly owned the patient. The inpatient care manager’s job ended at the door. The primary care office did not know the patient had been admitted. The specialist had a referral but no context. A well-scheduled appointment did not help if nobody noticed the patient had missed it.

The care managers did not describe it that way. They described a dozen small frustrations: phone tag, faxes that went nowhere, discharge summaries that arrived after the follow-up visit, patients who “fell off the radar.” Maya’s job was to see the shape underneath the complaints. Once she did, Brindlepath’s product strategy changed from better scheduling to clear ownership of every patient through every handoff, and its sales conversations changed from feature demos to a problem every chief nursing officer recognized.

This chapter is about finding that shape, before you spend a year building the wrong thing.

Why products fail

Products rarely fail because the team could not build them. They fail because the team built something the market did not value enough, at a price or cost that did not work, at a time when nobody was ready.

CB Insights, which has studied startup failures for over a decade, analyzed 431 venture-backed companies that shut down from 2023 onward. Running out of capital was nearly universal, cited in 70 percent of cases, but CB Insights is careful to note that it is usually the final cause of death, not the root one.1 Underneath it sat the familiar culprits: poor product-market fit, bad timing, and unit economics that never worked.

The lesson for product leaders inside established companies is the same. A new product line or major feature that fails rarely fails on engineering. It fails on the question this chapter is about: was this a problem worth solving, for people willing and able to pay for a solution?

Start with a problem statement

Every product effort should begin with a written problem statement. Not a solution statement dressed up as a problem (“Users need a dashboard”) and not a vision (“Transform care coordination”). A problem statement names who is affected, what they are trying to do, what gets in the way, and what it costs them.

A useful template:

[Who] need to [goal or job], but [obstacle], which leads to [consequence, ideally measurable]. Today they [current workaround], which falls short because [gap].

Here is Brindlepath’s original problem statement, and the one Maya wrote after her field work:

Before: “Care managers need a better way to schedule post-discharge appointments.”

After: “Outpatient care navigators at health systems need to know, within 24 hours, which recently discharged patients are now theirs and what those patients need. But ownership after discharge is unclear, and information arrives by fax, phone and portal days late. As a result, a meaningful share of high-risk patients miss their first follow-up visit, which drives avoidable readmissions that the hospital is penalized for. Today navigators work from spreadsheets and daily phone calls with inpatient staff, which do not scale past a few dozen patients and break whenever someone is out.”

The second version is longer, and much more useful. It tells the team who the user is (the outpatient navigator, not the inpatient care manager), what success looks like (knowing within 24 hours), what the business case is (readmission penalties), and what the real competitor is (a spreadsheet and the telephone). Every one of those facts will shape the product.

A good problem statement passes four tests:

  • It contains no solution. If it names a feature, rewrite it.
  • It names a specific person, not “users” or “the market.”
  • It describes a consequence someone cares about, in their terms: time, money, risk, outcomes.
  • It can be wrong. A problem statement is a hypothesis. If no evidence could disprove it, it is too vague to be useful.

Jobs, not features

The most durable way to understand a problem is to ask what job the customer is hiring a product to do. The jobs-to-be-done idea, popularized by Clayton Christensen, holds that people do not buy products for their features; they “hire” them to make progress in a particular circumstance.2 Levitt’s quarter-inch hole is the same idea, decades earlier.

Jobs have three layers:

  • Functional: the practical task. Know which discharged patients are mine today.
  • Emotional: how the person wants to feel. Confident that nobody is falling through the cracks on my watch.
  • Social: how they want to be seen. The navigator the physicians trust to have it handled.

Products that serve only the functional job are easy to copy and easy to replace. Products that serve all three become part of how someone sees their own work. Brindlepath’s navigators did not just want a list; they wanted to go home at night without the low-grade dread that a patient was in trouble and nobody knew.

When you interview customers, listen for the job behind the request. When someone asks for a feature, ask what they would do with it, and then what that would let them do, until you reach something they genuinely care about. That is usually the job.

Who has the problem, who pays, who decides

In consumer software, the person with the problem, the person who chooses the product and the person who pays are often the same. In business software they almost never are, and in healthcare the gap is extreme.

