How to Validate a Startup Idea in 48 Hours Without Writing a Single Line of Code

Pillar: Entrepreneurship • Cluster 1A • Primary keyword: validate startup idea

The Graveyard Nobody Talks About

I have built three products that failed before anyone paid me a dollar. Not because the code was bad. Not because the design was ugly. Because I was solving a problem that either did not exist, or that nobody cared enough to pay to fix.

The first one took me four months. The second took six. The third I killed after two weeks because I had finally learned something: the product is not the risky part. The assumption that people want it is the risky part. This is why the first principle in my AI-Native Founder Playbook is “validate before you build,” not “ship fast and see.”

42% of startups fail because they build products nobody wants. That is the single most cited cause of startup death, beating out running out of money (29%), wrong team (23%), and competition (19%). And here is what makes that number painful to read: product-market fit is not a mystery you discover after building. It is a hypothesis you test before.

The founders who survive do not have better ideas. They have better feedback loops. They talk to people before they write a line of code. They find out whether the problem is real, whether people already spend money solving it badly, and whether they will pay for a better solution.

And they do this fast. Not in three months of “market research.” In 48 hours.

I have run this process across multiple ventures now. The one that worked most recently was EasyClinicGrow, where I talked to 31 clinic owners in six days before I wrote a single line of backend code. For context on what that kind of upfront research unlocks, read about how Medvi scaled to $401M in year one with a one-person team, a result only possible because the founder knew exactly who had the problem and what they would pay. Fourteen of them described the exact same problem in almost the exact same words. That convergence told me more than any market research report would have.

This post is the framework I use. Every step. Every tool. Every signal that tells you whether to kill the idea or push forward.

The Real Problem with How Founders Validate

Most founders do not skip validation. They do fake validation.

Fake validation looks like this: you tell ten friends about the idea. They say “wow, that sounds amazing.” You post in a Facebook group and get 47 likes. You send a survey to your email list and 68% say they would “probably” or “definitely” use your product.

None of that is validation. That is social affirmation. People are polite. People lie on surveys. “Probably use” costs nothing to say and means almost nothing.

Real validation is uncomfortable. It means putting a price in front of someone and watching what they do. It is the reason the AI Wrapper Trap is so common: founders validate that a problem is real, build a thin solution on top of a foundation model, and discover only after launch that customers will not pay because the switching cost is zero. The willingness-to-pay test in this sprint catches that before you build. It means asking someone to give you their calendar for 30 minutes and pay attention to whether they reschedule or cancel. It means building a landing page for a product that does not exist yet and running $50 of ads to see if anyone clicks “Get Early Access.”

Only 40% of startups conduct formal market validation before launch, according to current research. The ones that do take up to three times longer to reach product-market fit than founders expect, mostly because the first version of the idea is almost never right, and you need real feedback to course-correct.

The other thing that kills founders: they conflate problem validation with solution validation. Confirming that a problem is real is not the same as confirming that your specific solution is the right one. I have seen founders who spent a month interviewing customers about a problem, built a product to solve it, and still failed, because the interviews validated the problem but not the solution shape.

The 48-hour sprint I am going to walk you through tests both, sequentially. First the problem. Then the solution concept. Then the willingness to pay. In that order, because each step should be a gate you pass before spending more time on the next one. If you have already built something and are trying to figure out whether you have the right product, the Founder Operating System framework has a section on decision auditing that is worth reading alongside this.

The 48-Hour Validation Sprint Framework

Before the framework, a quick note on who this works for: early-stage founders with an idea they have not built yet, or founders who have built something and are not seeing traction and want to re-validate before pivoting. If you have paying customers already, this is less useful. If you have nothing but an idea and a hunch, this is exactly right.

The sprint has four phases, each roughly 12 hours. You do not need a team. You need a laptop, a phone, and a willingness to talk to strangers about their problems.

Each phase has a gate. If you do not pass the gate, you do not move to the next phase. You either kill the idea, change the positioning, or change the problem you are solving and restart from Phase 1. The gate structure is what separates this from “doing some research and then building anyway.”

Hours 0-12: The Problem Exists Test

This is the phase most founders rush through, because talking to people feels slow compared to building. It is not slow. It is the only part that actually matters.

The goal is simple: confirm that the problem you are trying to solve is real, painful, and currently unsolved (or solved badly).

