# A market intelligence base of 1.9M industrial companies > How Red Bumerang turned thousands of one-off client databases into one searchable base of industrial companies, enriched from what each business makes. Source: https://redbumerang.com/en/case-studies/market-intelligence/ Our own systems / market data ## Where do we sell next, and why? Years of market knowledge used to die on delivery, one Excel file at a time. It now lives in one base that every engagement adds to and every engagement can question. *Red Bumerang’s own system. Figures as at 6 September 2026. Internal tooling is described, never named.* ## The first problem was ours We had spent years building market knowledge, and it died on delivery. Across {{fact:clients-cumulative}} clients we had built thousands of company databases — Europe and beyond. Each one took real work: collection, enrichment, and thousands of hours of classification done by hand. And each one was delivered as an Excel file. Static. Not searchable. No shared structure, no accumulated meta-information, nothing connecting the market we mapped this month to the one we mapped last year. Every engagement started the market from zero, and everything learned in the previous one was gone. Not lost as in deleted — lost as in unfindable. The knowledge existed, in thousands of files, in a form nobody could ask a question of. So we built one base instead of thousands of files. Every company we classify goes into it, enriched by the same automated pipeline that reads company websites for what a business actually makes and can do. It now holds {{fact:companies-classified}} industrial companies. ## The second problem is the client's A manufacturer decides where to sell next based on the founder's instinct — and cannot say why. "Let's go to Germany." It might be right. But ask why Germany, why now, and why not Poland, and the answer is usually a feeling, a trade fair, or the fact that the last German customer went well. This is the biggest unanswered question in industrial sales, and it is expensive in a way that never shows up as a line item. A wrong market costs a year of a sales team's effort before anyone admits it was wrong. So we pointed the base at the market itself. Around 500 sources — national manufacturing publications, statistics offices, industry feeds — are crawled continuously and classified into five kinds of buying signal, then attached to the companies and geographies they concern. - **FLOW** — What it captures: Production or supply chain moving from one region to another · Example: German automakers shifting production to Poland - **TREND** — What it captures: A sector growing or declining in a region · Example: Plastics production index drops 8% in Germany - **ENTRY** — What it captures: New investment, factory opening, or FDI announcement · Example: Samsung announces €3B chip plant in Dresden - **RISK** — What it captures: Insolvencies, closures, exits, sector stress · Example: Wave of insolvencies among German injection moulders Those five are not categories for filing things. They are the five ways a market tells you it is about to buy — or about to stop. A FLOW signal says your customer's production is moving and your sales territory should move with it. A RISK signal says a segment you are about to invest in is contracting. Read together, over time, they describe where the work is going. The classification runs on an open-source model on our own infrastructure, not a commercial API. At this volume that is the difference between a system that runs every day and one that is too expensive to leave switched on. The result answers both halves of the question. Not just where to go, but the evidence for why — which industries in which countries are actually moving, from sources a client can check. - **GROWTH** — What it captures: Market expansion or demand increase · Example: Romanian glass packaging demand up 29% ## Then clients started asking questions nobody could answer Once the base existed, it turned out to be the answer to questions we had not been asked yet. An energy group came to us with one: which Slovenian industrial companies have a large enough roof to be worth approaching for solar, and which already have panels? We had around 140,000 Slovenian companies in the base already. The question became: take those companies, add roof geometry from national aerial imagery, add an energy profile, add subsidy history — and produce a ranked list. No company in Slovenia could answer that question. Not because it was technically impossible, but because answering it required a classified base of companies to start from, and nobody had one. The requests kept coming in the same shape: subsidies received, de-minimis state aid, energy consumption classification. Each is a different question, and each is answerable in days rather than months because the base and the pipeline already exist. ## What it is, in one line We built infrastructure for ourselves, and it turned out to be the product. Red Bumerang provides companies with the data, the AI and the intelligence so that their sales process is on-point, efficient, and based on data rather than the CEO's feeling. ## Method What the numbers count. {{fact:companies-classified}} is the company count in the base, verified live on 6 September 2026. 360 is cumulative clients since October 2020. Around 500 refers to monitored signal sources, and around 140,000 to Slovenian companies in the base — both stated as approximate because both change. What is measured and what is judged. Company identity, classification codes and logistics scoring are deterministic. Reading a company website into a capability profile, and classifying a market signal into one of the five types, are model work — they are judgements about meaning. What is not claimed. No accuracy rate is published for signal classification or website enrichment. Confidence is recorded per record, which is not the same as a validated accuracy figure, and we do not present it as one. --- ## Bring us a process, or a market. Every system here started as a conversation about work that was already happening — how it actually runs, where it costs the most, and what the order underneath it looks like. That is the first meeting.