# Validate an Idea in One Week

> Before you build for months, spend seven days testing the one thing that must be true. A day-by-day plan, pass bars, and how to decide go, pivot or stop.

- Author: Tai Pham Phu
- Published: 17 Nov 2026
- Tags: Startups, Entrepreneurship, Product, Business, Small Business
- URL: https://phamphutai.com/2026/11/17/validate-an-idea-in-one-week/

Before Dropbox was a product most people could use, Drew Houston made a short video. It was a screen recording, about three minutes long, with Houston narrating while he showed how the software was meant to work. He filled it with in-jokes for the early adopters on Digg, the tech news site where it spread.

Eric Ries tells the story in a 2011 [TechCrunch guest post](https://techcrunch.com/2011/10/19/dropbox-minimal-viable-product/) and in *The Lean Startup*. He quotes Houston on the result: the beta waiting list went **"from 5,000 people to 75,000 people literally overnight."**

The video didn't prove Dropbox would make money. It didn't prove the hard engineering would work at scale. It proved one specific thing, cheaply: **a lot of people wanted this, badly enough to leave an email and wait.** That was the question Houston needed answered, and he answered it without shipping a public product.

This post is about doing the same for your idea, in one week. Not a perfect answer. A good-enough answer, before you spend months building.

---

## Why "build it and see" wastes months

Most first-time founders, especially engineers, validate an idea like this: build it, launch it, see what happens. I understand the pull. Building is the part we know how to do.

The problem is what "see what happens" tells you. Suppose you spend four months building, launch, and almost nobody signs up. Why? Nobody has the problem? They have it but won't pay? Wrong price, confusing page, wrong places? **A failed launch mixes all of these answers together.** You paid four months for a result you can't read.

There's a quieter cost too: after four months, you're attached, and you'll explain away weak signals instead of hearing them.

A week of tests flips this. Each test asks **one question**, gives a clear answer, and costs days, not months. If the idea dies, it dies cheaply. If it survives, you build with real evidence and, often, your first customers already waiting.

---

## Find your riskiest assumption

Every idea rests on a few things that must be true for it to work. Most of them are probably fine. One or two are the ones that could kill you. Those are what you test first.

A useful way to sort them comes from the design firm IDEO, which describes three lenses for any new idea. On its [design thinking site](https://designthinking.ideo.com/history), IDEO frames them as questions:

- **Desirability:** are we solving a problem that matters to people? *Do they want it?*
- **Viability:** can the solution succeed and last within real-world constraints? *Will it make money, at a price they'll pay?*
- **Feasibility:** can we build it with today's technology and skills? *Can we actually deliver it?*

For most small software ideas, **feasibility is the least risky lens.** You can probably build it. The ones that kill startups are desirability (nobody cares enough) and viability (they care, but won't pay enough to cover your costs). If you've read [Unit Economics in Plain English](https://phamphutai.com/2026/11/11/unit-economics-in-plain-english/), viability is where that math starts.

Here's a simple exercise for Day 1. Finish this sentence ten times: **"This only works if..."**

> This only works if tutors lose real time every week chasing payments.
> This only works if they'll pay $15 a month to stop.
> This only works if I can reach them without paid ads.
> ...

For each one, ask: **how sure am I?** and **if it's wrong, does the idea die?** The least certain one that would kill the idea is your **riskiest assumption**. Build the week around it.

**Test the thing most likely to kill you, not the thing easiest to test.** Founders often test the comfortable question (can I make a nice landing page?) and avoid the scary one (will anyone pay?).