Who to talk to and how to get them on the phone in under two hours

You need five to eight people who match your target customer profile. Not your friends. Not your family. Not your former colleagues who will be polite about everything. You need people who are living the problem right now.

The fastest ways to get them:

LinkedIn direct message. Search for your target persona by job title and industry. Message 20 people with something like: “Hey [name], I am doing research on [specific problem area] and noticed your background in [industry]. Could I buy you a 20-minute coffee chat to understand how your team handles this? No pitch, just listening.” Expect a 20-30% reply rate. Book five calls in two hours.

Reddit cold outreach. Find the subreddit where your target customer hangs out. Search for threads where people are complaining about the exact problem you want to solve. Message the person directly: “I saw your post about [problem]. I am researching this space. Would you be open to a short call? I will send you a $20 Amazon gift card for your time.” You will get five people booked before lunch.

Existing network, but correctly.. Not “hey does this sound like a good idea?” Instead: “I know you deal with [problem area] at work. I am trying to understand this better. Can I ask you some questions for 15 minutes? I want to learn about your experience, not pitch you anything.”

This problem-first approach is one of the core mechanics in the AI Opportunity Map 2026, which shows that the highest-margin AI products are the ones built on top of a deeply understood problem with a known buyer. One founder I know built a $3M ARR legal tech startup by doing exactly this before writing any code. He messaged 80 solo attorneys on LinkedIn in one afternoon. Nineteen responded. He had twelve calls booked by the next morning. By the end of those twelve calls, he had heard the same exact complaint eleven times: extracting clauses from contracts took hours, was error-prone, and had no good tool for solos who could not afford enterprise software. That convergence was his green light.

What to ask and how to listen

The Mom Test, popularized by Rob Fitzpatrick, remains the best framework here. The core rule: never ask about your idea. Ask about their life.

Questions that work:

• “Walk me through the last time you dealt with [problem area]. What happened step by step?”
• “What did you try to fix it? How did that work?”
• “How much time does this cost you in a typical week?”
• “What would it mean for your work if this problem just went away?”
• “What have you already paid for, trying to solve this?”

Questions that do not work:

• “Would you use a product that did X?” (Everyone says yes.)
• “Do you think this is a big problem?” (Obviously yes, you just described it.)
• “Would you pay for a solution to this?” (Hypothetical spend is meaningless.)

Take notes in real time. After the call, score the problem on three dimensions: pain intensity (1-10), frequency (daily/weekly/monthly), and money already being spent on workarounds.

What passing looks like

You pass Phase 1 if at least three of your five to eight people describe the same problem in similar terms, rate the pain at 7 or above, and are already spending time or money on an imperfect solution.

When 60% of the people you talk to describe the same problem in the same way, you have found your early adopters. That is the convergence signal you are looking for. Do not aim for consensus. Aim for a consistent, specific, painful pattern.

If you talk to eight people and they each describe a completely different problem, or they rate the pain at 4 out of 10, or nobody is currently doing anything to address it, those are kill signals. You either have the wrong target customer, the wrong problem framing, or an idea that is a vitamin rather than a painkiller.

Hours 12-24: The Demand Signal Hunt

Phase 1 tells you the problem is real in your sample. Phase 2 tells you whether the market is real at scale.

The question here is: are strangers already searching for, complaining about, or spending money on solutions to this problem? If they are, you have a market that already understands the problem. If they are not, you may be ahead of a market that does not exist yet, which is a much harder position to be in.

Reddit as a demand detector

Reddit has 121 million daily active users. More importantly for validation purposes, its users are brutally honest. They complain without a filter. They ask “why doesn’t something like this exist?” They post workflow breakdowns and workarounds. It is the best free demand research tool available.

The method: search Reddit for your problem keywords. Search Google using site:reddit.com plus your keywords. Look for threads where people are describing the problem, asking for solutions, or comparing tools.

The signal: if you find 5 or more threads in the past 30 days where people describe the same problem, that is real, organic demand. If you find a thread with 200+ upvotes where someone says “I cannot believe there is no good tool for X,” that is a gift. Screenshot it. It is your validation artifact.

One rule of thumb that has held up across multiple companies I have built and talked to founders about: the same complaint across five threads in the same month is demand. Not interesting. Demand.

