So You Want to Make Your Intranet Actually Useful
I spent about three years helping our campus roll out a replacement intranet at a medical center. It started as a document repository that looked like it was built in 2003, migrated content to a new platform, and eventually turned into something people actually opened before they realized what it was. The Penn Medicine intranet story is one of those projects that seems straightforward on paper and then you hit the reality of 40,000 employees across dozens of buildings who still have their bookmarks set to the old system. When I hear people talk about intranet transformation, they immediately jump to features and modules. The actual bottleneck is almost always different. It's about what happens after go-live. At Penn, the intranet team learned pretty quickly that getting people to use the new system required solving a trust problem first, not a technology problem. Most clinicians and staff had spent a decade or more navigating a portal that broke occasionally, loaded slowly, and buried everything under layers of navigation menus designed for administrators who liked categories. I watched a senior nurse manager refuse to switch her team because the old intranet had a shared folder they could bookmark and come back to. The new system had better search but the page structure was unfamiliar. She didn't care about the better search. She cared about consistency. The workaround we used was giving each department a personalized dashboard with their most-used links pinned to the front page. It took two weeks of configuration but she stuck with it after that. That pattern repeated across probably sixty departments before we had enough momentum.
How the Penn Approach Actually Works
Unlocking Collaboration How The Penn Medicine Intranet Is Transforming Workplace Communication
The framework Penn built around their intranet is built on three practical assumptions. First, that most communication in a medical center is already happening somewhere else, and the intranet should aggregate it rather than compete with it. Second, that the people doing clinical work don't have fifteen minutes to find policy updates, so everything important surfaces automatically. Third, that departments need ownership over their own sections without waiting for central IT to publish content. The platform itself uses a modular design where teams can create and manage their own subpages within governance guidelines. Content owners are identified on every page, which sounds simple but solves half the problems you'd get with outdated information. I once found a surgery scheduling policy from 2017 because no one had updated it and the content owner had retired three years earlier. The governance model at Penn requires content review every twelve months and flags pages as approaching their review date. It's not perfect. I saw several departments ignore the reminders and the system just kept nagging them quietly in the background. Collaboration features include shared workspaces where teams can co-author documents, leave threaded comments, and connect to SharePoint sites without leaving the intranet interface. This matters more than people usually expect because clinical teams often work across departments. A research project might involve someone from epidemiology, someone from the pharmacy department, and someone in biostatistics. Without a shared workspace, those conversations scatter across email and individual desktops. The intranet keeps them visible and searchable.
What Nobody Tells You About Implementation
The technical setup is the easy part. Getting people to adopt it reliably is where projects stall. I saw two different approaches at Penn and both had real costs. The top-down approach used mandatory training sessions and leadership memos. Within six months, compliance rates were high but engagement was shallow. People logged in, clicked through the required modules, and left. The bottom-up approach gave department champions access to training on how to build their own pages and customized their sections however they wanted. Adoption was slower in the first quarter but by month eight the number of active content creators had grown significantly and people were actually using the collaboration tools. The hybrid model Penn settled on combines elements of both. Department leads attend an orientation session, then their teams get self-service access to build out their own pages with templates and guardrails. The templates are important because they prevent every department from designing something completely different and make the whole system feel coherent. Penn has about twelve standard templates covering things like policy pages, event calendars, team directories, and project hubs. Most departments pick one or two and customize within those bounds. There's also an accessibility layer that many people overlook. The intranet has to meet WCAG standards and Penn does regular audits. This affects everything from color contrast on pages to how screen readers navigate the content. It's not optional at a medical institution and the penalties for non-compliance can be significant. I spent a full week working with the accessibility team on one problematic module because the original design team hadn't considered keyboard navigation at all. Fixed it, but it was a painful lesson in doing it right the first time.
Get the Full Details
The Search Problem
If your intranet search doesn't work well, nothing else matters. I've never seen a case where good search wasn't the deciding factor for whether people kept using the system. Penn invested heavily in their search indexing, including full-text search across all document types and a separate knowledge base for frequently asked questions. They also implemented a faceted search system that lets users filter by department, document type, date range, and audience. One specific edge case that caused real headaches: medical terminology. Searching for "chemotherapy" returns different results depending on whether the user is in oncology or billing. The solution was role-based result personalization combined with a thesaurus mapping so that related terms surface together. It's not perfect. I still see results I'd classify as irrelevant sometimes, but the team has been refining it for years and the accuracy is now reasonable enough that most users stop complaining about it.
Measuring What Actually Matters
Most intranet dashboards track page views and unique visitors. These numbers tell you nothing about whether the system is improving workplace communication. Penn tracks a few different metrics that are more useful. One is the percentage of critical communications distributed through the intranet versus other channels. Another is the average time to find a piece of information, measured through periodic surveys rather than system logs. A third is the number of cross-departmental collaborations initiated through the platform, which they track by looking at how many workspaces involve members from three or more departments. The hard truth is that none of these metrics are clean. The cross-department collaboration count, for example, doesn't distinguish between a productive research project and a frustrated attempt by someone to find a contact who should have been easier to reach. But it's a starting point. The alternative is tracking page views and telling your leadership that engagement is up when nobody has actually changed how they communicate with each other.
When the Intranet Won't Fix Your Problem
There are scenarios where building out the intranet further won't help. If your organization has a culture where people communicate exclusively through email and instant messaging, no amount of intranet investment will change that behavior overnight. The tool has to solve a real pain point people already feel. At Penn, the breakthrough moment came when they started routing the most annoying manual processes through the intranet. Things like submitting time-off requests, accessing badge renewal forms, and finding parking information. Once people experienced the relief of not having to call three different offices to complete those tasks, they became more willing to try using the collaboration features. Another limitation: the intranet cannot replace direct communication. Clinical handoffs, emergency coordination, and urgent patient-related matters should still happen through dedicated clinical systems, not through a general-purpose intranet. I've seen organizations try to force everything into the intranet and end up with a bottleneck that slows down real work. The Penn model keeps clinical systems separate and uses the intranet for everything that happens outside direct patient care interactions. If your organization is small enough that everyone knows everyone else, a full intranet platform might be overkill. A shared drives setup with clear naming conventions and maybe a basic team calendar could handle the same workload with less maintenance and lower cost. The Penn approach assumes an organization large enough that information asymmetry is a genuine daily problem.
Practical Steps if You're Considering This
Start by mapping where your people currently get information. Not where they say they get it from in a survey, but where you find them actually looking. I spent a week shadowing different roles just to see which browser tabs people had open during a typical morning. Nurses had clinical portals, staff schedules, and a mess of email. Administrators had spreadsheets and a shared drive nobody organized. Understanding that chaos is what you're trying to fix. Then identify your content owners before you build anything. Every department needs someone who is willing to keep their section current. If you can't find those people, the intranet will become a graveyard of stale information within a year. At Penn, some departments assigned this as a formal responsibility added to job descriptions. Others paid stipends. Neither approach was elegant but both produced results. Finally, resist the urge to migrate everything at once. I watched a team try to migrate thirty years of documents and give up halfway through because the volume was overwhelming. Penn migrated the most accessed policies and forms first, then worked through the rest in priority order. The pages that got the least traffic after six months were simply archived or removed. This meant the system stayed manageable and people weren't confused by hundreds of irrelevant results for simple searches.
