Home / Recipes / Catch supply chain disruption before it reaches the wires
Worked queryCatch supply chain disruption before it reaches the wires
/stories in semantic mode. The parameters matter less
than the reasoning behind them, so that is most of what is below.
The request
GET /stories
?q=plant closure production halt shipping delay port congestion
&mode=semantic
&language=deu,jpn,zho,nld
&from=<now - 7d>
# then, for one port you depend on: strikes within 75 km of Rotterdam
GET /stories
?event_type=labor_strike
&near=51.9244,4.4777
&radius_km=75
&from=<now - 7d>Why these parameters
The trade and domestic-language press leads the international wire on this by days. German industrial reporting names plants and revised output figures; Japanese coverage carries supplier detail during the March reporting period; Dutch outlets sit on a genuine trade chokepoint and report port disruption early. Querying only English means reading the summary after everyone has it. The second request pins the watch to one site: near matches articles mentioning a city within 75 km of the port, in any language, and event_type keeps the ones carrying the strike class. A class is precise rather than complete, so that request runs beside the semantic one rather than replacing it, and its partial flag says when part of the index has not yet been rebuilt with place and event fields.
The obvious approach, and why it loses
The obvious approach is an English query on supply chain disruption, which retrieves analysis and commentary rather than reports of specific incidents. The vocabulary is the reason: an incident is reported locally as a fire at a named plant, a strike at a named port, a halt to a named production line, and none of those phrases contain the words you searched for. The second instinct, watching your suppliers by name, is worth doing and covers only the tier you have names for — the disruptions that matter are usually two tiers down, at a supplier your supplier has not told you about.
When it returns too much, or too little
Too much means the concept list is pulling in routine industrial news; add near with a radius around the ports and plants you actually depend on, or a country or an outlet set for the corridors, rather than removing concepts, since each concept is covering a different failure mode. Too little is usually the language list rather than the window — the recipe leans on German, Japanese, Chinese and Dutch coverage for a reason, and dropping to English alone will return a fraction of it. If you need physical-disruption reporting specifically rather than policy, add the supply_disruption signal with a lower bound and raise it until commentary drops out.
How you would know it is working
Measure lead time against events you eventually learned about anyway. Take ten disruptions from the last year that reached you through a supplier, a customer or the wires, and check when each first appeared in the index. The median gap is what this recipe is worth, it is quantifiable, and it is the number to put in front of whoever is paying for it. Track the false-escalation rate alongside it, because that is what gets these systems switched off: count how many alerts led to an action that turned out to be unnecessary, and require a second independent outlet before anything expensive fires.
When this is the wrong tool
Early signals are early precisely because they are unconfirmed, so this recipe trades precision for lead time and there is no setting that gives you both. A domestic report of a production halt may be corrected, may be routine maintenance reported as news, or may be a local outlet repeating a rumour. If your downstream action is expensive or hard to reverse, wait for a second independent outlet before acting. This is also a poor fit for long-horizon planning, where structural data on capacity and trade flows is far more reliable than news. The value here is specifically in the days between something happening and the wire reporting it, and outside that window other sources are better.
Running it for real
Response fields, failure modes, adaptations and scheduling cost: Catch supply chain disruption before it reaches the wires, in the docs.