Restaurant group
Twelve sites, menus that change weekly, and a rule agreed on day one: no PDF ever appears on this website again.
The studio's redesign was warm and confident and completely undermined by the old site's menu strategy: a PDF per location, photographed from print artwork, updated whenever someone remembered. The group's twelve general managers were the real users, and none of them had ever been given a login worth having.
Where it started
Discovery mostly meant eating. We visited four of the twelve sites, watched what a Friday service does to the person running it, and pulled the old site's analytics, which showed 70% of menu views on phones and PDFs taking eleven seconds on venue wifi. The no-PDF rule was agreed in that first meeting and written into the quote, where it remains the only clause of the contract anyone at the group can recite.
We built the GMs' editing screens before the public menu templates, which is backwards by most schedules and was the right way round here. If the managers wouldn't use it, nothing downstream mattered. The first version failed the phone test inside a minute (too many fields above the fold) and the version that shipped is the second.
The build
Menus are entries. Sections, dishes, prices and dietary marks are fields, per location and per daypart. That's what makes them legible on a phone, findable by search engines, and updatable in the time it takes service to start. Locations carry structured opening hours that publish as schema.org data, so the hours on Google and the hours on the site come from the same field, which ends a whole genre of one-star review.
Ordering hands off to the group's existing platform by deep link, carrying the location with it, so the brand holds right up to the basket without us rebuilding a checkout that already worked.
The control panel got the most design attention of anything, because a GM on a phone between services needs to see their location, their menus and nothing else, with sections they can reorder with a thumb and a save button that does what it says. Craft's field layouts did most of it. Leaving things out did the rest.
One template layer, twelve venues
Craft's entry types did the structural work, but the speed came from what the twelve sites share: one layer of Twig, a settings entry per location for the things that differ, and nothing copied twelve times. A design amend lands everywhere at once instead of eleven times out of twelve, and it's why the thirteenth venue cost nothing in templates.
The menu templates are fluid from a phone at the pass up to the wall screens two venues run them on, with the type stepping up so a specials board reads from behind the bar. Motion barely features, on purpose. A GM with wet hands wants taps that respond now, so states change instantly, and the one animation, a section settling into place after a reorder, switches off under reduced motion. Dietary marks are icons with real text underneath rather than tooltips, because a screen reader can't hover and neither can a thumb.
Tested on phones, between services
Before launch we sat two general managers in front of a staging login and watched. Most field labels got renamed into kitchen language, a dietary-marks dropdown became tap-to-toggle icons, and the price field stopped accepting a pound sign because everyone kept typing one.
Forty menu changes a week would flatten a naive cache, so pages purge by location tag. A GM saving the lunch menu clears their own site's pages and nothing else, while the other eleven keep serving cached HTML. A menu save is live in a couple of seconds, which matters at 11:55.
Rollout was staged the way the group opens restaurants: softly. The flagship went first and ran for a fortnight with its GM as the pilot, producing three small amends and one convert who ended up doing our training for us. Then the other eleven venues launched in two batches over a single week, each GM loading their own menus as the training exercise. By the time the old site was switched off, every menu on the new one had been entered by the person who'd be maintaining it.
Our GMs update menus from the pass now. I honestly never thought I'd see it.
After launch
Menus now change about forty times a week across the group, and nobody emails the studio about any of it. When the group opened its thirteenth site this summer, the studio designed the launch page, a GM copied a sister venue's menus across in an afternoon, and we found out when the entry went live. The PDF rule has held.