Website design for museums and cultural institutions

Built for ten years, not two

Most websites are designed around the person who commissioned them. For an institution that is the wrong frame entirely. The director who signs off will have moved on before the site is replaced; the person maintaining it may not have been hired yet.

So the question is not what the site should look like this year. It is what structure will still make sense when the next exhibition programme, the next funding round and the next content editor arrive.

In practice that means: content types that match how the institution actually thinks — exhibitions, events, collection objects, people, press — rather than free-form pages that drift apart within a year. Navigation that does not have to be redesigned every time the programme changes. And a build that a competent editor can run without calling a developer.

Programme versus collection

These are two different websites sharing one address, and confusing them is the most common structural mistake.

The programme is time-sensitive: what is on now, what opens next, how to book, whether the café is open. It changes weekly and it is what ninety per cent of visitors come for.

The collection is permanent and slow: objects, provenance, context, scholarship. It is what the institution is for, and it is what search engines and researchers actually index.

A site that merges them into one feed serves neither. We keep them as separate paths that cross-link, so a visitor looking for opening hours is never asked to understand a catalogue record, and a researcher is never pushed into a booking funnel.

Multilingual by default

Cultural institutions are usually multilingual before they admit it — a bilingual city, a touring programme, a collection with captions in three languages, a funder in another country.

Retrofitting translation onto a monolingual build is expensive and always looks it: parallel trees that drift, missing pages, a language switcher that leads to a 404. Building the structure with language as a first-class dimension costs little more at the start and saves the entire problem later.

We build the language layer in from the first page, with proper hreflang, translated navigation and a fallback that degrades gracefully when a translation is missing rather than dropping the visitor on an empty page.

Accessibility is not a separate project

For a publicly funded institution, an inaccessible site is a straightforward failure of the brief — and in most jurisdictions a legal one. It is also the single area where retrofitting is most expensive, because it touches structure rather than surface.

We build to WCAG 2.2 AA as the baseline rather than as an audit to pass later: real heading hierarchies, keyboard-operable everything, focus that is visible, contrast that holds on a projector as well as a phone, and motion that stops when the visitor asks it to.

The same work makes the site faster and more legible for everyone, which is the part that rarely gets said out loud.

The press and the researcher are visitors too

Two audiences are almost always forgotten, and both are disproportionately valuable.

The journalist needs a press room that works at eleven at night before a preview: images at print resolution, captions that name what is in the picture, a contact who will answer, and a date on the release so it is obvious whether it is current. A press page that is four years old tells a journalist the institution has stopped doing things.

The researcher needs citations that resolve. An object page whose URL changes when the site is redesigned breaks every reference to it in every paper that ever linked to it. Stable identifiers are cheap to build and impossible to retrofit.

• A press room with downloadable images, credits and a dated release
• Stable, human-readable URLs for every object and exhibition
• A visible last-updated date on time-sensitive pages

What we build

Working with institutions

Procurement, brand guidelines, an existing CMS the IT department will not release, a photography archive with unclear rights — institutional projects carry constraints that commercial ones do not, and pretending otherwise is how projects slip.

We scope around them in the first conversation rather than discovering them in week six: what must stay, what can move, who signs off, and which of the existing systems is genuinely load-bearing.

The build is then staged so that something demonstrable exists early and can be shown to a board, a funder or a focus group before the whole thing is finished.