For a single Brindlepath deal, the cast typically includes:

  • The users: outpatient navigators and care managers, who live in the product every day.
  • The clinical champion: often a director of care management or a chief nursing officer, who feels the readmission problem and sponsors the purchase.
  • The economic buyer: a chief financial officer or chief operating officer, who signs off on the money and wants a return.
  • The technical gatekeepers: IT and information security, who must approve integration with the electronic health record and the handling of patient data.
  • The influencers: physicians, quality leaders and compliance officers, who can slow or stop a deal without ever being in the room.
  • The beneficiary who never logs in: the patient.

Each of these people has a different version of the problem. The navigator cares about her day. The chief nursing officer cares about readmission rates and staff burnout. The CFO cares about penalties avoided and headcount not added. IT cares about integration effort and security risk. A product strategy that speaks only to the user will win enthusiastic pilots and lose the purchase.

Map your buying group explicitly. For each role, write down what they need to believe to say yes, what would make them say no, and what evidence would convince them. You will use this map again for positioning in Chapter 4, for pricing in Chapter 6 and for go-to-market in Chapter 13.

How big is the problem?

Not every real problem is worth solving. You need some way to compare problems before committing a team to one. I use three questions:

  1. How often does it happen? A problem that occurs every day for every user beats one that occurs once a quarter.
  2. How much does it hurt? Measure in the customer’s currency: hours lost, money spent, risk carried, outcomes harmed.
  3. What are they doing about it now? If people have built elaborate workarounds, or are already paying for something imperfect, the problem is real and the budget exists. If they shrug, it may be real but not urgent.

Plotting candidate problems by frequency and severity gives you a quick, honest picture of where to focus. Figure 2.3 shows the exercise Maya ran with her team after the field work. The positions are from that exercise, using the team’s own scoring; they are illustrative, not survey data.

The original scheduling product addressed a real but occasional pain. The ownership problem was frequent, severe and expensive. The map made the pivot obvious in a way no argument had.

Sizing the market

Once you have a problem worth solving, estimate how many people have it and what they would pay. Investors and executives will ask for three numbers:

  • Total addressable market (TAM): everyone who has the problem, if you won every one of them.
  • Serviceable addressable market (SAM): the part you can realistically reach with your product, channels and geography.
  • Serviceable obtainable market (SOM): the share you can plausibly win in the next few years.

Build these numbers from the bottom up, not from analyst reports. Top-down estimates (“the care-coordination market is $X billion; if we get 1 percent…”) are easy to produce and impossible to act on. A bottom-up estimate starts with countable units and a price.

For Brindlepath, the bottom-up version looked like this: there are a few hundred health systems in the United States with enough post-discharge volume and readmission exposure to care; each employs a number of outpatient navigators and care managers; Brindlepath prices per navigator seat with a platform fee. Multiplying the number of target systems by a realistic number of seats and a realistic price gives a market estimate the sales team can build a plan around and the finance team can check. The exact figures matter less than the method: every number in the chain is something you can verify by talking to customers.

Know the real competition

When teams list competitors, they usually list companies that sell similar software. That is the smallest part of the competition. The biggest competitor for most new products is the status quo: the spreadsheet, the phone call, the weekly meeting, the person who “just knows.”

For Brindlepath, the competitive set included:

  • Direct competitors: other care-coordination platforms.
  • The incumbent platform: the electronic health record vendor, which offered basic care-management modules that were “good enough” for some buyers and already paid for.
  • The status quo: spreadsheets, shared inboxes and daily huddle calls.
  • Doing nothing: accepting the readmission penalties as a cost of doing business.

Each requires a different answer. Against direct competitors, you win on fit and outcomes. Against the incumbent platform, you win on depth and speed. Against the status quo, you win by making the cost of the workaround visible and the switch painless. Against doing nothing, you win only if the problem is urgent enough that someone’s job depends on fixing it.

When you study competitors, study their customers more than their features. Read their reviews, talk to people who chose them and people who left them, and find out what job they were hired for and where they fall short.

Constraints are part of the problem

Every market has rules that shape what a solution can be. In healthcare, they include privacy law such as HIPAA in the United States, the need to integrate with electronic health records through standards like HL7 and FHIR, clinical safety expectations and, for some software, medical device regulation. Fintech has its licensing and anti-money-laundering rules; education has student-privacy laws; most enterprise software faces security reviews and procurement requirements.

Treat these constraints as part of the problem definition, not as obstacles discovered at the end. Ask early:

  • What data will the product touch, and what rules govern it?
  • What systems must it connect to, and how open are they?
  • Who has to approve it before a customer can buy or deploy it?
  • Could any part of it be regulated as more than ordinary software?

