LP

Getting Started with MiniLab

A ten-minute walkthrough of how to take a product hypothesis from a rough idea to a confident Go, Iterate, or Kill decision — using real users, real ads, real money.

What MiniLab is for

You're trying to find out whether a product idea is worth building. MiniLab helps you do that without building the product first. Instead of writing code, you:

  1. Write down a positioning idea — who is this for, what problem does it solve.
  2. Put up a landing page that pretends the product exists.
  3. Send paid traffic to that page.
  4. Watch what real visitors do — click, scroll, sign up, pay a deposit.
  5. Decide whether to build it, change the angle, or drop it.

Everything in MiniLab is built around that loop. You'll see the same five stages reflected in the dashboard, in chat, and in the reports.

Who does what

MiniLab gives you three teammates you can talk to in chat. Each owns part of the journey. You don't have to remember which is which — when you describe what you want, the right one steps up. But it helps to know what each is responsible for.

Researcher
Coming up with the idea, validating demand, writing the brief.
LP-Builder
Designing, building, and publishing the landing page.
Growth-Manager
Running paid ads, watching the numbers, calling Go or Kill.

When a stage is done, the idea moves to the next role automatically. You can always see which stage an idea is at on the Ideas board.

The journey, stage by stage

Stage 1 — Have an idea (Researcher)

Open the Ideas page. You can either type a hypothesis yourself ("a private cloud for families to back up phone photos, $399") or ask the Researcher in chat to brainstorm. It can pull pain points from Reddit, X, and the web, and propose positioning angles you might not have thought of.

The Researcher will produce:

  • The idea card — name, direction, the core pain point it solves.
  • One or more angles — different ways to pitch the same product to different audiences. You'll choose which to test (or test more than one).
  • A research report — what people are saying about this problem today, who's competing, why now.
  • A landing-page brief — what the page should say, what it should look like, what the headline should promise.

When you're happy with the brief, mark the idea ready and the LP-Builder takes over.

Stage 2 — Put up a landing page (LP-Builder)

The LP-Builder takes the brief and produces a real, hosted landing page.

Ask the LP-Builder in chat to:

  • Pick a visual style — it shows you two directions side-by-side. Choose one.
  • Generate the page — copy, hero image, sections, call-to-action.
  • Set up tracking — every visitor, click, and conversion is recorded.
  • Wire up payments — for products that take a refundable deposit, you'll get a payment link.
  • Publish to a real URL — first a test link for you to review, then the production URL.

Once the page is live, the idea moves to the Growth-Manager.

Stage 3 — Run ads & decide (Growth-Manager)

This is the longest stage. The next section is dedicated to it.

The Growth-Manager in detail

Once a landing page is live, the Growth-Manager's job is to buy traffic to it, learn from what visitors do, and recommend a verdict. Here's what that looks like day to day.

Setting up the campaigns

Before any money is spent, the Growth-Manager works with you to define:

  • Platform — Google Search, Google Performance Max, Meta Feed, or several at once.
  • Landing page — if the idea has multiple pages or A/B variants, you pick which one to send ads to.
  • Angle — if the idea is being tested with multiple positioning angles, each one usually gets its own campaign so you can compare them fairly.
  • Budget — how much per day, how much in total, when to start and stop.
  • Audience — countries, languages, interests, search themes.
  • Bidding strategy — Maximize Clicks for early learning, Maximize Conversions once you have signups, and so on.
Before any campaign goes live, the Growth-Manager will show you the full plan and wait for you to confirm. Nothing is launched without your OK.

Writing the ad creative

For each campaign, the Growth-Manager produces:

  • Headlines and descriptions — short text variants that get shuffled and tested by the ad platform.
  • Images — generated to match the page's style, then composed with overlay text, brand block, and call-to-action so they're ready to ship (not raw mockups).
  • For Meta, complete ad copies — primary text + headline pairs designed to go together.
  • For Google Performance Max, asset groups — bundles of headlines, descriptions, images, search themes, and audience signals.

You'll see every piece of creative in chat before it goes out. You can approve, ask for tweaks, or reject and ask for a different angle.

Launching and steering the campaigns

Once approved, the Growth-Manager:

  • Pushes campaigns, ad groups, and ads into Google Ads or Meta Ads.
  • Pauses campaigns that are learning poorly or burning budget.
  • Raises or lowers daily budgets as performance evolves.
  • Pauses individual keywords or ads that are bleeding spend.
  • Resumes things when conditions change.

