I design pages around how people actually decide, then build and ship them myself.

The problem

Poker players already have solvers that tell them the right play. What they don't have is a way to remember it at the table. But “study more” is a hard sell: it sounds like homework.

Decisions, and the psychology behind them

• Lead with a loss, not a feature. “That hand you lost last week. Did you study it, or forget it?” Every player has that hand. Loss aversion does the selling; the product is the answer.
• Show the real thing. The hero background cycles through real app screens, not illustrations. People trust what they can already see working.
• Remove the risk of trying. “No card needed” next to every button, a free plan with no expiry date, and an FAQ that answers the real fears (“What happens to my data if I don't pay? Nothing is deleted.”).
• Scarcity, but honest. A founder price for the first 100 subscriptions, stated with the exact number, so it is a real offer and not a fake countdown.

Built for speed and search

Plain HTML and CSS, no framework, WebP screenshots in desktop and mobile sizes, and cookie-free analytics. Separate Spanish and English pages with hreflang, a sitemap and Search Console.

What happened

512 ms

LCP, median (p75: 836 ms)

Green

all Core Web Vitals (LCP, INP, CLS)

72

visits in the first 7 days, before any promotion (some of them mine)

Honest note: the page is days old and hasn't been promoted yet, so there is no conversion rate worth quoting. I won't invent one. The first public launch posts go out this week, and I'm measuring visit → sign-up by language.

A checklist that ticks itself

• Three steps, not a tour. One tiny first action (“Review it: 2 minutes”) instead of ten tooltips. A small first win lowers the cost of starting.
• Step 3 is the habit. Spaced repetition only works if you come back, so “Come back tomorrow” is a step in its own right, with the reason next to it.
• It reads real data. Steps tick themselves when you actually do them; nobody has to tick a box. The progress bar uses the goal-gradient effect: people speed up as they near the end.
• It gets out of the way. It hides itself when done, and can be dismissed at any time.

A founder list without pressure

The sign-up form explains the founder price in one sentence and offers an unticked checkbox. No pre-ticked consent, no dark patterns: if someone joins the list, they meant it. It can be changed later in the account settings.

How it's measured

An SQL query reports how many new accounts reach each of the three steps and how many join the founder list. Only totals, no personal data. The numbers start this week.

What already works

• The BMI slider. It opens on a realistic example that shows “Likely eligible” before you've typed anything, so the visitor sees a yes before being asked to commit.
• Proof you can check. Every regulator badge links to its public register. That beats “trusted by” logos in a category full of fear of fake pills.
• Weekly pricing and a fair cookie choice. £/week makes the cost easy to picture, and Accept and Reject are given equal weight.

Honest limits

I have no access to Bolt's data, so these are hypotheses to test, not conclusions. Some may already have been tested, and some wording is limited by UK rules on advertising prescription-only medicines, so every copy change here would go through clinical and compliance review first.

What I'd want to know in week one: where the drop-off sits between the home page, the plans and question 16, and what new patients say about the second charge.