The five-second testThree cold readers see the first screen for five seconds and must answer: who is this for, what do they offer, what would you do next. All three, all correct, or the copy gets fixed.
One clear next step per viewportExactly one primary call to action visible at a time, labeled with what you get — never “Submit.”
An outcome headline, eight words or fewerOne H1 per page, and it names what you get — not what we do.
Proof earlySomething real and checkable — a published checklist, a real document, a measured number — within the first two screens.
Pricing one click away, at mostEvery page reaches the published price list in one click. Money pages carry pricing inline.
A visible pathService pages show how the engagement actually runs — mapped, built, grown — not vague promises of partnership.
Two typefaces, used preciselyThe type system is locked. No page invents its own typography.
Contrast verified with tools, not eyesEvery text-and-background pair is measured against the WCAG AA standard. Accessibility is checked per page, never sampled.
Motion that respects the readerAnimation stays inside a strict budget, and anyone who sets “reduce motion” gets the full page, still and complete.
Speed gates on real devicesLoad and layout-stability targets measured on production, on mobile. Local numbers never publish; the footer scores wire to real measurements only.
Works one-handed on a phoneNo horizontal scroll at small screens, thumb-reachable buttons, and a nav you can use while holding coffee.
Zero hype vocabularyAn automated sweep blocks the puffery words agencies lean on. If a sentence needs hype to work, the sentence gets rewritten.
Calm where trust is earnedPricing and intake pages run quieter by design: no countdowns, no fake scarcity, no exit popups — the footer publishes the count, and it’s zero.
Every claim real and sourcedEvery number traceable to a source. Empty proof slots stay honestly empty until real results exist — nothing is invented to fill a design.
Two of the fourteen can only be finished in the real world: the speed gate is measured on production at launch, and the five-second test needs real cold readers. Both are run — and their results published — rather than assumed.