You can ask in chat at any time: "pause the Family Vault Meta campaign", "double the budget on the Google one", "kill keyword X" — the Growth-Manager will confirm and act.

Pulling the numbers every day

The Growth-Manager syncs daily metrics from Google Ads, Meta Ads, and Google Analytics into MiniLab. For each campaign, ad, keyword, and the landing page itself, you get:

  • Impressions, clicks, click-through rate (CTR).
  • Spend, cost per click (CPC).
  • Conversions (signups, deposits), conversion rate (CVR), cost per acquisition (CPA).
  • Landing-page behavior — page views, scroll depth, CTA clicks, bounce rate.

These roll up by angle, so you can compare "Family Vault" vs. "Photographer's Vault" head to head.

Tracking presale orders and email leads

When a visitor pays a deposit, the order is captured automatically — amount, email, which campaign sent them, which ad they clicked. When someone just submits their email, that lead is captured too. Both lists live on the Orders page and in your lead exports.

Producing the weekly report

Once enough data is in, the Growth-Manager writes an analytics report: what worked, what didn't, which angle is winning, where the funnel is leaking, what to try next. You'll get this in chat and attached to the idea.

Calling Go, Iterate, or Kill

This is the whole point. The Growth-Manager will:

  • Watch CTR (does the ad grab attention?), CVR (does the page convince?), and presale count (do people actually pay?).
  • Compare results against the thresholds defined in the original plan.
  • Recommend one of three verdicts:
GoBuild the product. The numbers clear the bar.
IterateTweak the angle, the page, or the audience and run another week. There’s a signal, but it’s not strong enough yet.
KillStop spending. The market isn’t there, at least not at this price or with this pitch.

The Growth-Manager recommends. You approve, with a written reason, and the idea is marked Approved or Rejected on the Decisions page.

What the Growth-Manager won’t do (by design)

  • Edit the landing page itself. Copy changes, design changes, and payment-link setup all go back to the LP-Builder. The Growth-Manager doesn't touch the page, so it's always clear who owns what.
  • Issue refunds. Refunds happen in the Stripe dashboard manually. Ask ops if you need one.
  • Skip your confirmation. Every campaign launch, every budget change, every Go/Kill verdict needs your explicit OK in chat.

A typical week, in one paragraph

Monday morning you open MiniLab and pick an idea from the Ideas board. You ask the Researcher to brainstorm angles, pick one, and approve the research report. By Wednesday the LP-Builder has a landing page live at a real URL with tracking and a payment link. Thursday the Growth-Manager drafts ad copy, images, and a $20/day Google Search campaign — you approve it, it goes live in the afternoon. Friday through next Wednesday the Growth-Manager syncs metrics daily and posts a brief update each morning in chat. The following Thursday you get the analytics report: CTR 3.4%, CVR 6.1%, two presale deposits. The Growth-Manager recommends Iterate — same product, sharper headline. You approve the new headline. Two weeks later: Go.

Where to look in the dashboard

  • Ideas — Every idea you've started, what stage it's at, who's working on it.
  • Campaigns — Every ad campaign, its budget, status, and daily numbers.
  • Ad Materials — All headlines, descriptions, images, and ad copies — ready to ship or already live.
  • Orders — Every presale deposit, including the ad that drove it.
  • Analytics — Cross-cutting view of CTR, CVR, CPA, and presale count by angle.
  • Decisions — The Go / Iterate / Kill verdict log.
  • Chat panel — Where you actually talk to the Researcher, LP-Builder, and Growth-Manager.

Tips for new users

  • Talk in plain language. "Launch a $30/day Meta campaign for Family Vault, US audience" is enough. You don't need the platform's jargon.
  • Always read the confirmation message. Every team member shows you what it's about to do before doing it. Glance at budgets, audiences, and status changes before you say yes.
  • One angle per campaign. When testing multiple angles, give each its own campaign so you can compare them cleanly.
  • Wait a full week before deciding. Ad platforms need 3–7 days to learn. Numbers in the first two days will lie to you.
  • Trust the dashboard, not memory. When in doubt about status, owner, or past decisions, look at the Ideas board.

That's all you need to start. Open the Ideas page, pick a hypothesis, and go.