Website builders · checked 2026-10-19

Can I vibecode Unbounce?

KINDA· weekend project

For one person with an agent, the core is small: pages served from a static host, a form that stores or forwards leads, and a split that sends visitors to version A or B and counts conversions. The weekend goes into the experiment part you would trust ad money to: visitors who always see the same version, bot filtering, a significance test that doesn't invite peeking, and text filled from ad URL parameters. What stays behind is the editor your marketers use without you, popups and sticky bars, a thousand integrations and the AI traffic routing.

$29/moprice

$348you'd save a year

weekendbuild time

—people replaced it

Before you start

What you need on hand.

DifficultyIntermediate · Some system setup or keys.

TimeOne to two hours for the first version; DNS and a server take longer

Claude Code, Codex or CursorAny of them can run the prompt.
Node.js 22better-sqlite3 compiles a native module, so build tools must be installed.
A small serverA $5 VPS or any host that runs Node; ad traffic must be able to reach it.
A domainYou point a subdomain such as go.example.com at the server.
SMTP credentialsOptional: the leads page works without email.

Before you buildUnbounce has no free plan, only a 14-day trial, so there is nothing cheaper to try first. If your pages get under 500 visitors a month, Starter at $29 may already be enough.

1. Install one agent: Claude Code, Codex, Cursor.
2. Make an empty folder and open it in the agent.
3. Paste the prompt below, press enter, and answer its questions as it builds.

The prompt

Build me a personal landing-page server with A/B testing to replace Unbounce. Requirements:

- Node.js 22 + TypeScript: Fastify serving static HTML from ./pages/<page>/<variant>.html.
  `npm start` runs http://localhost:4860, with /admin behind one password from .env. I edit
  the pages as plain HTML with an agent; there is no visual editor.
- Variants: each page folder holds a.html, b.html and more; a manifest.json sets the split
  percentages, the conversion goal (a form submit or a click on [data-goal]) and an optional
  end date.
- Assignment: the first visit picks a variant by weight and keeps it in a first-party cookie
  for 90 days, so a visitor always sees the same version. Skip bots (`isbot`) in all counts.
- Storage: better-sqlite3 in ./data/pages.db: visitors (hashed cookie id, variant, utm
  fields, first_seen), conversions, and leads (form fields as JSON, source URL).
- Forms: any <form data-lead> posts to /lead; validate with zod, include a honeypot field and
  a per-IP rate limit (`@fastify/rate-limit`), store the lead, then send it by email through
  nodemailer SMTP from .env and, if set, as JSON to WEBHOOK_URL.
- Dynamic text: elements marked data-utm="utm_term" are filled from the query string at
  render time, with a default when the parameter is missing, HTML-escaped.
- Results in /admin: per page and variant, visitors, conversions, rate, and a two-proportion
  z-test at 95%. Show "not enough data" until each variant has 100 visitors and 7 days have
  passed, so nobody calls a winner early.
- Leads page: search, CSV export and delete. Single user. No accounts, no telemetry;
  cookies are first-party only.
- Out of scope: a visual editor, popups and sticky bars, bandit algorithms and AI routing,
  payments.
- README: serving it behind Caddy on a VPS with a real domain, a sample cookie-notice
  sentence, why low traffic needs long tests, and a warning that a stats number is not a
  promise.

Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.

Share on X

Max MiskowSponsoredAI agents, automations and MVPs, built for your company.Visit site ›

Then keep going

Next prompts, same folder.

The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Unbounce.

Add popups and sticky bars to the landing-page server in this folder. Requirements:

- An overlays/ folder with one JSON file per overlay (type popup or bar, HTML, trigger,
  pages it applies to) served by a /overlay.js script that pages include with one tag.
- Triggers: after N seconds, after scrolling X percent, or when the pointer leaves the top
  of the window on desktop. A cookie caps each overlay at once per visitor per 7 days.
- Overlay forms post to /lead like page forms and count as conversions for the page.
- Closing with Escape, a visible close button, and no overlay on the first 3 seconds.
- Update the README.

Add multi-step forms to the landing-page server in this folder. Requirements:

- A <form data-steps> splits into steps by <fieldset>, with next and back buttons and a
  progress bar, in vanilla JS with no build step.
- After each step, POST the answers so far to /lead/partial with a random lead id kept in
  sessionStorage; store them as a partial lead and mark it complete on the last step.
- A leads filter for complete and partial, with the last step reached for partials.
- Escape every stored value when it is displayed, and delete partials older than 30 days.
- Update the README with a note about collecting data before consent.

Make lead delivery reliable in the landing-page server in this folder. Requirements:

- Write every lead to an outbox table before replying to the visitor, then deliver to
  WEBHOOK_URL in the background.
- Retry failures with exponential backoff up to 8 times; sign each request with an
  HMAC-SHA256 header over the raw body using WEBHOOK_SECRET.
- An admin page listing pending, delivered and failed deliveries, with a retry button.
- Email me a summary when a delivery fails for good.
- Update the README with a sample receiver.

Add a weekly results email to the landing-page server in this folder. Requirements:

- Every Monday at 08:00 in the time zone from .env, email one summary: per running page, the
  visitors, conversions and rate of each variant and the days left before the minimum test
  length is reached.
- Declare a winner only when the existing 95% test passes and both minimums are met;
  otherwise write "keep running".
- Show new leads count for the week and the top three sources by utm_source.
- Add a --send-now flag for testing. Update the README.

