validation·10 min read·
Smoke-test landing page for UK SaaS with Claude Code (2026)
Before the pre-sale, before the Mom Test calls, before the build - there is the landing page. Not a marketing page for a product you already have, a smoke test for a product you have not built. This is the UK indie hacker walkthrough: the five sections that actually validate, the Claude Code prompt that drafts the whole page in one session, the Tally form that captures intent in pounds, and the 100-50-3 rule that tells you whether to keep going or kill the idea.
Before the pre-sale, before the Mom Test calls, before the first line of build code - there is the page.
Not a marketing page for a product you already ship. A smoke test for a product you have not built yet. One URL, five sections, three signals to read. The whole point is to spend an afternoon discovering whether anyone will pay for the thing you are describing before you spend six weekends building it.
The good news: a UK indie hacker can stand the entire test up live in one Claude Code session and a free Vercel deploy. The bad news: most founders skip this step and go straight to the build, then wonder six weekends later why nobody signs up. Skip this step at your peril - the smoke-test page is the single cheapest piece of validation you can run.
Marketing page vs smoke-test page - why the distinction matters
A marketing landing page exists to convert visitors who already know what your product does. It has screenshots, feature lists, pricing tiers, testimonials, and a clear path into the app.
A smoke-test page exists to discover whether the product should be built at all. It has no screenshots (there is no product), no feature lists (you are testing a hypothesis), no pricing tiers (you are testing one price anchor), and no path into the app. The CTA leads to a founding-member form or a Stripe payment link, not a sign-up flow.
Confusing the two is the most common smoke-test mistake. You see a beautiful indie SaaS landing page on a launch thread, copy the structure, and end up building a marketing page for a product that does not exist. The visitor cannot tell what you are actually offering, so they bounce, and you learn nothing about demand. Strip everything that does not serve the single job of reading the signal.
The five sections that actually validate
You need exactly five sections. More is procrastination; fewer is not enough to read the signal.
1. Problem headline (one sentence)
Name the specific pain in the specific audience's language. Not "the modern SaaS workflow is broken" - that is anyone's headline. Something like "UK Etsy sellers spend 40 minutes a week on GPSR paperwork they could automate in 10."
The test of a good problem headline: someone with the problem reads it and feels seen; someone without it reads it and bounces immediately. That filtering is a feature - you want unqualified visitors to leave, because their behaviour will pollute the signal.
2. Three-line outcome (the product promise)
Three short lines, no more. What it does, who it is for, and the headline benefit in pounds or time.
- Line 1: what it does ("an AI assistant that drafts GPSR safety packs from your Etsy listings")
- Line 2: who it is for ("for UK Etsy and Amazon makers exporting to the EU")
- Line 3: the benefit ("ten-minute drafts, ten-pound monthly tier, no agency fees")
No screenshots. No mock-ups (unless they are wireframe-stylised and explicitly labelled as such). Mock-ups are the most common path to soft signals - visitors say they want the polished thing they see, then walk away when the bill arrives.
3. GBP price anchor
This is the section most founders flinch on. They do not want to show a price because they have not decided on one, or because they worry it will scare people away. Both impulses are wrong.
Showing the founding-member price is the single biggest signal multiplier on the page. A visitor who clicks 'join the founding members' on a page that already displays GBP 60 for lifetime access has told you something a thousand vague email captures cannot. Hide the price and you learn that some people are curious. Show the price and you learn that some people will pay.
The format that works:
Founding members: GBP 60 one-off, lifetime access
(intended retail: GBP 12/month - founders lock in lifetime, forever)
Limited to the first 20 sign-ups.
The discount versus intended retail is the bait. Lifetime price-lock is the gratitude play. The scarcity number is honest pressure.
4. Founding-member CTA
One button. Big, clear, single colour, single verb. "Become a founding member" or "Reserve my lifetime spot". Not "Learn more". Not "Get in touch".
The CTA leads to one of two places:
- A Tally form that captures name, email, and one or two intent questions (covered below), OR
- A Stripe payment link if you are pre-selling immediately (see Pre-sell your UK SaaS with Stripe payment links before you build)
If you are still validating, the Tally route comes first - you want to collect the intent signal and run the Mom Test interviews before opening the Stripe link to strangers.
5. Social-proof slot
Even an empty slot, labelled honestly, beats fake testimonials. "Founding members so far: 0 - be number one" reads better than three stock photos of people you have never met. As real sign-ups land, replace the counter with names + roles ("Sarah - solo Etsy seller, Bristol; James - finance partner, London").
Honesty is a brand position UK buyers respond to. The US-default "join 10,000 happy customers" pattern lifted from generic SaaS templates lands flat on a homemade-feeling indie page; people can smell the lift.
