Signal notebooks
The public event that says a buyer is about to need what you sell, one industry at a time.
Most prospecting starts from what a company is: its size, its industry, the software it runs. That tells you whether someone could buy. It says nothing about when.
These are notes on the other question. Something happens in public, and it is published: a register entry, a zoning plan, a permit, a filing. Each note takes one industry and works out which event matters, where it is published, what filter turns a feed into a shortlist, and what we got wrong the first time.
One rule. An industry gets a note only after we have actually run the search. Where a probe failed, it is listed below as a failure rather than written up as a method.
Commercial real estate
A company that has just registered has nowhere to sit
A foreign company registers a branch, and its address is somebody else’s
Renewable energy
The signal belongs to the neighbourhood, not the company
A solar park is built next door, and the landowner watches the rent arrive
Accounting and corporate law
Six per cent of one day’s new companies shared a single address
A company is registered, and for a few weeks it has no accountant
Finance and banking
We went looking for a signal and found an address
A company gets an LEI, and a register confirms where it actually is
Construction and fit-out
The open-data version had the names removed
A building permit is published months before anyone is hired
Staffing and recruitment
Everyone sells “they are hiring”. The signal is that they failed to
A role has been reposted, which means the last attempt did not work
Insurance and benefits
The obligation is public. The headcount that triggers it is not
A company crosses an employee threshold and owes duties it does not know about
Logistics and freight
The register exists, and it can only answer questions you already know
A company takes warehouse space, months before its first shipment
What eight probes had in common
One thing turned up in three of these independently, in three countries and three unrelated datasets, so it is worth naming on its own. The curated, official, documented version of a public dataset is usually missing the field that makes it useful.
Zurich publishes its building permits as open data, and the export has every name stripped out. The same records through the gazette API carry the client and the architect. The UK planning API returns the literal string “See source” where the applicant should be. The German labour agency’s website collapses every older job posting into “30+ days ago”, while its API returns the two dates that tell you a role has been reposted.
In each case somebody built a friendlier view and dropped the column that answers the interesting question. If a source looks useless, check whether you are reading the source or a summary of it.
The second recurring lesson is about timing. A source that proves a need is real often proves it has already been met. Completed projects are the best documented ones. In renewable energy the landowners with perfect evidence had already signed with a developer. In construction the same trap is a contract award notice rather than a call for tenders. Find the field that separates deciding from decided, and filter on it first.