What you lose

A drag-and-drop editor teammates can use without touching code
Popups, sticky bars and triggers such as exit intent or scheduling
1000+ integrations; you get a webhook and email instead
AI traffic optimization, benchmarks and audience insights
Hosting built for ad bursts, and a support team to call

Why people still pay

Marketers pay to publish and test a page the same afternoon without asking a developer, and to know it will stay up when an ad budget sends a crowd. The editor, the hosting, the integrations and the experiment reports arrive as one tool, with hosting included in every plan.

Moatexecution polish integrations infrastructure scale

Do the math

Unbounce or your own copy?

Over

Your usage

Count my time12h ×$/h

Unbounce, 1 yr$348

Your own copy, 1 yr$44running + build session

You keep$305pays for itself in the first month

Estimate · Runs on a $5 VPS or a free tier of a Node host; SMTP at lead volumes fits free tiers. The build is token usage on your agent plan.

Leaving Unbounce

Take your data, then cancel.

1 · Take your data out of Unbounce

1Leads: open a page's Leads tab, filter by date range and download the CSV; the All Pages screen can download leads by date range per client.
2Pages: download them as .unbounce files from the page options. That format only re-imports into Unbounce, and only Classic Builder pages can be downloaded, not Smart Builder pages.
3We found no HTML export: plan to rebuild each page elsewhere and keep the text, images and form field names.
4Popups, sticky bars and conversion statistics have no export in what we read: screenshot the numbers you need.

2 · Cancel Unbounce

1Click your avatar at the top right and choose Account Management, then Subscriptions.
2Click Cancel My Plan, review the clients, popups, sticky bars and features that go away at renewal, and click Continue.
3Fill in the feedback form and click Cancel My Plan. Only owners and administrators see the button.
4You keep paid features until the end of the billing cycle, then drop to a free plan.

The help center says the account and lead data may be deleted after 90 days if you don't upgrade again, so export leads first.
Cancelling keeps the account; deleting it is a separate request to support and can't be undone.
The pricing page and the cancel article don't mention refunds: assume none for the current period.
Keep DNS pointing at Unbounce until the new pages are live, then switch the record and watch the form submissions arrive.

Checked 2026-10-19 against Unbounce's docs: 1, 2, 3, 4

The escape hatch

Don't feel like building it? These are already free.

WebstudioOpen sourceAGPL-3.0 visual website builder, billed as a Webflow alternative that can be hosted anywhere; its animation package is proprietary.★ 9.1k · active Oct 2026 · self-host, real ops
WebVisit ↗MauticOpen sourceGPL-3.0 marketing automation with landing pages, forms and email campaigns; a PHP app you host yourself.★ 11k · active Oct 2026 · docker to self-host
WebVisit ↗GrowthBookOpen sourceOpen-source A/B testing and feature flags with test statistics built in; MIT core, some enterprise parts are paid. It runs tests, it doesn't build pages.★ 8.5k · active Oct 2026 · docker to self-host
WebVisit ↗

All 5 free alternatives to Unbounce ›

The numbers

Unbounce pricing

Plan	Monthly	Annual, per month	What you get
Starter	$29	$22	5 landing pages, up to 500 visitors a month, 1 user, 1 root domain, drag-and-drop builder, lead forms, free hosting.
Build	$99	$74	Unlimited pages, up to 20k visitors, 1 user, 1 root domain, popups and sticky bars, multi-step forms, custom code, scripts and pixels, AI copywriting, 1000+ integrations.
Experiment	$149	$112	Unlimited A/B tests and variants, up to 30k visitors, 3 users, 2 root domains, manual traffic split, confidence intervals, dynamic text replacement, conversion reporting.
Optimize	$249	$187	Up to 50k visitors, 5 users, 3 root domains, AI traffic optimization, scheduling, advanced triggers, industry benchmarking, audience insights.
Concierge and Agency	custom	—	Concierge: from 100,000 visitors, 15 users, 5 domains, implementation help and a customer success manager. Agency: from 50,000 visitors, 5 users, 10 domains, listing in the partner directory.

Free tier: Trial only: one user, no custom domains and some traffic limits (not itemized on the page).
Billing: Monthly or yearly in USD; yearly saves 25%. Upgrades take effect at once; downgrades from Build, Experiment or Optimize wait for the end of the billing cycle.
Hidden costs: A visitor is counted once a month per person. Going over the visitor or conversion cap doesn't stop pages: Unbounce adds a one-time overage fee based on the extra usage. Starter's 500 visitors and missing scripts and pixels mean most ad campaigns start at Build ($99), and A/B testing starts at Experiment ($149). Onboarding, page migration, redesign and consulting are paid professional services.
Source: unbounce.com

Also on the list

Leadpages$29/mo · Website buildersYES

Carrd$1.58/mo · Website buildersYES

Framer$15/mo · Website buildersKINDA

Questions

Unbounce, briefly

Not answered here? Tell us and we'll add it.

Ask a question ›

No free plan: a 14-day trial with no card, limited to one user, no custom domains and some traffic limits.

Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: A drag-and-drop editor teammates can use without touching code, Popups, sticky bars and triggers such as exit intent or scheduling.

About $29 a month on the Starter plan, $348 a year (checked 2026-10-19).

A drag-and-drop editor teammates can use without touching code; Popups, sticky bars and triggers such as exit intent or scheduling; 1000+ integrations; you get a webhook and email instead; AI traffic optimization, benchmarks and audience insights; Hosting built for ad bursts, and a support team to call. If any of those matter to you, keep paying.