The Claude Code session that drafts the whole page
The five sections plus the bare-bones Next.js + Tailwind scaffold come out of a single Claude Code session. Open Claude Code in a fresh folder and run:
claude -p "Generate a single-page Next.js 16 + Tailwind landing page for a
UK SaaS smoke test. Five sections in this order:
1) Problem headline (one sentence, large type, navy on cream)
2) Three-line outcome paragraph
3) GBP price anchor block - founding-member price, intended retail, scarcity
4) Single CTA button linking to a Tally form URL placeholder
5) Social-proof counter slot, empty by default
Product details:
- Name: [PRODUCT_NAME]
- Problem: [PROBLEM_SENTENCE]
- Audience: [WHO_IT_IS_FOR]
- Founding price: GBP 60 one-off
- Intended retail: GBP 12/month
- Scarcity: first 20 founders
Style: deep navy on warm cream background, burnt orange accent for the CTA,
no images, no screenshots, no mock-ups. EN-UK spelling throughout.
Output the page.tsx file and the tailwind.config.ts. Nothing else."
Edit the four bracketed placeholders, paste the output into a new Next.js project (npx create-next-app@latest), and you have a working page in fifteen minutes. The whole stack is free: Next.js is open source, Vercel hosts it on the hobby tier at no cost, and a domain runs GBP 8 to GBP 15 a year.
If you want the copy sharper, run a second session against the draft:
claude -p "Sharpen the headline and the three-line outcome on this landing
page. The headline must filter out unqualified visitors - someone without
the problem should bounce immediately. The three lines must each carry one
specific noun (a tool name, a number in pounds, a UK regulator name). No
generic SaaS adjectives. EN-UK direct voice, no hype.
[PASTE PAGE.TSX HERE]"
The model is good at this. Two iterations and the page reads like a builder wrote it, not a marketer.
Capturing intent: Tally + the price-band question
The Tally form behind the CTA is where the real signal lives. Three fields, no more.
- Name (required)
- Email (required)
- The price-band question (required, single-choice radio): "What do you currently spend on this problem? a) Nothing, I tolerate it b) Under GBP 10/month in tools c) GBP 10-50/month in tools or time d) Over GBP 50/month or it costs me clients."
That third question is the one that earns its keep. Email captures from option (a) are weak signals - someone curious enough to fill a form but with no current spend rarely converts to paid. Captures from (c) and (d) are gold - someone already paying for an inferior workaround is the highest-intent buyer you can find. Sort your Mom Test interview queue by this answer.
Tally is free up to 100 submissions a month, hosted at tally.so, and integrates with Vercel-deployed pages via a simple URL embed or external link. No code, no integration, no backend.
The 100-50-3 rule - what good looks like
A smoke-test page on its own tells you nothing. The page plus 100 targeted visitors tells you everything you need to decide.
The yardstick:
- 100 targeted visitors in a week. Not 100 random impressions from an untargeted ad - 100 visitors who genuinely have the problem, sourced from one specific community where that audience lives (a subreddit, a LinkedIn niche, a founder WhatsApp group, a Discord, an Indie Hackers London Slack channel).
- 50 dwelling beyond 30 seconds. Vercel Analytics (built into the hobby tier, cookieless, free) gives you bounce rate and time on page. Below 50% engaged is a headline problem - your problem statement is not filtering or not landing.
- 3 high-intent actions. A founding-member click, a Tally form submission with option (c) or (d) on the price-band question, or, ideally, a pre-sale.
Hit those three numbers and the signal is real - move to the Mom Test interviews (Mom Test customer interview questions for UK indie hackers) and then the pre-sale.
Miss them, and the bottleneck is one of three things, in order:
- Audience source. The 100 visitors were not targeted. Pick a sharper community and re-run.
- Offer. The headline does not name the pain, or the price anchor is wrong. Edit the page.
- Page. The layout is confusing, the CTA is buried, or the copy hedges. Run a second Claude Code session.
Founders flip the order and blame the page first. Almost always the bottleneck is audience source. Sharper community, sharper signal.
Driving the 100 - where the targeted visitors come from
The page is half the test. The other half is where you point the URL. For a UK micro-SaaS, four sources tend to earn their keep:
- A relevant subreddit thread - not a self-promo post (auto-removed), but a genuinely helpful comment on a thread where someone has the problem, with the page URL as a follow-up in DMs to anyone who upvotes
- A LinkedIn UK post in your niche, written as a problem statement with the page link at the end
- A founder WhatsApp or Slack group - direct shares with permission, asking for honest feedback first and traffic second
- An Indie Hackers UK or London meetup - in person or in the Slack, framed as a feedback request
Avoid paid ads in week one. The signal is too noisy and the GBP per qualified visitor is too high for a smoke test. Save that for after you have validated and are scaling.
A note on what NOT to add to the smoke-test page
Three temptations to resist, every one of which dilutes the signal:
- A blog or content section. You are not testing whether people read your blog, you are testing whether they will pay. Strip it.
- A "request a demo" route. Demos imply a product. You do not have one. The button leads to a founding-member commitment, not a sales call.
- A pricing comparison table. Three tiers of nothing is still nothing. One price, one tier, one decision.
Each of these is a marketing-page idea bleeding into a smoke-test page. They make the page feel more legitimate and read the signal more weakly. Resist.