Keyword research as a proxy for market size

Open Google Keyword Planner (free) or Semrush’s free tier (five searches per day). Type in the core problem your product solves. Look at three things:

Volume. How many people are searching for this per month? Under 1,000 searches and you have a niche or nascent market. Between 1,000 and 10,000 is a focused market with real demand. Above 10,000 and there is clearly a market, but there are also likely competitors.

Intent. Are the searches informational (“what is X”) or transactional (“best tool for X,” “buy X software”)? Transactional searches mean people are actively looking to spend money. That is what you want.

Trend direction. Open Google Trends for the same keywords. Is the trend line flat, rising, or declining? Rising over the last two years with a spike in the last six months is the best signal. That means you are entering a market as it is heating up, not after it has peaked.

Where else to look

Beyond Reddit and search, three other places are worth 30 minutes each:

Job boards. Search Indeed, LinkedIn Jobs, or similar for job postings that mention your problem domain. If companies are hiring to manually do the thing you want to automate, that is a demand signal and a market sizing tool at once. Twenty job postings a month for “contract review specialist” tells you exactly how big the pain is and how much companies are willing to spend on it.

Product Hunt and G2 reviews. Search for the closest existing solutions. Read the one-star and two-star reviews on G2 or Capterra. Those reviews are your product roadmap. Every complaint is a feature gap. Every “this tool is too expensive for a small team” is pricing intelligence.

Competitor pricing pages. If there are competitors, look at their pricing pages. If they are charging $500/month enterprise contracts, that tells you the market pays for this. If the cheapest plan is $200/month, there may be a gap for a lower-price-point product for smaller buyers.

What passing looks like

You pass Phase 2 if you find organic demand signals: Reddit complaints, search volume above 1,000 monthly searches with transactional intent, or job postings indicating companies are hiring for the problem you are solving.

Kill signal: you search Google, Reddit, and G2 and find nothing. No search volume, no complaints, no job postings. This happens with brand-new categories where the market does not yet understand the problem. That is not automatically a death sentence for the idea, but it means you are doing category creation, which is 10x harder than category entry. Make that choice consciously.

Hours 24-36: The Smoke Test

Phases 1 and 2 were about confirming the problem. Phase 3 is about testing whether strangers will engage with your proposed solution, without you being in the room to sell it to them.

This is the smoke test. The idea: build a minimal representation of your product concept, get real people to see it, and measure their behavior.

The landing page test

You can build a landing page for your not-yet-built product in under two hours. Tools like Carrd (free), Framer (free tier), or Webflow let you create a credible, one-page product description with a call to action.

The page needs only four elements: a headline that names the problem and the benefit, a three-bullet description of what it does, one call to action (“Get Early Access” or “Join the Waitlist”), and no mention of pricing yet.

Do not overthink the design. The Buffer team validated their social media tool with a landing page that had one headline, two sentences of description, and a “Plans and Pricing” button. When you clicked it, the page said the product was still being built and asked for your email. The sign-up rate from that page told them there was real demand before they wrote a single backend function.

Dropbox did something slightly different. Drew Houston recorded a three-minute demo video of a product that did not fully exist yet. He posted it to Digg. Overnight signups went from 5,000 to 75,000. That video cost nothing to make and validated demand at a scale no focus group could have matched.

Build the page. Then drive traffic to it.

How to drive traffic to a landing page in 24 hours with $50

You do not need a big ad budget. You need enough traffic to get a statistically meaningful signal.

For B2C ideas: run $50 in Facebook or Instagram ads to your exact target audience. Target by interest, age, and behavior. Watch the click-through rate and email capture rate, not the “likes.”

For B2B ideas: post to LinkedIn with a genuine description of the problem you are solving. Share it in three relevant LinkedIn groups. Post to the relevant subreddits with a “roast my landing page” framing, which gets more engagement than a pitch and will still drive traffic.

For local ideas: run a Google Ads campaign targeting your city plus your keywords for one day. $50 can generate 100-200 clicks on a local search campaign.

What you are measuring: the click-through rate on your ad (measures how compelling your headline is), the email capture rate (measures how compelling your value proposition is), and whether anyone books a discovery call through the page.

A CTR above 2% on cold traffic is a good signal. An email capture rate above 20% on landing page visitors is excellent. Below 1% CTR is a red flag about the messaging or the audience, not necessarily the idea.