---

## The evidence ladder

Before the plan, one idea that makes every test easier to read: **not all "yes" answers are equal.** Some cost the person nothing. Some cost them something real.

| Signal | What it costs them | Strength | What it tells you |
|---|---|---|---|
| "Great idea!" | Nothing | None | They're being kind |
| Email signup | A few seconds | Weak | Curious enough to click twice |
| Time commitment (a call, a trial) | An hour or more | Medium | The problem is real enough to give up time |
| Deposit or pre-order | Some money, refundable | Strong | They'd pay to have it solved |
| Paid at full price | Real money, kept | Strongest | You have a customer |

Each rung is harder to fake than the one below it. In [Your First 10 Customers](https://phamphutai.com/2026/11/11/your-first-10-customers/) I described the same idea through Rob Fitzpatrick's *The Mom Test*: count commitments, not compliments. This week is about **pushing people up the ladder** as far as they'll go, and writing down where they stop.

---

## The example we'll follow

> **Fictional example: LessonLoop.** A founder, call him Duc, wants to build a simple tool for independent tutors. It sends parents a bill after each lesson, reminds them when it's late, and shows the tutor who still owes what. His guess: tutors with 10 or more students lose hours each month chasing payments and feel awkward doing it. Planned price: $15 a month.

(LessonLoop is invented. Every number below is an illustration, not data from a real company, including any of mine.)

Duc's Day 1 list has three risky assumptions:

1. **Desirability:** tutors feel this pain often enough to want a fix.
2. **Viability:** they'll pay about $15 a month for it.
3. **Reach:** he can find tutors cheaply, through groups and referrals.

The riskiest is number 2. Lots of tutors probably complain about late payers. Fewer may pay a monthly fee when a notebook and a few awkward messages sort of work. So his week has to end with money on the table, not just emails.

---

## The 7-day plan

One test per day, sometimes two. Some tests run in the background for several days (the landing page keeps collecting visitors while you do other things). For each one: what it proves, what it doesn't, and a **pass bar you write down before you start.**

If you set the bar after seeing the result, you'll set it just below whatever you got.

### Day 1: Write it down

Write your riskiest assumptions, the tests for each, and a pass bar for each test. Write what you'll do if a test fails. Keep it to one page.

Duc's pass bars:

| Test | Pass bar (set on Day 1) |
|---|---|
| Conversations | At least 6 of 10 tutors describe chasing a payment in the last month |
| Landing page | At least 5% of visitors leave an email |
| Deposit | At least 5 tutors pay a $20 refundable deposit |
| Concierge | At least 2 of 3 tutors want to keep going and pay |

**Proves:** nothing yet. **Does:** stops you from fooling yourself later.

### Day 2: Customer conversations

Talk to 8 to 10 people who should have the problem. Ask about their past, not your idea: "When did a parent last pay late? What did you do?" I covered how to run these calls, and where to find people, in [Your First 10 Customers](https://phamphutai.com/2026/11/11/your-first-10-customers/), so I won't repeat it here.

Two things to add for a validation week:

- **End every call with an ask.** "Can I send you something to try next week?" or "Who else should I talk to?" Note who says yes.
- **Write down their exact words.** You'll use them on Day 3's landing page.

**Proves:** the problem exists, how people describe it, and what they use today. **Doesn't prove:** that they'll pay, or that strangers feel it as strongly as the people you know. **Pass bar:** Duc's was 6 of 10 with a recent, specific story.

### Day 3: Landing page (a fake-door test)

A **landing page** is a single web page that describes your product and asks visitors to do one thing. A **fake-door test** means the "door" (a sign-up or pricing button) leads to something that doesn't fully exist yet. You measure how many people try to walk through it.

Buffer started this way. In a [2011 post on the Buffer blog](https://buffer.com/resources/idea-to-paying-customers-in-7-weeks-how-we-did-it), founder Joel Gascoigne describes a **two-page test**: a page explaining the idea, and a second page where people could leave an email to hear when it launched. He tweeted the link and asked for opinions. Then he added a pricing page between the two, so people had to pick a plan before leaving their email. That one extra click tested two things: whether people were keen enough to click again, and which price they'd choose. Some picked paid plans. Only then did he build the product, in about seven weeks of evenings and weekends. He writes that his first paying customer came within four days of launch.