The hand-off into the rest of the validation sequence
The smoke-test page is the first of three artefacts in the validation kit:
- The smoke-test page (this post) - read the demand signal in one week
- The Mom Test interviews (next post) - find out which signal-givers will actually buy
- The Stripe pre-sale link (prior post) - convert the strongest interviews into the first three founding-member sales
Each step costs almost nothing. Each step is reversible. Each step is calibrated to filter you out fast if the demand is not there - which is the cheapest gift a validation framework can give a UK indie hacker. Six lost weekends building the wrong product is the only failure mode you cannot afford; an afternoon spent on a smoke-test page that produces a clear "no signal, kill it" answer is a win.
The whole sequence sits inside the broader weekend-vibe-coding pattern: spin up the page Friday evening, drive traffic Saturday and Sunday, read the signal Monday, run the calls Tuesday and Wednesday, fire the Stripe link Thursday. By the following Sunday you either have three founding-member sales and a green light to build, or a quiet week and a clean kill - and either outcome is worth the time. See the weekend vibe-coding playbook for UK builders for the broader cadence.
Ship the page, drive the 100, read the signal
The page is the cheapest, fastest piece of validation in the kit. Five sections, one Claude Code session, free Vercel hosting, a Tally form behind the CTA. The hard part is not the build - it is the discipline of using the page to read a real signal rather than to soothe yourself that you are working on the product.
Spin it up this weekend. Drive 100 targeted visitors. Read the three numbers. Whatever the answer, you will be a week closer to either the right product or the next idea.
Want the data that tells you which idea is worth building in the first place? IdeaStack publishes data-backed UK business idea reports - keyword volumes, competitor pricing, SERP analysis, and builder prompts - every week. Browse the latest free reports.
Frequently asked
What is the difference between a smoke-test page and a normal marketing landing page?
A marketing landing page sells a product you already ship - the headline, the screenshots, the testimonials, the pricing tiers all describe something a buyer can actually open and use. A smoke-test page sells a product you have not built yet. The job is not to convert, it is to discover whether anyone will pay for the thing you are describing before you spend six weekends building it. That single shift changes almost every design decision: you do not need feature lists, you need one clear outcome statement; you do not need pricing tiers, you need one price anchor in pounds; you do not need login or onboarding, you need one form that captures intent or a Stripe payment link that captures pounds. The smoke test exists to read a signal cheaply, not to look polished, and any element that does not contribute to reading that signal is noise.
How long should a UK indie hacker spend on the smoke-test page before declaring it ready?
A single Claude Code session and one focused afternoon - target the whole thing standing up live in three to six hours. The cap matters because the page is a probe, not a product, and over-investing in it is the most common smoke-test mistake. People polish the design for days, agonise over the copy, A/B test the headline before they have any traffic, and convince themselves they are working when really they are procrastinating on the harder job of finding real visitors. The discipline is to ship a credible-but-imperfect page fast, then spend the rest of the time driving 100 real visitors to it. If you finish the page in an afternoon and the signal is unclear after a week, the answer is more traffic and sharper interviews, not a prettier page.
Should the price on the smoke-test page be the real intended price, or a placeholder?
The real intended price, in GBP, anchored to a recurring number you would actually charge. Putting a placeholder defeats the whole experiment because price is one of the three signals you are trying to read - if a visitor sees no price they cannot tell you whether your number is too high or too low, and a clicked CTA gives you almost no information. Show the founding-member rate clearly (a number around GBP 39 to GBP 99 one-off, or a discounted monthly rate such as GBP 6 versus an intended GBP 12), explain why it is discounted, and let visitors self-select. A visitor who clicks 'join the founding members' on a page that already shows the price has told you something a thousand email captures cannot - that the number does not scare them off.
Do I need a domain name and proper hosting, or will a Vercel preview URL do?
A proper domain helps and is worth the small spend, but a Vercel preview URL is enough to start. Bounce rate and trust suffer on a vercel.app subdomain - visitors clocking the unfamiliar URL are more likely to leave before reading, and that pollutes your signal. A .co.uk or .io that matches the product name removes one excuse for them to bounce. Spend the GBP 8 to GBP 15 a year, point it at Vercel in five minutes via DNS, and your signal becomes cleaner. The exception is the very first day - if you are just sharing the page with a handful of warm contacts to sanity-check the copy, the preview URL is fine. Once you start running real traffic, get the domain.
How many visitors do I need before the signal is real?
About 100 targeted visitors, not 100 random ones - and the 'targeted' word is doing most of the work. A hundred visitors who genuinely have the problem you are solving (because they came from a relevant subreddit, a niche LinkedIn post, a founder DM, or a specific community thread) will tell you in a week what 10,000 untargeted impressions from a broad ad would not. The 100-50-3 rule is the rough yardstick: of 100 targeted visitors, you want at least 50 to dwell beyond 30 seconds (someone read the headline and stayed) and at least 3 high-intent actions - a founding-member click, a Tally form submission with the price-band question answered, or, best of all, a pre-sale. Hit those numbers and the signal is real. Miss them, and the problem is either the audience source, the offer, or the page itself - in that order.