The Wizard of Oz approach for complex products

Some products are hard to represent on a landing page because the value is in the workflow, not the concept. For those, the Wizard of Oz test is more useful.

The idea: simulate the product manually, with the customer not knowing how it works behind the scenes. They think they are using software. You are actually doing it by hand.

Wealthfront’s early version had the team manually generating investment reports that looked system-generated. The customer saw a clean PDF recommendation. A human analyst produced it. This let them validate whether customers would act on the recommendations before they built the algorithm.

For an AI-powered tool, this might mean doing the AI task yourself for the first ten customers. For a data aggregation product, it might mean manually pulling data from five sources into a spreadsheet and presenting it in a clean Google Doc. The customer experience is simulated. The demand signal is real.

What passing looks like

Phase 3 gate: you get 10 or more email sign-ups from cold traffic (strangers you did not already know), or you do a Wizard of Oz with five customers and all five say they want to continue using it.

If you run $50 in ads and get 200 visitors but zero emails, that is useful information. Your headline is wrong, your value proposition is unclear, or your target audience is off. You can iterate on the messaging and rerun. The question is not “did we pass the gate perfectly on the first try” but “are we learning fast enough to find a version that passes.”

Hours 36-48: The Willingness to Pay Test

This is the phase most founders skip because it feels too early. That instinct is wrong. Willingness to pay is the most reliable validation signal you can collect, and collecting it before you build is not premature. It is smart.

Here is why: email sign-ups are cheap. Clicking a link is free. Saying “yes I would use this” costs nothing. But paying $50 upfront for a product that does not exist yet is a real commitment. It separates people who are curious from people who have a problem painful enough to spend money on.

If even 5 to 10 people will pre-pay for your product before it exists, you have passed a significant validation threshold. Founders who collect $10,000 to $50,000 in pre-orders before building can self-fund initial development and have a much stronger position with investors.

How to ask for money before you have a product

For B2C products: offer a lifetime deal or a founding member discount. “We are launching in 60 days. I am offering 20 founding member spots at $49 (compared to the $29/month regular price) to the first people who sign up. You get access as soon as we launch, plus a lifetime 40% discount on the regular plan.”

This works for several reasons. The scarcity (20 spots) is real and creates urgency. The discount is meaningful. The framing is honest about where you are in development. And the ask is small enough to be a real test without being so small it is meaningless. I have seen founders get 15 founding member pre-sales in 48 hours using exactly this message posted to their target subreddit.

Use Stripe with a Typeform or a simple Gumroad page. It takes 30 minutes to set up. If you feel uncomfortable taking money for a product you have not built, use the framing “early access deposit” and be explicit about the refund policy if you decide not to build it. Most serious early adopters will accept this without hesitation.

The letter of intent for B2B

For B2B products, pre-sales are harder because purchasing decisions require procurement cycles and approvals. Use a letter of intent instead.

A letter of intent is not a contract. It is a non-binding document where a potential customer says “if this product is built to the following specification, we intend to purchase it at this price.” It takes 20 minutes to write and 10 minutes for a buyer to sign.

The value of an LOI is not legal. It is psychological and informational. A buyer who signs an LOI has moved beyond curiosity into something close to commitment. They have had an internal conversation about the purchase. They have imagined it in their workflow. They have agreed to a price. That is far more valuable signal than “yes this sounds interesting.”

I have seen B2B founders collect three to five LOIs in the final phase of a 48-hour sprint. That is enough to start building with confidence. The validation threshold: $5,000 to $10,000 in presale revenue or binding commitments before you write code.

What passing looks like

Phase 4 gate: at least three people have given you money or signed a letter of intent. Not “they said they would.” Not “they sounded really interested.” Money in your Stripe account or a signature on a document.

If you cannot get three paying commitments from twenty qualified prospects, you have a willingness-to-pay problem. Common reasons: the price is wrong (usually too high), the timing is wrong (need it but not right now), or the problem is real but not painful enough to pay today.

Each of those is a different pivot. Price is adjustable. Timing is addressable with a payment structure. If the problem is not painful enough to pay for, you may have a nice-to-have, not a need-to-have. That is not necessarily a kill signal. Some products start as nice-to-haves and become need-to-haves as the market evolves. But be honest with yourself about which one you are building.

