The FORGE Validation Sprint: A 7–14 Day Plan (with templates)

Most posts on validation give you a framework. This one gives you a calendar.

This post turns the FORGE validation framework into a seven-day sprint you can start on Monday. If you want more data, extend to 14 days. Either way, by the end you will have a clear answer: build, or move on.

What the sprint covers

Seven days. One idea. A structured process for finding out whether genuine demand exists before you write a line of code.

The sprint has three phases:

• Days 1–2: Understand. Frame the problem and define who has it.
• Days 3–5: Test. Deploy a fake door and run discovery interviews.
• Days 6–7: Decide. Analyse your signals and make the call.

For founders with more time or a more complex problem, the extension adds another week of interviews and a second round of landing page traffic. The decision still happens on day 14.

Days 1–2: Understand

Day 1: Frame the problem

Before you test anything, you need to be precise about what you are testing.

Write one sentence that describes the problem, the person who has it, and the cost of not solving it. Be specific. "Freelancers waste time on invoicing" is not a problem statement. "Independent consultants spending more than three hours per week reconciling client payments across multiple platforms" is.

Then write down your assumptions. What would have to be true for this to be a business? At minimum: the problem is common, it is painful, people are actively looking for a better solution, and they would pay for one. Write those assumptions down before the sprint starts. Your job is to test them.

Day 2: Define your ICP

Your ideal customer profile is not a demographic description. It is a behavioural one.

What are they doing right now to solve the problem? What tools are they using? What are they searching for? Where do they gather? Use LinkedIn filters, subreddit searches, and job posting data to count how many of them exist. If you cannot find 10,000 potential customers using public data, your ICP is probably too narrow or the problem is not widespread enough.

This day is research. Find the community forums, the job posts, the competitor review pages. That is your raw material.

Days 3–5: Test

Day 3: Build and launch your fake door

A fake door is a landing page for a product that does not exist yet. The goal is to see whether people in your ICP will exchange their email for the promise of a solution.

The page needs three things: a clear problem statement, a description of the outcome your product delivers, and a single call to action. Do not describe features. Describe the result.

Forge lets you deploy a static landing page in minutes from a git push. Set up a simple page, connect it to a form tool for the email capture, and push it live. Total build time should be under two hours.

Set your target: 50 qualified visitors over the next two days to get a meaningful conversion rate. Qualified means people who match your ICP, not traffic from your personal network. The fake doors post covers how to design a landing page test that gives you real evidence rather than inflated sign-up counts.

Days 4–5: Drive traffic and run interviews

Drive traffic through two channels.

Paid. Run a small ad. £50 to £100 on Google or Meta, targeted at the segment you defined on day 2. Keep the ad copy aligned with the landing page. You are not optimising for the lowest cost per click; you are optimising for qualified visitors.

Direct. Find 10 to 15 people in your ICP through LinkedIn, Reddit, or community forums. Send a short, honest message: you are testing an idea and would value 20 minutes of their time.

Run five discovery interviews on days 4 and 5. Use the objection-first method: open with the reasons your product probably will not work for them, then listen. The people who push back on your objections are your real signal. Those who nod and disengage are not.

Record and transcribe every call. Otter.ai works well for this. You will need the raw transcripts on day 6.

Rob Fitzpatrick's The Mom Test is worth reading before your first interview. The core rule: never ask anyone if your idea is good. Ask about their life, their current tools, and their frustrations instead.

Days 6–7: Decide

Day 6: Analyse your signals

By the end of day 5 you should have a landing page conversion rate, five interview transcripts, and your original list of assumptions from day 1.

Work through the signal hierarchy. Map each piece of evidence to a tier. Landing page conversions are tier 2 signals at best. Interview engagement and explicit objection-pushback are tier 3. Any pre-order or pilot payment is tier 4.

Look for patterns in the interview transcripts. Did the same objection come up repeatedly? Did anyone push back hard on your disqualifiers? Did anyone ask how to sign up before you mentioned it?

You can use Claude or Perplexity to cluster your interview notes by theme. Feed in the transcripts and ask for the three to five most common concerns and the most common positive responses. This is one of the places AI compresses what would otherwise take hours of manual analysis.

Day 7: Make the Go/No-Go call

Apply the threshold test from the demand testing post:

Go: At least one tier 4 signal, at least two interviews where the prospect challenged your objections, and a clear picture of your ICP. Start building a minimal version for that segment, not a general product.

Pause: Strong tier 3 signals but no tier 4. Run a concierge pilot. Manually deliver the service for a paying customer before automating anything. Paul Graham's essay Do Things That Don't Scale covers why manual delivery before automation is the right move at this stage.

No-go: Only tier 1 and tier 2 signals. The idea may be interesting, but you do not have evidence of a market. Return to day 1 with a sharper problem statement or a different ICP.

Write the verdict down. If you go, write a single paragraph describing what you are building and for whom. If you pause or stop, write down what you learned and which assumption was wrong. Either outcome is more valuable than six months of building without asking.

The 14-day extension

If you are not getting enough qualified traffic or interviews in the first week, extend the sprint. Use days 8 to 12 to run five more interviews and refine the landing page based on what you heard. Do not add features to the page. Sharpen the problem statement.

Day 13 is a second signal analysis. Day 14 is the final Go/No-Go.

If you still cannot collect a tier 4 signal after 14 days and 50 qualified visitors, the problem either is not painful enough or is not widespread enough. That is the answer. It is a good one to have before you build.

---

The sprint format keeps you from one of the most common founder traps: building for months before finding out whether anyone wants what you are making. Seven days is short enough to feel urgent. The decision framework at the end is clear enough to act on. If you want to understand the full validation toolkit before running the sprint, validation without an MVP covers every pre-build test in detail. And if the sprint surfaces that you need better interview technique, customer discovery in the AI era covers how to capture and analyse what you hear. Run the sprint. Make the call.