A good fake-door page has:

1. **A headline in your customers' words.** Duc's: "Stop chasing parents for lesson payments."
2. **Three short lines** on what it does.
3. **A price.** Even a rough one. Without it, you're testing curiosity, not demand.
4. **One button**, then a second step that collects an email.

**The honesty note.** A fake door must not trick people. The second page should say plainly: **"We're not ready yet. Leave your email and we'll tell you when it is."** Don't take card details for a product that doesn't exist. Don't charge anyone. Don't pretend there are users you don't have. The test still works when people know it's early, and you keep the trust of the exact people you hope will become customers.

Send the page to the places your customers gather: the groups and communities from your Day 2 list, plus anyone you talked to. Leave it running through Day 6.

**Proves:** a stranger understands the offer and cares enough to leave an email, at a stated price. **Doesn't prove:** they'll pay. Emails are the second rung of the ladder. **Pass bar:** Duc's was 5% of visitors.

### Day 4: Ask for a deposit

This is the test for viability, and the one most founders skip because it's uncomfortable.

Email everyone who signed up, and call the most interested people from Day 2. Make a real offer: **"I'm starting with 10 tutors next month. A $20 deposit holds your spot, and it counts toward your first month. If I don't launch, or you change your mind before then, you get it back in full."**

Notice what this is and isn't. It's a real payment, for a real promise, with a clear refund. It isn't a fake charge on a fake checkout. People know exactly what they're paying for.

Zappos started with a version of this. Founder Nick Swinmurn told [*Fortune*](https://fortune.com/2012/09/05/nick-swinmurn-zappos-silent-founder/) in 2012 that he asked a shoe store if he could photograph its shoes and put them online, offering: **"if people buy them, I'll buy them from you at full price."** He held no stock. Customers paid real money for real shoes; he just didn't buy them until someone did.

**Proves:** people will part with money for this, at roughly this price. **Doesn't prove:** they'll keep paying every month. **Pass bar:** Duc's was 5 deposits.

### Day 5: A concierge test

A **concierge test** means delivering the product's result by hand, and the customer knows a person is doing it. (I described it in more detail in [Your First 10 Customers](https://phamphutai.com/2026/11/11/your-first-10-customers/).)

Duc picks three tutors who left deposits. For one week, he does the job himself: they send him each lesson by message, he sends parents a short bill and a payment link, and he follows up on late ones. One shared spreadsheet, no app.

This is where you learn what no landing page can tell you. Maybe tutors don't want automatic reminders, because some parents are friends. Maybe the real pain is knowing who owes what, not chasing.

**Proves:** the result is valuable when it actually arrives, and which parts matter. **Doesn't prove:** you can deliver it profitably at scale. That's the next stage. **Pass bar:** Duc's was 2 of 3 tutors asking to keep going and pay.

### Day 6: A paid-ads smoke test (only if cheap)

A **smoke test** sends paid traffic to your landing page to see if strangers, who have never heard of you, react the same way as your communities did.

Do it only if it's cheap, and read it with care. **Small samples lie.** If 2 out of 20 visitors sign up, that looks like 10%. But with only 20 visitors, the true rate could plausibly be anywhere from about 3% to 30%. One more or one fewer signup swings the picture.

So a tiny ad test can tell you if a headline gets **any** clicks at a sane cost, not your real conversion rate. If you can't afford a few hundred visitors, skip this day and have more conversations instead.

**Proves:** whether strangers click, and a rough cost per click. **Doesn't prove:** much else at small sizes. **Pass bar:** a sign-up rate in the same ballpark as your organic traffic, and a cost you could live with.

### Day 7: Decide

Put every result next to its pass bar. Then make one of three calls:

- **Go:** the riskiest assumption passed, with money on the table. Start building, and build only what the concierge test showed mattered.
- **Pivot:** the problem is real, but something specific failed: the price, the customer type, or the solution. Change that one thing and run another week.
- **Stop:** people don't feel the problem, or won't pay anything to fix it. Write down what you learned, and move on. A week is a cheap price for that answer.

