How we work
SubsidySignal is a service, not a self-service tool. This page describes how we build and maintain a target map for a client, where the information comes from, what AI does and does not do, and where the limits are.
The service process
It starts with a call in which we learn what you sell, to whom, in which markets and at what deal size. From that brief our team builds a personal target map: organizations that recently received a grant, an investment or a public contract and plausibly need your services. The map is updated weekly in your client portal. For each organization we prepare the approach — the right person, the reason to talk now, the message — and, where we have a way in, we make the introduction. We do not run mass mailings or advertising on your behalf.
Which official sources we monitor
Our monitoring platform ingests three official source families: CORDIS (Horizon Europe project and participant data, refreshed monthly), the European Commission Financial Transparency System (yearly beneficiary datasets), and TED contract award notices (daily). Each source's coverage, cadence and licence is documented on the data sources page. Every funding fact keeps its source URL and retrieval timestamp.
Our network
Part of our information comes from our own network of industry contacts and advisers who share market intelligence — for example that a budget has been approved, which priorities a funded programme has, or who decides. We treat such information as a lead to verify, not as a fact: it enters your map after our team has checked it against public records where possible, and it is labeled as network intelligence rather than as an official record.
How data is normalized
Organization names arrive in many spellings (Siemens AG, SIEMENS AKTIENGESELLSCHAFT…). We normalize legal forms, match by domain and registration identifiers where available, and use fuzzy matching with a stored confidence score. Every funding event keeps its source URL, retrieval timestamp and raw payload — provenance is never discarded.
What AI does — and does not do
Our monitoring platform uses a language model to read the stated objectives of a funded project and map them to a fixed taxonomy of spend categories, each with a probability, expected timeframe and written reasoning. These are predictions, labeled as such, and they are one input for our analysts. AI does not decide who goes on your map, does not write your outreach, and does not contact anyone.
Human review
Every organization on your map has been looked at by an analyst who understands your offer. We check that the funding is real and recent, that the organization plausibly buys what you sell, and that there is a sensible reason to talk now. Signals that do not pass that review do not reach you, however high the model's score.
Fact vs. inference
Amounts, recipients, dates and programmes are FACTS from official records, always linked to source. Sector classification and project interpretation are INFERENCES. Future purchases are PREDICTIONS — probabilities, never certainties. Network intelligence is labeled separately. The portal labels all of these, everywhere.
What confidence means
Confidence (0–100) expresses how well the source material supports the analysis: a detailed project description with explicit procurement plans scores high; a one-line grant record scores low. Our analysts use it as a triage aid; it is never the sole reason an organization is added to or removed from your map.
Known limitations
FTS data is published annually, so those events can lag up to 18 months. CORDIS refreshes monthly, not in real time. Multi-partner projects make per-organization amounts approximate when only totals are published. Predictions can be wrong. Network intelligence is partial and depends on where our contacts are active. And we cannot promise meetings or deals — we can promise that every organization on your map has a real, recent budget and has been reviewed by a person.
Update frequency
TED: daily. CORDIS: monthly (matching the official dataset refresh). FTS: yearly (matching official publication). Platform enrichment runs daily; your map is reviewed and updated weekly, and immediately when our network surfaces something time-critical.
Want a map of organizations like this, built for what you sell?
Built from official European public funding and procurement sources.