FOUNDING STORES
no card · onboarding in waves
- ▸ Full product — nothing gated
- ▸ Direct line to the founders (support@ghostbot.ai)
- ▸ First notice + preferred terms when paid plans arrive
[ 01 ] A/B TESTING, GHOSTWRITTEN
GhostBot reads your product pages, drafts the variants, and measures every test in revenue — not clicks. You approve what ships. Losers pause themselves.
FREE DURING EARLY ACCESS · NO CARD · US SHOPIFY STORES FIRST
Animated demo of the GhostBot pipeline: the AI drafts a test variant, storefront traffic splits, purchases are attributed to variants, and a statistically significant winner is declared. All numbers simulated.
[ 02 ] THE PROBLEM
Every store owner knows A/B testing prints money. Almost nobody runs it. The tools hand you an empty editor and wish you luck — you invent the hypotheses, write the variants, and squint at significance calculators.
So changes ship on gut feel, and nobody ever finds out what that redesign actually cost. GhostBot exists to do the part nobody has time for: it writes the tests, and it does the math.
[ 03 ] WHAT'S IN THE BOX
DRAFTS · YOU APPROVE
GhostBot reads your product pages and drafts test variants worth running — headlines, CTAs, emphasis — grounded in what's actually on the page, not a generic template.
GUARDRAILED
Nothing touches your storefront without your approval, and guardrails block unsafe changes outright: no rogue scripts, no off-site links, no broken layouts.
MEASURED IN $
Orders are attributed to the variant that produced them, server-side. Winners are declared on real revenue with statistical stop conditions — not on click-throughs.
CONSENT-AWARE
Respects your store's consent settings automatically via Shopify's Customer Privacy API. Deletion requests cascade through analytics. One ~3KB async script.
[ 04 ] HOW IT WORKS
Install the Shopify app and flip one toggle in your theme editor. A setup checklist walks you through it — no code, no theme surgery.
GhostBot drafts experiments for your product pages. You see the exact change, side by side, and approve in one click.
Traffic splits automatically. When a result is statistically real you get a verdict in revenue — and losing variants pause themselves.
[ 05 ] THE MATH
Every visitor's variant assignment is recorded from first paint — you can't judge a test on traffic it never saw.
Add-to-carts and checkouts are tied back to the assignment that produced them, per visitor, not per pageview.
Orders are attributed server-side from Shopify order webhooks — real dollars per variant, in your currency, not projections.
Significance stop conditions decide when a test is done. Under-performing variants are paused automatically.
No p-hacking dashboards. No “trending toward significance.” A test ends when the math says the result is real — and the verdict is denominated in revenue.
[ 06 ] EARLY ACCESS
FOUNDING STORES
no card · onboarding in waves
THE FINE PRINT, UP FRONT
GhostBot is in early access, onboarding US Shopify stores in waves. Founding stores get the full product, a direct line to the people building it, and first notice (plus preferred terms) when paid plans arrive.
[ 07 ] FAQ
Only what you approve: targeted text and emphasis changes to your existing pages, applied at render time by a small script. Guardrails reject unsafe edits (scripts, off-site links, layout-breaking styles) before they can even reach your review queue. Turn the embed off and your store is untouched.
The storefront script is roughly 3KB compressed and loads asynchronously — it doesn't block your page from rendering. No heavy replay recording, no bundle of trackers.
The script checks Shopify's Customer Privacy API and only collects analytics when consent allows it — if a visitor declines, tracking stays off and stored identifiers are wiped. Store and customer deletion requests from Shopify cascade through our analytics automatically.
The full product, free, while we onboard founding stores. Paid plans come later — early-access stores get advance notice and preferred terms before anything changes. No card required now.
No. GhostBot runs the split, does the math, and only calls a winner when the result is statistically real. You see plain-language verdicts with the revenue impact — not p-values to interpret.
US Shopify stores first — that's where we're focused during early access. EU/UK support is planned; join the waitlist and we'll tell you the moment your region opens.
[ 08 ] EARLY ACCESS