The rule I'd suggest: **a pass needs at least one rung of real money.** Emails alone are a "maybe," never a "go."

---

## The week, as a picture

<figure class="diagram">
<svg viewBox="0 0 680 390" role="img" aria-label="Seven-day validation plan as a timeline: Day 1 assumptions, Day 2 talk, Day 3 landing page, Day 4 pre-sell, Day 5 concierge, Day 6 optional ads, Day 7 decide. Below it, an evidence ladder rising from compliment to email signup, time, deposit and paid." xmlns="http://www.w3.org/2000/svg">
<defs><marker id="arr-v1w" viewBox="0 0 10 10" refX="9" refY="5" markerWidth="7" markerHeight="7" orient="auto"><path d="M0,0 L10,5 L0,10 z" fill="currentColor"/></marker></defs>
<text x="20" y="24" font-size="14">One week, one test a day (fictional LessonLoop plan)</text>
<rect x="20" y="40" width="84" height="80" fill="none" stroke="currentColor"/>
<text x="62" y="62" font-size="13" text-anchor="middle">Day 1</text>
<text x="62" y="82" font-size="11" text-anchor="middle">Assumptions</text>
<text x="62" y="104" font-size="11" text-anchor="middle" class="muted">pass bars</text>
<rect x="112" y="40" width="84" height="80" fill="none" stroke="currentColor"/>
<text x="154" y="62" font-size="13" text-anchor="middle">Day 2</text>
<text x="154" y="82" font-size="11" text-anchor="middle">Talk</text>
<text x="154" y="104" font-size="11" text-anchor="middle" class="muted">time</text>
<rect x="204" y="40" width="84" height="80" fill="none" stroke="currentColor"/>
<text x="246" y="62" font-size="13" text-anchor="middle">Day 3</text>
<text x="246" y="82" font-size="11" text-anchor="middle">Fake door</text>
<text x="246" y="104" font-size="11" text-anchor="middle" class="muted">emails</text>
<rect x="296" y="40" width="84" height="80" fill="none" stroke="currentColor"/>
<text x="338" y="62" font-size="13" text-anchor="middle">Day 4</text>
<text x="338" y="82" font-size="11" text-anchor="middle">Pre-sell</text>
<text x="338" y="104" font-size="11" text-anchor="middle" class="muted">deposits</text>
<rect x="388" y="40" width="84" height="80" fill="none" stroke="currentColor"/>
<text x="430" y="62" font-size="13" text-anchor="middle">Day 5</text>
<text x="430" y="82" font-size="11" text-anchor="middle">Concierge</text>
<text x="430" y="104" font-size="11" text-anchor="middle" class="muted">paid use</text>
<rect x="480" y="40" width="84" height="80" fill="none" stroke="currentColor" stroke-dasharray="4 3"/>
<text x="522" y="62" font-size="13" text-anchor="middle">Day 6</text>
<text x="522" y="82" font-size="11" text-anchor="middle">Ads (maybe)</text>
<text x="522" y="104" font-size="11" text-anchor="middle" class="muted">click rate</text>
<rect x="572" y="40" width="88" height="80" fill="currentColor"/>
<text x="616" y="62" font-size="13" text-anchor="middle" class="on-fg">Day 7</text>
<text x="616" y="82" font-size="11" text-anchor="middle" class="on-fg">Decide</text>
<text x="616" y="104" font-size="11" text-anchor="middle" class="on-fg">go or stop</text>
<line x1="20" y1="138" x2="656" y2="138" stroke="currentColor" stroke-width="1.5" marker-end="url(#arr-v1w)"/>
<text x="384" y="164" font-size="11" text-anchor="middle" class="muted">page keeps collecting visitors, Days 3 to 6</text>
<line x1="204" y1="146" x2="564" y2="146" stroke="currentColor" stroke-dasharray="2 3"/>
<text x="20" y="188" font-size="14">The evidence ladder: each rung is harder to fake</text>
<rect x="20" y="320" width="120" height="40" fill="none" stroke="currentColor"/>
<text x="80" y="345" font-size="12" text-anchor="middle">Compliment</text>
<rect x="148" y="290" width="120" height="70" fill="none" stroke="currentColor"/>
<text x="208" y="310" font-size="12" text-anchor="middle">Email signup</text>
<text x="208" y="326" font-size="11" text-anchor="middle" class="muted">a few seconds</text>
<rect x="276" y="260" width="120" height="100" fill="none" stroke="currentColor"/>
<text x="336" y="280" font-size="12" text-anchor="middle">Time</text>
<text x="336" y="296" font-size="11" text-anchor="middle" class="muted">a call, a trial</text>
<rect x="404" y="230" width="120" height="130" fill="none" stroke="currentColor" stroke-width="2"/>
<text x="464" y="250" font-size="12" text-anchor="middle">Deposit</text>
<text x="464" y="266" font-size="11" text-anchor="middle" class="muted">refundable money</text>
<rect x="532" y="200" width="128" height="160" fill="currentColor"/>
<text x="596" y="220" font-size="12" text-anchor="middle" class="on-fg">Paid</text>
<text x="596" y="236" font-size="11" text-anchor="middle" class="on-fg">kept money</text>
<text x="20" y="380" font-size="11" class="muted">Weaker on the left, stronger on the right. A "go" needs at least one money rung.</text>
</svg>
<figcaption>Each day pushes people one rung higher, from a kind word to money they won't ask back.</figcaption>
</figure>