The answers change your cost, your timeline and sometimes your entire approach. A product that needs deep integration with five different electronic health records is a different business from one that works on a standalone basis. Better to know before you build. Chapter 11 covers trust and compliance in depth.

Brindlepath in practice: sizing the problem

When Maya presented the new problem statement to the board, the first question was the one every board asks: how big is this? She answered with a bottom-up estimate, built with Priya Nair, the CFO, and checked against conversations with a dozen health systems.

The chain of reasoning was simple. The team counted the health systems in Brindlepath’s target regions with enough post-discharge volume and readmission penalties to feel the problem, and found a few hundred. Field interviews suggested that a typical system in that group employed somewhere between twenty and sixty outpatient navigators and care managers who would use the product. At Brindlepath’s planned price of $1,800 per seat plus a platform fee, a typical customer would be worth roughly $70,000 to $150,000 a year. Multiplying the number of reachable systems by a conservative share and a typical contract value gave a serviceable market in the tens of millions of dollars of annual revenue, before any expansion into physician groups or health plans.

The number itself mattered less than the conversation it enabled. Because every link in the chain was something a board member could question and a salesperson could verify, the debate moved from “is this big enough?” to “which assumptions are weakest?” The weakest turned out to be the number of navigators per customer, which varied widely. So the team made it a discovery question for the next round of interviews, and refined the estimate a quarter later.

The same exercise exposed something uncomfortable about the old scheduling product. Fewer buyers felt that pain strongly, the budget for it was smaller, and the record vendor’s module already did “good enough” scheduling for many. The bottom-up math, as much as the field work, made the case for the new strategy.

Common mistakes

Falling in love with a solution. Teams often start with an idea and then look for a problem to justify it. The tell is a problem statement that quietly describes the product. Brindlepath’s original scheduling tool was a solution in search of the real problem.

Asking customers what they want. Customers are experts in their problems and unreliable designers of solutions. “Would you use this?” invites a polite yes. “Tell me about the last time this happened” invites the truth.

Talking only to the people who like you. Existing champions confirm what you believe. Prospects who said no, and customers who left, tell you what you are missing.

Ignoring the buyer. A product that delights users but cannot win over the economic buyer, IT and compliance will collect enthusiastic pilots and few contracts.

Sizing the market from the top down. “One percent of a ten-billion-dollar market” convinces nobody who has sold enterprise software. Build the estimate from countable customers and a realistic price.

Discovering constraints last. Regulation, integration and procurement requirements discovered after the product is built can double its cost or delay it by a year.

Try this

Write the problem statement for the most important thing your team is working on, using the template in this chapter. Then show it to three people: a customer, a colleague in sales or customer success, and an engineer. Ask each of them one question: “What is wrong with this?” Rewrite it after each conversation. If the statement changes substantially, you have learned something important before spending another week building. Finally, add the buying group: list each role, what they need to believe, and the evidence that would convince them.

How AI changes this

AI is genuinely useful in problem discovery, as long as you use it for breadth and keep humans on depth.

Use AI to scan wide. An assistant can read hundreds of support tickets, app-store reviews, sales-call transcripts, forum threads and competitor release notes, and cluster them into candidate problems. Work that once took weeks now takes an afternoon. Treat the clusters as leads, not conclusions.

Use AI to pressure-test your thinking. Ask a model to argue against your problem statement, list the stakeholders you have missed, or role-play a skeptical CFO. It will not always be right, but it is a fast, tireless sparring partner.

Use AI to draft the market model. Give it your bottom-up assumptions and ask it to build the spreadsheet and the sensitivity analysis. Then check every assumption yourself.

Do not use AI as a substitute for the field. A model trained on the internet knows what has been written about a problem, which is the average, often outdated view. The insight that changed Brindlepath’s strategy came from standing behind a navigator at 4:30 p.m. and watching her call three offices to find out who owned a patient. No dataset contained that. The most valuable discovery still happens in person, and it is the part of the job that AI makes more important, not less, because everyone else now has the same desk research you do.

Watch for synthetic certainty. AI summaries of customer feedback tend to smooth over disagreement and produce tidy themes. Real problems are messy. When a cluster looks too neat, read the raw material.

Chapter checklist