Tool 08 · $27
Get it for $27
One page that tests itself and sends traffic to the winner.
Build a landing page that serves several versions, learns which one sells, and shifts visitors toward it automatically. Tied to real purchases in Stripe, not clicks. Runs on Cloudflare Workers.
We ran a landing page test with a well-known tool. It declared variant B the winner after a week, on click-through to checkout. We switched everything to B.
Two weeks later the Stripe table (Tool 01) showed revenue per visitor had dropped. B got more people to the checkout page. Fewer of them paid. The tool had optimised for the wrong event because it could not see the right one.
So we built a page that could. A small worker on Cloudflare serves the variants, Stripe tells it which variant produced each payment, and it shifts traffic toward the one that sells. No dashboard, no licence, no click-through proxy.
We built the first version ourselves. It has run every test on our site since.
The starter worker, config and an example variants file are what this starter gives you.
Testing tools score the event they can see on your page: a click, a form, a scroll. The event that matters happens on Stripe, after the page, and most tools never hear about it.
A page tied to purchases is slower to decide, because sales are rarer than clicks. It is also right. And a page that shifts traffic as it learns wastes less on the loser while it decides.
The worker is small on purpose. Variants in one file, weights in one config, purchases reported by a Stripe webhook.
One worker in front of your page. Visitors get a variant, Stripe reports the sale with the variant on it, the weights move. You read the config when you want to know who is winning.
The starter worker is complete. You add your variants to the file, set the config, and deploy with Wrangler. Claude Code walks through the Stripe webhook from the guide.
Worker, config and variants file are documented so a developer can deploy it in an hour and you can edit variants afterwards without them.
Plus the guide itself: 5 pages on what to vary first, how long to wait before trusting a weight, and the one Stripe setting that makes the webhook fire.
You write a new headline or offer section as a variant and add it to the file. The worker starts showing it to a slice of traffic. If it sells, it earns more traffic. If not, it fades.
The Monday review (Tool 06) gets a new row: which page variant each payment came from. Ads and pages are now tested by the same judge.
This is step 12 of the sixty: the landing page as a system that learns, instead of a file you argue about.
Most agencies run the ads and rent everything else. Sixty Steps builds all sixty as real development on the client's own domain, then reports every week against what settled in Stripe.
These ten tools are the steps you can set up yourself in an afternoon. This one is step 12: a landing page that tests itself against purchases.
I built analytics models for banks and miners before this. The habit that stuck is not trusting a dashboard that grades its own work. Every tool on this page exists because we needed it on our own account first.
Dear reader,
If you have read this far, you have probably already decided whether $27 is a risk worth taking. It is less than one month of a testing tool. So the rest of this page is not about the starter. It is about who is selling it to you, and why we sell the steps one at a time.
We are an analytics firm that got into Meta ads by accident, found that nobody was measuring them properly, and started fixing that one step at a time. This is one of those steps.
My name is Ryan Richardson. For 6 years I built analytics models for banks and mining companies through a firm called Onwards Analytics. The brief was always the same: two systems that should agree do not, and someone senior needs to know which one to trust before the Monday meeting.
I had run Meta ads on and off for about five years. Once we ran them for our own offers, I was staring at the same problem I had been paid to solve for years, except now the two systems were Ads Manager and Stripe, and the person who needed to know was me.
So I did what I had always done. I traced every step between a stranger seeing an ad and a payment settling in Stripe, and wrote each one down. There were sixty. That list became a book called The Book Is the Funnel, and then a business called Sixty Steps that builds those steps for other people.
Three rules. We apply them to our own account before anyone else's.
Stripe is the scoreboard. Every report, decision and invoice is measured against money that settled. Platform dashboards are inputs. If a number on one of our pages cannot be traced to Stripe, it is marked as reasoning.
Everything lives on your domain, in your name. Pages, checkout, email and tracking are real code you own. There is no page-builder licence, and nothing stops working if you stop working with us.
We use it before we sell it. Every test on our own site runs through this worker. The example variants are ours, with copy shortened.
That is also why the starter is $27. We are not trying to make money on it. We are trying to get your page judged by the same thing that judges your ads, because a page optimised for clicks quietly costs you sales.
Each tool is one step from the book, packaged as the exact files we use: a template, a sheet, a script outline or a prompt, plus a short guide. They cost between $4.95 and $27. There is no software to subscribe to and nothing that phones home.
Three of them measure. One tags each Stripe payment with its ad. A weekly review joins those tags to Meta results and tells you what to make next. A six-lever diagnostic scores your whole funnel and finds the weak one.
Two of them screen. A three-gate scoring sheet judges an ad before it costs money. A blind panel of AI buyers scores the same ad independently and reaches a consensus.
Two of them research. One maps how your top 10 rivals look and where the empty space is. One pulls any competitor's ads and ranks them by how long they have run.
Three of them build. 40 ads uploaded from one spreadsheet. A landing page that tests its own variants against purchases. A pipeline that produces 20 scored offers in an afternoon.
None of them replaces the other fifty steps. Each one makes one step faster, or makes one number honest. This page sells the self-testing page, because the landing page is where most of the sixty steps happen and almost nobody tests it against money.
The files arrive by email within a few minutes. You add two variants of one section, set the config, deploy the worker and point the Stripe webhook at it. From then on the page tests itself.
Then nothing happens. There is no onboarding sequence, no upsell call, no countdown. If you want the other tools you will find them on the site. If you want us to build the rest of the sixty steps for you, the site explains how that works and what it costs, and it starts with a free teardown that might say no.
If the starter does not do what this page says, reply to the receipt email and say so, any time in the first 60 days. You will get your $27 back and I will want to know what went wrong.
P.S. If you are testing a page by clicks today, pull revenue per visitor for each variant from Stripe before you switch. It is the cheapest way to find out whether your tool has been lying.