---

## Duc's week, in numbers

Here's how the fictional LessonLoop week might go. The numbers are invented but the arithmetic is checked.

**Day 2.** Duc talks to 10 tutors. 7 describe chasing a payment in the last month, in detail. Two say they keep a notebook of who owes what and still lose track. **Pass** (bar was 6 of 10).

**Days 3 to 6.** He posts the landing page in four tutor groups and sends it to everyone from Day 2. Over four days it gets **400 visitors**.

| Step | Calculation | Result |
|---|---|---|
| Visitors | from groups and referrals | 400 |
| Email signups | 400 × 7% | 28 |
| Deposits ($20, refundable) | 400 × 2% | 8 |
| Deposits as share of signups | 8 ÷ 28 | about 29% |
| Deposit money held | 8 × $20 | $160 |

The landing page **passes** (7% against a 5% bar). The deposits **pass** (8 against a bar of 5).

A note on how sure he can be. With 28 signups out of 400, the true signup rate is very likely somewhere between about **5% and 10%**. That's a usable range. For deposits, 8 out of 400 gives a range of roughly **1% to 4%** of visitors. Wide, but clearly not zero. (These ranges come from a standard statistical formula, the Wilson interval, at 95% confidence. You don't need the formula; you need the habit of asking "how many people is this based on?")

**Day 5.** Three of the deposit tutors get the concierge service for a week. Two ask to keep going at $15 a month. The third says she likes it but only has six students, and the notebook is fine for her. **Pass** (2 of 3), plus a useful lesson: **the customer is the tutor with many students, not every tutor.**

**Day 6.** Duc spends **$60** on ads at about $0.40 a click: 150 clicks, 6 signups. That's 4%. Lower than his organic 7%, but with only 150 visitors the true rate could be anywhere from about 2% to 9%. It cost **$10 per signup**. He notes it and moves on. It neither confirms nor kills anything, which is the honest result of a small ad test.

**Day 7. The call: go, with a narrower customer.** The riskiest assumption (will they pay?) has passed with real money: eight deposits and two tutors ready to pay monthly. The concierge test changed the target to tutors with 10 or more students. Duc's next step is to keep serving the paying tutors by hand while he builds the smallest version that removes his own manual work, and then to go get the rest of his [first 10 customers](https://phamphutai.com/2026/11/11/your-first-10-customers/).

