How to Validate a Startup Idea Before You Build It or Raise Money

You can find out if people will pay for your idea before you spend $10,000 on an app or a month on a pitch deck. It takes 4 to 8 weeks, your time and a small ad budget. You talk to the people who have the problem, put up a page that promises a fix, ask for money, and then do the job by hand for a few customers. Before each step, you write down the result that means “keep going”.

Building something nobody wants is one of the most common ways a startup dies. CB Insights looked at 431 VC-backed companies that shut down since 2023 and found poor product-market fit in 43% of them (CB Insights, 2026). Running out of money topped the list at 70%, but the report calls that the final cause of death, not the root problem.

The five steps at a glance

Step	What you do	Time	Keep going when
1. Problem interviews	Talk to 20 or more people who have the problem	1-2 weeks	Most of them already spend time or money on a fix
2. Landing page test	Put up a one-page offer with a sign-up or “buy” button	1-2 weeks	You beat the sign-up rate you wrote down before launch
3. Pre-sales	Ask for a pre-order, a deposit or a signed letter of intent	1-2 weeks	Strangers pay, not only friends
4. Manual MVP	Deliver the result by hand to 5-10 customers	2-4 weeks	They come back and pay again without being chased
5. Decide	Build the MVP, change the idea or stop	A day	The next question needs real software to answer

Steps 1 and 2 can run at the same time. After that, order matters. Each step costs a bit more than the one before, so a weak result stops you early, while it’s still cheap.

Step 1: Talk to people about their problem, not your idea

The first test is free. Find people who have the problem and ask how they deal with it today.

Skip friends and family. They like you, so they’ll like your idea. Look where your customers already spend time: industry groups, forums, LinkedIn, local businesses, people who left reviews of a competitor.

How many is enough? Todd Jackson of First Round Capital wrote that about 20 dog owners were enough for one consumer idea he worked on, and that for B2B products he prefers 50 or more (Lenny’s Newsletter).

Ask about the past, not the future. “Would you use an app that does X?” gets a polite yes. “Tell me about the last time this happened” gets you what really happened. Rob Fitzpatrick’s book The Mom Test is built on this idea.

Questions that get honest answers

• When did this last happen to you? What did you do?
• What have you tried to fix it? What did that cost?
• What’s annoying about how you do it now?
• Who else deals with this, and who pays for the fix?
• Can I show you something rough next week?

Leave your pitch for the end, or skip it. The last question is a small test of its own: people who really have the problem say yes and give you a time.

What a strong answer sounds like

A strong answer describes the problem before you bring it up. The person has already paid for something or built a workaround in spreadsheets, and they ask when they can try your fix. A weak answer sounds like “cool idea” or “I’d probably use that”, followed by silence. Count that as a no, however nice it sounded.

Write down the exact words people use. They’ll become the headline of your landing page in step 2.

Step 2: Put up a landing page before the product

A landing page test, also called a smoke test or a fake door test, checks if strangers want your offer enough to click. You make one page that says what the product does and who it’s for, with one button: “Join the waitlist”, “Get early access” or a price. The product behind the button doesn’t exist yet. People who click see an honest note that it’s coming and a box for their email.

Buffer started this way. Its founder, Joel Gascoigne, first put up two pages to see if people would consider the app at all. Then he added a pricing page between them to see which plan people clicked. He built the product in 7 weeks and had his first paying customer 4 days after launch (Buffer).

How to run yours:

• Use the words from your interviews in the headline.
• Get traffic you didn’t hand-pick: a small ad budget, posts in groups where your customers hang out, cold emails.
• Track two numbers: how many visitors leave an email, and how many click a paid plan.
• Write down your target before you start. There’s no universal “good” sign-up rate, and if you set the bar after you see the result, you’ll always pass.

AI site builders can make this page in an hour. Free “AI idea validators” are a different thing: they score your idea from its description. That score is a guess about strangers. The landing page shows you what strangers did.

Step 3: Ask for money

Joining a waitlist costs a visitor nothing. Paying costs them something, which is why a payment is the stronger signal.

Ways to pre-sell:

• Pre-orders at an early-bird price, with a full refund if you don’t deliver.
• A paid pilot for a business customer: a few weeks of the service for a fixed fee.
• A letter of intent: a signed note that a company plans to buy at a stated price once the product works. It’s weaker than cash, but it counts in B2B.
• A deposit that holds a spot in the first group of users.

A payment link and an honest description are enough. Tell people what they’ll get and when.

If nobody pays, ask why. “Too expensive”, “not right now” and “my boss would have to approve it” are three different problems. The first points to the price, the second to how urgent the problem is, the third to who you’re selling to.

Step 4: Do the job by hand first

Before anyone writes code, deliver the result yourself to a few paying customers. You can do it in the open or behind a screen.

	Concierge MVP	Wizard of Oz MVP
What the customer sees	A person doing the work for them	A product that looks automatic
What happens behind it	You do the work in the open	You do the work by hand, out of sight
Best for	Learning what the job really involves	Testing if people will use a product, not a service
Watch out for	Customers paying for your attention, not the result	Promising a speed you can’t keep up by hand

Paul Graham wrote that when you only have a few users, you can do by hand things you plan to automate later. One of his examples is Stripe, whose founders set up the early “instant” merchant accounts for users by hand, behind the scenes (Paul Graham, Do Things That Don’t Scale).

