When you type "tech journalist" into a traditional search box, you get exactly that — journalists whose profile contains the words "tech journalist". But what about the writer at Wired who covers AI policy? The Bloomberg reporter who specialises in semiconductor supply chains? The Guardian correspondent who broke the last three major cybersecurity stories? None of them described themselves as a "tech journalist". Yet they're exactly who you need.
This is the fundamental problem with keyword search — and it's the reason we built our journalist discovery engine around semantic search instead.
What Keyword Search Actually Does
Traditional search is a matching exercise. It looks for your exact words — or close variations of them — inside a database. The underlying logic hasn't changed much since the 1970s: terms in your query are compared against terms in documents, and results are ranked by how often and where those terms appear.
For finding products on an e-commerce site, this works reasonably well. For finding the right journalist to cover your story, it falls apart fast.
Journalists rarely describe themselves the way PR professionals search for them. A reporter who covers climate tech might describe their beat as "clean energy transition" or "net zero policy" or "carbon markets". If you search for "sustainability journalist" you might miss all three. And the inverse problem is just as painful — a search for "finance writer" will surface thousands of results who technically match the words but would never touch your fintech story.
The Volume Problem Makes It Worse
There are hundreds of thousands of journalists across the UK and internationally. In a keyword-based system, narrowing that list means adding more filters — publication, location, job title, and so on. Each filter eliminates results mechanically, with no understanding of whether the people being removed were actually relevant.
You end up with a false precision: a short list that looks targeted but was actually produced by elimination rather than relevance. You've filtered out noise, yes — but you've also filtered out the people you actually needed to find.
What Semantic Search Does Differently
Semantic search doesn't match words. It matches meaning.
Under the hood, text — your query, a journalist's profile, their articles, their beat descriptions — is converted into a numerical representation called an embedding. This embedding encodes the conceptual meaning of the text, not just the words used to express it. When you search, the system finds content whose meaning is closest to the meaning of your query, regardless of whether the exact words overlap.
This is how a search for "fintech disruption" surfaces the banking correspondent who wrote about digital wallets, the payments reporter who's been covering open banking, and the startup journalist who profiled three neobanks last quarter — even if not one of them used the phrase "fintech disruption" in their profile.
The technology is grounded in the same advances that power modern large language models. It understands synonyms, related concepts, industry jargon, and contextual relationships in a way that pure keyword matching simply cannot.
How We Use It in Journalist Discovery
PressReacher's journalist discovery doesn't start with a list and a filter panel. It starts with your story.
You describe what you're pitching — in plain English, the way you'd explain it to a colleague. Our system extracts the underlying topics, beats, and themes from that description. It then runs a neural search across our database and across the wider web, finding journalists whose actual coverage — not just their job title — aligns with what you're working on.
The result is a relevance-ranked shortlist of journalists who have genuinely written about the things your story touches on. Not journalists who happened to use the right keywords in their bio. Journalists who have demonstrated, through their published work, that they cover this space.
We also surface recently discovered journalists — reporters who may not yet be widely known in PR circles but whose recent output makes them a strong match for your pitch. Semantic search is particularly good at this, because it finds people based on what they write, not how prominent they are.
The Difference in Practice
Imagine you're launching a mental health app aimed at young professionals. A keyword search for "mental health journalist" returns a broad, noisy list. You'll need to manually review dozens of profiles to work out who actually covers workplace wellbeing versus clinical psychiatry versus charity fundraising.
With semantic search, you describe your story: a digital tool helping young professionals manage work-related anxiety. The system understands that this intersects with workplace wellbeing, mental health in employment, the pressures of always-on culture, and digital health technology. It finds journalists who have written about those themes — from lifestyle angles, business angles, and health angles — and surfaces them ranked by relevance.
The shortlist is smaller, and it's right. You spend your time pitching, not filtering.
Why This Matters for PR
A mismatched pitch isn't just a wasted email. It's a reputational cost. Journalists talk, and repeatedly receiving irrelevant pitches is the fastest way to end up permanently ignored — or worse, publicly called out.
The precision that semantic search enables isn't just a workflow improvement. It's better for the journalists on the other end. When the pitches they receive are genuinely relevant to their beat, the whole ecosystem works better. Journalists engage more, responses improve, and coverage actually happens.
Keyword search was built for a world where databases were small and queries were simple. Journalist discovery — finding the right person in a profession with enormous diversity of specialism, across thousands of publications — is not a simple problem. It needs a search approach that actually understands what you're looking for.
That's what semantic search delivers. And it's what powers discovery on PressReacher.