A **pivot** would have looked different: say 8% signups but only one deposit. The problem is real; the price or offer isn't. Duc would test a per-student price or a different buyer, like tutoring centres, for another week. A **stop**: 3 of 10 tutors with a recent story, 1% signups, no deposits. The problem isn't painful enough, and he moves on having built nothing.

---

## Common mistakes

### 1. Setting the pass bar after the results

"We got 3% signups, which is actually pretty good for a new product." Maybe. But you'd have said the same about 2%. **Write the bars on Day 1.** If you change them, write down why, before you look at more data.

### 2. Stopping at the email rung

Email lists feel like progress, but an email is cheap to give and easy to forget. **Always ask for something that costs money or real time** before you call it validated.

### 3. Faking the door too well

A fake checkout that takes card numbers, a "sold out" message for a product that doesn't exist, invented user counts on the page. These can get you sharper numbers, and they burn the trust of the exact people you want as customers. **Say it's early. Take only refundable deposits for a clear promise.** The honest test is good enough.

### 4. Reading too much into tiny numbers

Two signups from twenty visitors isn't "10%." It's two people. **Always write the count next to the percentage,** and be suspicious of any conclusion based on fewer than a few dozen people taking the action.

### 5. Testing only friends

Friends sign up, even leave deposits, because they like you. Kind, and almost useless. **At least half your evidence should come from strangers** who fit your customer profile.

### 6. Running the "week" for three months

The point is speed. If a test isn't giving clear answers, change the test or make the call with what you have. **Another month of maybe is exactly what this week is meant to save you from.**

---

## Why one week is enough

A week won't tell you whether you'll build a big company. It can tell you whether the **one thing most likely to kill your idea** is true, and that rarely needs months. Houston's video, Gascoigne's two pages and Swinmurn's photos of someone else's shoes had the same shape: test the biggest unknown with the smallest thing, and let real people answer with their time, email or money.

**Don't ask people if they like your idea. Give them a chance to act on it, and watch what they do.**

---

## Key ideas

- **"Build it and see" mixes many answers into one unreadable result.** Small tests each answer one question, in days.
- **Find your riskiest assumption first.** Sort with IDEO's three lenses: desirability, viability, feasibility. For most software ideas, viability (will they pay?) is the scary one.
- **Climb the evidence ladder:** compliment, email, time, deposit, paid. Each rung is harder to fake.
- **Set every pass bar on Day 1,** before you see any results.
- **Fake doors must be honest.** Say it's not ready, collect emails, take only refundable deposits, never fake charges.
- **Small samples mislead.** Always report the count with the percentage, and treat tiny ad tests as rough.
- **End the week with a decision:** go, pivot or stop. A "go" needs at least one rung of real money.

---

*Sources: Eric Ries, [How DropBox Started As A Minimal Viable Product](https://techcrunch.com/2011/10/19/dropbox-minimal-viable-product/) (TechCrunch, October 2011, quoting Drew Houston); Joel Gascoigne, [Idea to Paying Customers in 7 Weeks: How We Did It](https://buffer.com/resources/idea-to-paying-customers-in-7-weeks-how-we-did-it) (Buffer blog, February 2011); IDEO, [Design Thinking: history and the three lenses](https://designthinking.ideo.com/history); Dinah Eng, [Zappos's silent founder](https://fortune.com/2012/09/05/nick-swinmurn-zappos-silent-founder/) (Fortune, September 2012); Rob Fitzpatrick, [The Mom Test](https://www.momtestbook.com/). Confidence ranges in the worked example are 95% Wilson intervals, computed for this post.*

*Related: [Your First 10 Customers Won't Come From Marketing](https://phamphutai.com/2026/11/11/your-first-10-customers/), [Product-Market Fit: How to Know You Have It](https://phamphutai.com/2026/11/17/product-market-fit-how-to-know/), and [Unit Economics in Plain English](https://phamphutai.com/2026/11/11/unit-economics-in-plain-english/).*
