Morphos / Journal
Websites that have to last
A heritage site is a decade-long commitment. Designing for the funding cycle rather than the institution's life is how they end up rebuilt every three years.
Cultural and heritage organisations commission websites under different constraints from businesses: a grant funds the build, a committee approves the direction, and the institution expects the site to still be standing after the people who commissioned it have moved on. That changes what "good" means.
Four requirements that follow
None of them is about visual style, and all of them are decided early.
What institutional longevity requires
- Content that survives a redesign — text stored in a form that can be exported, not locked inside a page builder. Institutional sites outlive their tools.
- Accessibility as a deliverable, not an audit — for publicly funded bodies this is usually a legal requirement, and it is much cheaper built in than retrofitted.
- Multilingual where the audience genuinely spans languages — with proper hreflang, so each version ranks in its own market rather than competing.
- An exit that does not require the original studio — accounts owned by the institution, documentation of every service, and no dependency that cannot be handed over.
What changes when the client is an institution
| Decision | Commercial site | Institutional site |
|---|---|---|
| Motion | Whatever serves the brand | Restrained — accessibility and device range dominate |
| Language | Often one | Whatever the audience actually needs, with hreflang |
| Content storage | Wherever is convenient | Exportable, portable, documented |
| Handover | Accounts and a document | A full exit plan the institution can execute alone |
| Review cycle | Annually | On staff or committee change, so the site must explain itself |
The honest constraint
An institution that needs to last a decade should be suspicious of anything fashionable. That is not a call for dullness — it is a call for decisions that will still be defensible when the people who made them have left.
The language nobody plans for
Institutions plan for the languages their audience speaks and forget the ones their own content is already in. A collection catalogue carries object titles in the original language, inscriptions, place names, artist names that were never in the language of the site.
If the content model assumes one language per field, those records end up mangled: translated proper nouns, lost diacritics, romanised titles that no researcher will ever search for. The fix costs nothing at design time and is close to impossible later.
Keep original-language fields as their own fields. Display them alongside the translation rather than replacing it. It is a small decision made once, and it is the difference between a catalogue that scholars can use and one they cannot.