The Kill or Continue Decision

After 48 hours, you have real data. Here is how to read it:

The common failure mode: founders who get mixed signals across the four phases decide to “build and see.” That is almost always the wrong call. Mixed signals mean something specific is wrong, and building will not fix it. Find out what the specific problem is before spending three months coding.

If Phase 1 passes but Phase 4 fails, you have a pain-point without a willingness to pay. The problem may be real but not acute enough to spend money on. Adjust your target customer, the severity of the problem you are addressing, or your price point.

If Phase 3 fails but Phase 1 and 2 pass, you have a messaging problem. The demand is real, but your description of the solution is not connecting. Rewrite the headline and value proposition. Run the smoke test again with a different framing.

The 7 Validation Signals Ranked by Reliability

Not all validation signals are equal. Here is how I rank them, from most reliable to least:

The pattern: signals that cost the other person something (money, calendar time, a signature) are reliable. Signals that cost them nothing (a like, a survey response, a “sounds great!”) are not. Weight your validation decisions accordingly.

The Contrarian Take: Most Validation Is Theater

I want to be honest about something: most founders who do “validation” are doing it to feel better about a decision they have already made, not to actually test whether the decision is right.

I have done this myself. I had an idea I was excited about. I talked to eight people. Six of them seemed interested. I declared it “validated” and built for four months. The product failed. When I looked back, I had been asking leading questions. I had been picking the people most likely to be positive. I had been interpreting vague interest as enthusiasm.

Real validation is designed to kill your idea. You are looking for reasons not to build, not reasons to build. The goal of Phase 1 is to find out if anyone cares. The goal of Phase 4 is to see if anyone cares enough to pay. Every step is a filter. You want the filter to be as harsh as possible.

There is a second piece of theater worth naming: the validation spiral. Some founders use the validation process as a way to avoid building. They interview 50 people. They run three landing page tests. They spend six weeks “validating” when the real blocker is fear of building something imperfect. Validation is not a substitute for building. It is a gate in front of building. 48 hours is the right timeframe. Six weeks is procrastination with extra steps.

The third thing most people get wrong: validation does not de-risk the execution. You can validate that a market exists, that people have the problem, that some of them will pay, and still fail completely because you built the wrong solution, priced it wrong, chose the wrong channel, or ran out of money before you reached the customers who would have bought. Validation is not a guarantee. It is a probability-increasing activity. Knowing the difference keeps you from either skipping it (fatal) or relying on it too heavily (naive).

What to Do Monday Morning

This is a 48-hour sprint. Here is how to actually run it this week.

Tonight (2 hours): Write your problem hypothesis. Not the solution, the problem. “I believe that [specific type of person] struggles with [specific problem] when they try to [specific goal]. They currently solve it by [current workaround] which costs them [time/money/frustration].” One paragraph. No solution language yet. Send 20 LinkedIn messages and 10 Reddit DMs to people who fit your target customer profile. Ask for 20 minutes tomorrow morning. Offer a $20 gift card.

Tomorrow morning (4 hours): Run 5-8 discovery calls using the question framework above. Take notes. Score each person on pain intensity, frequency, and existing spend. Check for convergence. If 3 or more people describe the same problem unprompted, you have passed Phase 1.

Tomorrow afternoon (4 hours): Reddit mining, Google Keyword Planner research, competitor G2 review audit. Looking for organic demand signals. If you find 5 threads and 1,000+ monthly searches with transactional intent, you pass Phase 2.

Tomorrow evening (2 hours): Build the landing page on Carrd. One headline, three bullet points, one call to action. Do not spend more than two hours. Post it in three relevant communities. Put $50 behind a Facebook or Reddit ad for your exact target audience.

Day two, morning (4 hours): Check the landing page numbers. Count the emails. If you are above 10 cold sign-ups, move to Phase 4. If not, rewrite the headline and re-post.

Day two, afternoon (4 hours): Call back the most engaged people from your discovery calls. Present a simple pre-order. “I am building this. Founding member price is $X. I have 10 spots. Want one?” For B2B, offer a letter of intent. Count the yeses.

Day two, evening (1 hour): Look at your scorecard. Four phases. Four gates. Where did you pass? Where did you fail? Make the kill or continue decision based on the data, not your gut. If you passed all four, start building. If you failed one, diagnose why and decide whether to iterate or kill.

The whole sprint costs about $50 in ads, a $20 gift card per interview (optional), and 48 hours of focused effort. The companies that skip this spend four months building and then run the same test involuntarily, at launch, when the data comes too late to change direction.