This step fits AI startups well. If your product will use AI to write, sort or summarize something, do that job yourself with off-the-shelf AI tools, check every result, and send it to the customer. You’ll find out whether the output is worth paying for before anyone builds the real thing.

Then watch what customers do. The best sign is a customer who comes back and pays a second time without a reminder from you, and maybe brings a colleague. That’s as close to proof as you can get before there’s software.

The numbers that mean “go”

Set the bar for each step before you start, then compare.

Step	Go signal	Change or stop signal
Interviews	Most people describe the problem on their own and already spend money or hours on it	Polite interest and no workaround today
Landing page	You beat the target you wrote down	Clicks only come from people you know
Pre-sales	Strangers pay or sign a letter of intent	“Let me know when it’s ready”
Manual MVP	Customers come back and pay again	One-time use, no referrals
After launch	40% or more of users would be “very disappointed” without the product	Under 40% and not rising

The last row is the Sean Ellis test. You ask users how they’d feel if they could no longer use the product. Ellis found that companies where fewer than 40% answered “very disappointed” usually struggled to grow. Superhuman started at 22% and got to 58% within three quarters (First Round Review). You can only run this test once people use a real product, which is one more reason to keep the first version small.

What investors want to see before pre-seed

Pre-seed money now goes to fewer startups. In Q2 2026, US startups on Carta raised $3.19 billion across about 11,500 pre-seed deals. A year earlier, $3.22 billion went into 14,825 deals (Carta). That’s about the same money spread over roughly a fifth fewer deals.

An idea and a deck rarely clear that bar on their own. The steps above give you the evidence pre-seed investors look for:

• Interview notes that show a painful problem, in your customers’ words.
• A sign-up rate from traffic you didn’t hand-pick.
• Money: pre-orders, deposits, paid pilots or signed letters of intent.
• What you learned and changed along the way, which shows you move fast.
• A plan for who builds the product and how soon.

The last point trips up many non-technical founders, because investors will ask who builds it. A fractional CTO can own that answer part-time, from the technical plan to the first release, without a full-time salary.

When to build an MVP, and when to wait

Build when people have paid or committed, you know the one job the first version must do, and your next question can only be answered by working software, such as “do people come back in week three?” or “will they pay $20 a month without me doing the work?”

Wait if you haven’t talked to 20 people yet, if your only fans are friends, or if you can still test the idea by hand.

Once the answers are in, the first version doesn’t have to take months. With us, an MVP costs $5,000 to $15,000 and takes 2 to 4 weeks. Our guides on what an MVP costs and how long it takes to build an app show where the money and the weeks go. If your product runs on AI, read what an AI app costs as well.

Testing goes on after launch: GhostMinutes

GhostMinutes is an AI app that turns meetings into notes. The founder’s plan was to put a paid product in front of users fast and then find out which audience and price worked. We built the iOS app in 2 weeks and a web app a week later. Then we spent a couple of weeks testing features and pricing with the founder’s marketing team:

• separate landing pages for sales teams, consultants and people who transcribe video;
• a short quiz that suggests a plan;
• four plans, from Free to Business, paid monthly or yearly;
• one-time packs of minutes that never expire.

GhostMinutes reached $1,500 in monthly recurring revenue (GhostMinutes case study). A landing page couldn’t have told the founder which plan people pick when they pay inside a working app. These tests could.

EasySpk, an AI voice keyboard for iPhone, took a similar path. It went from naming to the App Store in 2 weeks, got 1,000+ downloads in its first days, and then moved to 1-week sprints that tested growth ideas (EasySpk case study).

FAQ

How do you validate a startup idea without building anything?

Talk to 20 or more people who have the problem. Put up a one-page offer with a sign-up or price button. Ask for money through pre-orders, deposits or letters of intent, then deliver the result by hand to a few paying customers. If strangers pay and come back, you have enough evidence to build.

How many customer interviews do you need?

About 20 is a common starting point for a consumer product. For B2B, 50 or more is safer, because businesses differ more in how they work and buy. You can stop when you keep hearing the same problems in the same words.

What is the difference between a concierge MVP and a Wizard of Oz MVP?

In a concierge MVP, the customer knows a person does the work. In a Wizard of Oz MVP, the customer sees what looks like a finished product, while a person does the work behind it. A concierge MVP teaches you the job, and a Wizard of Oz MVP tests if people will use a product.

Is a waitlist enough to prove demand?

No. A waitlist shows interest, and joining one costs nothing. Pre-orders, deposits and paid pilots are stronger, because people give up money. Use your waitlist to find the people you pre-sell to.

Do investors fund an idea without an MVP?

Some do, mostly when the founders have built and sold companies before. Most pre-seed investors want evidence: interview notes, sign-ups, pre-sales or early revenue. In Q2 2026, US pre-seed money on Carta went into about a fifth fewer deals than a year earlier, so the bar for an idea alone is high.

How much does it cost to validate a startup idea?

Interviews and pre-sales cost your time. A landing page costs little with today’s site builders, plus whatever ad budget you choose. A manual MVP costs your time and the tools you use. If the tests pass, an MVP with us costs $5,000 to $15,000 and takes 2 to 4 weeks.

Know what to build first

Describe your idea in a few sentences. Our AI estimate turns it into a scope, a budget range and a timeline in about four minutes, so you can see what a first version would cost before you commit.

Estimate my MVP