Craft Craft that ships. that ships.

Before anything was designed, I sat down with four serious lifters. The same answers kept coming back: they already track their training, they judge how hard a set was by the reps left in the tank, and every app they'd tried buried that under gamification. One ask kept surfacing, too. They wanted to watch their strength hold, or slip, through a cut. That locked the core: RIR at the centre of logging, and strength over time shown honestly.

Once early users were in a private beta, a feature-prioritisation survey decided what got built next. The winners became the app's progress views, which read training from several angles: strength trend per lift, personal records, and muscle-group balance over time.

The name does a lot of the work. A ram's horn grows a ring for every year it lives, and training is the same: every session is one more notch. So the mascot is a ram (I call him The Keeper) and the logo is a notched horn. The look is a system I call Dark Ledger: warm near-black, monospace numerals, clean prose, depth from soft frames over a faint ruled background. One rule holds it together: colour always has to mean something.

The whole app runs on one move: log a set, see what it was worth. Because it knows how hard each set felt, it can tell a real personal best from a lucky day.

I designed everything and directed AI dev tools (Claude Code, Cursor) through the build. The design lives in the code, so there's no hand-off where details die between a Figma file and the real app.

The hard part was deleting. I built a streak counter, liked it, and killed it, because a streak is the exact game Notch is trying to end. The exercise library went from 873 to 438, keeping the lifts a serious program actually uses. Features I was excited about moved to after launch. Anything that couldn't justify its place didn't get one.

The job was to win over cold traffic: people landing from a search or an ad who'd never heard of the brand. The old homepage did the opposite. It threw sixty-plus challenges at you in one dense grid, every card just saying “LEARN MORE.” No pitch, no “how it works”, nothing telling you where to look first. And the things that actually sell the product, like the 450k-member community, the medal you earn, and the licensed challenges (Lord of the Rings, Harry Potter), were buried or missing. Show someone everything and they pick nothing.

Before redrawing anything, I went to the analytics, and they told a clear story. Most first-timers arrived cold from a search or an ad, met a wall of sixty-plus challenges, and left without picking one. They weren't short on options, they were short on a path. So I designed for one person: a casual walker who wants a reason to move and a medal to show for it. Everything after this is for her.

I stopped treating it like a catalogue and started treating it like a path. Where does a first-timer get stuck, and what do they need to see next? Four things guided the redesign.

The new homepage actually walks you through it. A clear pitch at the top, then a five-step “how it works”, then the proof: 450k members, an 87.93% completion rate, and logos from Forbes, Wired and Runner's World. After that, a handful of popular and licensed challenges, reviews, a money-back guarantee, and the FAQ. The full catalogue is still there, one click away under “View all challenges.” It just stopped being the first wall you hit.

The catalogue got the same thinking. The old All-Challenges page was a flat grid of medal thumbnails on white. It sold the medal; the feeling of earning one never came through, and your eye had nowhere to land. The redesign leads with photos of real people holding their medals, a line reminding you that over a million others have done this, cleaner cards, and a few nudges like bundles, discounts and “new” tags. Now it reads like a trip worth taking.

While the pages shipped, I designed the company design system on the shadcn framework: tokens and components in Figma that matched what developers actually built. One source of truth, used by the dev, campaign and creative teams. It kept every page after this one consistent, and it's part of why the pricing work below could move fast.

Three plans, three tables, three buttons. Every table pitched its own call to action, so the page competed with itself. People had to work out the differences between plans before they could act, at exactly the point in the funnel where hesitation costs the most.

Compress the choice. One selector holds the plans, picking one is a tap, and a single call to action sits under it. The comparison still exists, it just stopped being a layout. The page asks one question at a time: which plan, then go.

It took four or five A/B rounds to get right. The selector, the copy around it and the defaults were adjusted round by round, each version tested against the last one live. The design that shipped is the one the numbers picked.

Conversion through the pricing step rose 11%. The lesson travelled well beyond this page: pages stop competing with themselves when they ask one decision at a time.