Communications teams sit on the most relational data in any organisation: who wrote what, in which outlet, about which narrative, quoting which spokesperson, in response to which announcement, noticed by which policymaker. Then they store it in the flattest formats ever devised – clippings, PDFs, spreadsheets, inboxes. Knowledge graphs exist for exactly this kind of communications problem.
That mismatch is the subject of this post: what a knowledge graph is, why communications work is unusually suited to one, and what actually changes when you connect the material instead of filing it.
What a knowledge graph is
A knowledge graph stores things and the relationships between them, rather than documents in folders. A journalist is a thing; so is an article, an outlet, a narrative, a bill, a spokesperson, a statement. The graph records how they connect: this journalist wrote that article, which advanced that narrative, which cites your CEO’s statement from March.
The reason this matters now is that language models reason far better over structured relationships than over piles of flat documents. Give a model a folder of coverage and it summarises; give it a graph and it can answer the questions communications people actually ask. Who has covered us on this topic before? How did this narrative develop? What did we say last time?
There is a deeper reason too. A press release records what you decided to say; it does not record why you said it that way, what you considered and rejected, or what the lawyers struck out. That reasoning lives in the conversations before the document – and a graph, unlike an archive, has somewhere to put it.
Documents are the output of knowledge work, not the knowledge itself.
Why knowledge graphs fit communications work
Plenty of business functions can use a knowledge graph. Communications is unusual in that the relationships are the entire job. A clipping means little on its own; its meaning is its connections – to the journalist’s previous coverage, to the narrative it advances, to the event it responds to, to what competitors said the same week.
Three shapes we run in production:
Monitoring becomes narrative intelligence. A flat monitoring feed tells you what was published today. A graph connects today’s articles to the stories they belong to, so the daily output is not a list of links but a map of narratives as they develop – which are accelerating, which have gone quiet, which just crossed from trade press into nationals.
A legislature becomes a briefing. Parliamentary output is natively graph-shaped: bills, amendments, committees, members, statements, all cross-referenced. Connected properly, both chambers’ daily activity becomes a focused briefing in the inbox before the morning meeting.
The archive starts answering. Years of releases, statements, and Q&As, connected by topic and time, become something a drafting assistant can actually use – not just your voice, but your history, and what you have never said and should not start saying now.
The part nobody tells you: capture is not free
The graph only knows what enters it.
Coverage, transcripts, and published documents can be ingested automatically, and should be – capture that depends on people remembering to file things will starve. But the most valuable layer, the reasoning behind decisions, does not arrive automatically. Someone has to record why the statement took the shape it did, and recording the why takes time that busy people do not naturally spend.
We know because we run our own company on one, and we argue about exactly this internally. Our working answer: automate everything that can be automated (the meetings, the monitoring, the documents), make recording the rest as light as a sentence in the flow of work, and accept that the graph is an asset you tend, not a system you install.
The other honest caveat: not everything needs a graph. If your question is “find me the boilerplate paragraph”, flat search is fine. A graph earns its keep where relationships and time carry the meaning – narratives, stakeholders, positions that evolve. If a spreadsheet answers your questions, keep the spreadsheet.
Where to start
Not with an enterprise knowledge programme. Pick the single workflow where connected beats filed most obviously – for most communications teams, that is monitoring, because the data already arrives daily and the pain of flat coverage lists is felt every morning. Get one graph-backed briefing into people’s inboxes, let the memory accumulate for a quarter, and the second use case will suggest itself: the archive is already half-connected by then.
That compounding is the real argument. A monitoring subscription is worth the same every month. A knowledge graph is worth more every month, because everything it ingests makes everything else it knows more useful. The longer it runs, the more it knows – and at some point it stops being a tool and starts being your organisation’s memory.
We run our own company on the pattern described here, and deploy it for communications, public affairs, and finance teams. If you want to see what a living company memory looks like after a year, we wrote that up too.
Leave a Reply