FAQ: Validating a Startup Idea Before Building

How many customer interviews do I need to validate a startup idea?

Start with five to eight interviews in Phase 1. The goal is pattern recognition, not statistical significance. When 60% of the people you talk to describe the same problem in similar terms, you have found a real signal. Beyond 15-20 interviews in a single session, you typically stop learning new things and start hearing confirmation of patterns you already identified. The 20-interview rule from SaaStr is a useful heuristic: do at least 20 customer interviews before committing to build anything, but you can get the core signal from the first 8.

What is a smoke test for a startup idea?

A smoke test is an experiment where you present the concept of a product that does not yet exist and measure how real people respond to it. The most common form is a landing page with a call to action (like “Get Early Access”) that you drive traffic to via ads or community posts. The click-through rate and email capture rate tell you whether your value proposition resonates with strangers, without you being in the room to explain or sell it. Buffer’s famous pre-launch landing page is the canonical example: they described a product that did not exist and collected emails for months before writing backend code.

Is a landing page enough to validate a startup idea?

No. A landing page alone tests one thing: whether your messaging resonates with cold traffic. It does not test whether the problem is real (Phase 1), whether there is market-scale demand (Phase 2), or whether people will actually pay (Phase 4). A landing page that collects 50 emails is encouraging but not conclusive. Use it as one gate in a four-phase process, not as the whole process. The founders who get into trouble are the ones who hit 200 email sign-ups and declare themselves validated before testing willingness to pay.

What is a letter of intent and should I use it for B2B validation?

A letter of intent (LOI) is a non-binding document where a potential customer states that they intend to purchase your product if it is built to a specified set of requirements at a stated price. It is not a contract. The buyer can walk away. But signing one requires an internal conversation, a review of your proposed spec, and an agreement on price, which filters out casual interest and surfaces real intent. For B2B products with deal sizes above $1,000 per year, LOIs are far more reliable validation signals than email sign-ups or survey responses. Three signed LOIs in your first 48-hour sprint is a very strong green light.

What is the difference between problem validation and solution validation?

Problem validation confirms that a specific type of person has a painful, frequent, and currently unsolved (or poorly solved) problem. Solution validation confirms that your specific proposed solution is the right way to address it. Most failed products fail at solution validation, not problem validation. The founders confirmed the problem was real but built the wrong solution shape, the wrong delivery mechanism, or the wrong pricing model. The 48-hour sprint validates both: Phase 1 is problem validation; Phases 3 and 4 are solution validation via the smoke test and willingness to pay test.

What if nobody responds to my outreach for interviews?

Low response rates on outreach usually mean one of three things: your target customer is too hard to reach through the channels you are using, your outreach message is too long or too pitch-heavy, or the problem you described in your message is not resonating. The fastest fix is to shorten your message to two sentences, lead with the problem rather than your idea, and offer something in return (a gift card, your expertise on a related topic, a reciprocal introduction). If you still get zero responses from 50 messages, consider whether you have correctly identified your target customer. The people living the problem most acutely are usually findable; you may just be looking in the wrong community.

Can I validate a startup idea without spending money?

Yes, but it takes longer. The $50 ad spend in Phase 3 buys speed. It puts your concept in front of cold traffic within hours. Without it, you rely on organic posts in communities, which take longer to get traction and introduce selection bias (the people in your communities are not the same as the general market). If budget is a real constraint, post your landing page to Reddit, Hacker News, Product Hunt, and LinkedIn with a genuine “I am testing this idea, tell me what you think” framing. Expect it to take two to three days to get meaningful traffic rather than 12 hours. The rest of the sprint, including customer interviews and letters of intent, costs nothing except time.

How do I know when to kill an idea versus when to iterate?

Kill the idea when the problem itself is not painful enough (Phase 1 scores averaging below 6 out of 10), when there is no existing demand signal online despite thorough searching (Phase 2), or when nobody will pay even at a very low price point with a very clear value proposition (Phase 4). Iterate when the problem is real but the solution description is not connecting (fix messaging), when the smoke test fails but interviews are strong (fix the framing of the product), or when the wrong people are responding to your outreach (fix the target customer definition). The most common mistake is killing an idea after one phase failure without diagnosing whether it is a problem failure or a framing failure. They require different responses.