Intranet knowledge management: linking published pages to what the organisation really knows

The intranet: where knowledge was published… then forgotten

Almost every organisation has an intranet: quality procedures, internal memos, team pages, safety instructions, internal FAQs. It was often built on a content management system with no API, and it has one peculiarity: everybody knows it exists, and nobody knows any more which pages still reflect how work is really done.

The intranet tells you what was published. It does not tell you who can apply the published procedure, how many people master it, or which critical knowledge is missing from it. Answering those questions takes a knowledge map — connected to the intranet rather than copied from it.

Knowledge mapping, in plain words

A knowledge map is not a list of documents. It is the list of what the organisation knows how to do — pieces of knowledge — linked to the processes and tasks they serve. For each piece of knowledge it says:

• who holds it, with a mastery level from 0 to 5, where 5 means "can teach it";
• how critical it is: criticality combines vulnerability (how many holders, how hard it is to transfer, confidentiality) and relevance (usage, fit with objectives, historical importance);
• which documents describe it — and this is where the intranet comes in.

It differs from classic skills mapping in one respect: risk is not declared, it is computed from assessed mastery, and it rises on its own when a holder leaves.

Reading an intranet without migrating it

The "Website / intranet" connector is the universal driver for anything a browser can display. It accepts three ways of pointing at pages:

1. A sitemap: the address of the sitemap.xml file, and every page it lists is read.
2. A list of addresses, one page per line.
3. Following links, for intranets without a sitemap: an address ending in "/*" is read together with every page it links to on the same site; "/**" goes one level deeper. Links to another site, anchors, files (PDF, images, archives) and addresses already seen are ignored.

If the intranet requires sign-in, the connector presents a username and password, or a token on its own. They are encrypted in the server's vault and never shown again. The readable text of each page is extracted — headings, paragraphs, tables — menus and scripts are ignored, and a page is only read again when its content has changed.

Step by step

1. Administration › Sources › Connect a system › Website / intranet.
2. In Scope, one address per line. Example: https://intranet.example.com/procedures/* reads the procedures home page and every procedure it lists.
3. If needed, the access: username and password, or a token.
4. Check, then Connect, then Sync.
5. Sync a second time: unchanged pages are not read again. That is the proof the connector will not hammer your server.

The limit is 300 pages per connector, followed links included. For a larger intranet, create one connector per section — procedures, quality, human resources — each with its own refresh frequency.

Confinement, and the closed intranet

With the edition installed on your servers, an intranet on your private network is read as it is, without opening any route to the internet. A public website is allowed by name, host by host, and every call is written to the outbound log. With the hosted service, only pages reachable from the internet can be read; for a closed intranet, two routes remain: the sovereign edition, or the inbound API, through which your own system sends its pages.

From page to map

Each page read is compared with the knowledge catalogue. When the full name of a piece of knowledge appears in the page — in its title the suggestion is strong — a link is suggested, and the knowledge manager confirms or rejects. Nothing is linked automatically. Once confirmed:

• each knowledge sheet lists the intranet pages that describe it, with their link;
• search and the assistant find those pages alongside validated sheets;
• red-zone knowledge with no page at all shows up plainly;
• pages that match no knowledge point to a gap in the catalogue.

The connector does not judge whether an intranet page is still accurate: that is not its job. But by showing which pages critical knowledge depends on, it tells you which ones to review first.

Building the map from the intranet: the method

1. Connect one section that matters, for instance maintenance procedures.
2. Read the titles: they form a first inventory of what the organisation believes it knows.
3. Name the knowledge in the catalogue, in the teams' own words, then click "Recompute links": every page already read runs against today's catalogue, without reading the intranet again.
4. Assess mastery for the people concerned, from 0 to 5, first declared then measured through targeted tests: it is the one piece of data the machine cannot invent, and it is what gives criticality its meaning.

The guide on building a knowledge management system from scratch covers the full order, from org chart to assessment campaigns.

What this connector does not read

Files linked from the intranet — PDFs, slide decks, spreadsheets — are not read by this driver, which sticks to pages. If they live in SharePoint or a document management system, connect them through the matching connector, which reads Word, PDF, PowerPoint and Excel. Connectors are included from the Professional plan upwards, and in the sovereign edition.

SharePoint is a trademark of Microsoft Corporation. KnowledgeCapital is neither affiliated with nor endorsed by that company.