A research centre website has to work for readers who want different things and arrive at the same homepage: a funder checking credibility, a postdoc looking for a position, an industry partner wondering whether you can test something, a journalist on deadline. Institutional sites are often built as though those readers are one reader, which serves none of them well. Here is how we approach it.
Start by writing down who actually visits
Before anyone talks about design, sit down with two or three colleagues and list every group that comes to your site. For SSPC, the pharmaceutical research centre we worked with, that meant funding agencies, academic collaborators, pharma companies looking for research partners, prospective PhD students, and the centre's own researchers, who mainly needed to find their publications and update their profiles.
Then write one sentence for each group describing what a successful visit looks like, specific rather than a phrase like "learn about our research". Something closer to "finds the named contact for industry collaboration and emails them". If you cannot write that sentence, either the audience is not real or you have not decided what you want from them, and both are worth knowing before you spend money.
The exercise takes an afternoon, costs nothing, and usually surfaces something uncomfortable: the homepage serves your own staff and board better than it serves anyone else, because they are the people who reviewed the content. Funders and partners rarely get a say in what a homepage says, so nobody speaks for them when it is signed off.
One site, several front doors
The instinct with several audiences is a homepage that addresses all of them at once, usually with four or five equal boxes, and it reads as a menu rather than a position. The alternative we use states plainly what the organisation does and who it is for, then routes quickly into sections written for one group each.
AIESEP is an international association for physical education research, and its members, its conference attendees and its journal readers want different things from the same site. It works because those paths separate early rather than blending into a general purpose middle. Somebody arriving to renew membership does not have to read the research agenda first.
There is a governance benefit as well, because a section with a defined audience also has a defined owner. Arguments about whether something belongs on the homepage turn into arguments about which section it belongs in, and those are much easier to settle. We have sat through enough content meetings to know this saves weeks.
The publication archive is harder than it looks
Every research organisation we work with underestimates the publication archive, which looks like a list and behaves like a structured database, with author relationships, funding acknowledgements, journal metadata, open access status, and filters that researchers use constantly and nobody else touches. If it is wrong, your most cited work becomes the hardest thing on the site to find.
The questions that decide it are dull ones, and worth asking early: where does the data come from, and can it be pulled from an existing repository rather than typed in twice? Who adds a new paper, and how long does it take them? Does an author page update itself when a publication lands, or does somebody have to remember?
If you do nothing else this month, export your publication list to a spreadsheet and check it against the last two years of actual output. The gap tells you whether the problem is the website or the process behind it, and quite often it is the process. Our Irish practice, Brainstorm, established in Limerick in 2010, has written about how these content structures get built for universities.
Build for the person editing it in 2032
Institutional sites have unusually long lives, which changes what a good build looks like. A commercial site might be replaced in four years, while a university department site often runs for eight or ten, edited throughout by staff who were not there when it was built. Page builders that let anyone construct any layout degrade under that, because thirty people making reasonable decisions still produce an incoherent whole.
We build a constrained set of page types instead: a news post, a person profile, a project page, a publication, an event. Each has defined fields, so filling one in correctly is the easiest available path and the layout cannot break by accident. The University of Hertfordshire work involved this kind of thinking, because a lot of people touch the content over time.
Ask any studio you are considering what happens when a new communications officer starts and has to publish something on their first day with no training. If the answer involves a manual, the system is more complicated than it needs to be. You want to hear that a news post takes about ten minutes to work out.
Governance, sign off and what actually delays projects
Research institution projects rarely run late because of design or development, and almost always run late because content approval passes through a director, a communications team, a funder's brand guidelines and sometimes a board. Each stage is reasonable on its own, and stacked they add months. A typical project of ours runs ten to sixteen weeks.
The fix is to name one person, in writing and at the start, who can approve content on behalf of the institution. They can consult whoever they like, but the studio deals with them. If your organisation cannot appoint that person, that is a real constraint, and it should change how the project is scheduled rather than surfacing in week nine.
It also helps to agree what is out of scope for review, because nobody needs the executive team debating the wording of a cookie notice. We put a rough content calendar in the proposal showing which pages are needed in which week, and it doubles as an early warning system for the weeks after it.
What a project at this scale actually covers
For a research centre or a membership body, a project in the $12,000 to $30,000 range usually covers a defined site with a proper content structure, a publication archive, page types your team can use without help, and design work that puts the institution alongside its peers. Combined brand and website programmes run $30,000 to $80,000, and those are the cases where the identity itself has drifted.
It does not cover a custom research data platform, an integration with every system the institution runs, or a fix for a content problem that is really a staffing problem. If nobody's job includes updating the site, the new site will look excellent for six months and then age where everyone can see it.
Dave Curtin will tell you when the smaller version is the right call, and sometimes that means restructuring the existing information architecture and leaving the visual design alone for another year. Our writing on planning what a site needs to say covers the groundwork we go through before any design starts.
Three things you can do this week
Open the site on a phone and try to find the contact details for research collaboration in under thirty seconds, with a timer running. Institutions tend to be good at describing their work and vague about how to start a conversation, and the repair is usually a named person with an email address.
Second, take your five most important pages and read them aloud to somebody outside your field, marking every sentence where they look confused. Research organisations write for peers by default, which is right for a journal article and wrong for a page that a funding officer will read on a Tuesday morning.
Third, check whether the site says who funds you, who governs it, and when the content was last reviewed, because those mundane signals do a lot of quiet work and they are often missing. Do the three things and you will either have improved the site for nothing, or you will have a clearer brief for whoever you hire next.