One GLP-1 brand to three. Every stage of the funnel under test.
US GLP-1 telehealth brand, seven-figure annual revenue, paid traffic from Meta, TikTok and search. We came in as the growth and build team for the whole funnel, from ad click to refill.
60
Landing page variations
20
Checkout variations
10
Price points tested live
3
Brands on one system
15
Products launched
Under NDA. Brand and revenue stay out. Scope and scale do not.
Where the money was leaving
Intake, checkout, upsells, portal and reporting all sit on the client's own stack. Three brands share it, so anything built once ships to all three.
Every ad landed on the same page. One headline, one price, one checkout, one product.
There was no second version of anything, so there was no way to learn what worked, and every dollar of traffic bought the same result as yesterday.
Growth meant buying more of the same traffic.
Shipped, then tested
60 landing pages
Each one tied to a traffic source and an offer, all reporting to one number.
10 price points
Tested on live traffic with monthly and multi-month plans and partner offers.
20 checkouts
Each variation ran against the one before it. The winner stayed.
Upsell pages
Offers after checkout on the saved card. Offers swap without a rebuild.
3 brands, 15 products
One system, one team. A product launch is a config, not a project.
Custom A/B system
Conversion rate per stage and a before/after on every launch, read off payments that stuck.
Admin analytics
Revenue, promos and product mix in one view for every launch.
Weekly releases
Landing pages, intakes and checkouts each keep a live variation at all times.
What changed
The brand went from one page, one price and one product to a funnel that runs a test at every stage and reads the answer in revenue.
Three brands and 15 products run on one system with one team.
Landing pages, intakes and checkouts each keep a live variation. The winner stays, the next one starts.
Duration. Ongoing. Weekly releases since the first brand went live.
60
Landing page variations
20
Checkout variations
Asked on the scope call
How do you move that fast?+
AI does the typing, a senior engineer decides what gets typed and reads every line before it ships. In the time a normal build puts one landing page live and tests it, you get three to five whole themes running against each other, so the same month of ad spend buys several shots at a winner instead of one. Nothing touching patient data ever runs unreviewed.
How do you know a winner actually won?+
The count that decides is the charge Stripe settled, not a checkout page view the platform's report calls a conversion. A variation stays up until it has enough orders to separate from the other one. If it never separates, we say so and delete it.
How many landing pages do we end up with?+
As many as you have traffic worth separating, and maintaining them is one job, not fifty: every page renders off one brand config, so a price or an offer changes once and lands everywhere. A source that cannot pay for its own page does not get one.
