ESG and controversy monitoring

The problem

Controversies surface in local-language coverage and specialist trade press long before they reach international reporting or a ratings provider, and by then the position has usually been taken.

How does a news API support ESG and controversy monitoring?

Cross-language monitoring against a holdings or supplier list, grouped by event so a single incident does not register as a wave.

What the workflow looks like

1. Build the watch list from holdings or suppliers as entities.
2. Use related_entities on a holding to see which organizations, places and topics its coverage keeps naming; a new name near the top is a reason to read the evidence articles, not a finding in itself.
3. Monitor the languages of the markets each operates in, not the language of your office.
4. Separate incident coverage from commentary by checking outlet spread and recency.
5. Record the first-seen timestamp, which get_story carries for the whole story — for stewardship, when you could have known matters.

What will a news API not do for ESG and controversy monitoring?

This is a detection layer, not an assessment. It will tell you an allegation is being reported and cannot tell you whether it is substantiated, material or already remediated. Every hit needs a human before it informs a position.

The mistake teams make here

The characteristic error is ranking companies by controversy volume without normalising, which ranks press freedom and press size rather than conduct. A holding in a market with a dense independent press will generate more adverse coverage than an identically behaved holding in a market without one, and the ordering that falls out of a raw count is close to the inverse of what a stewardship process wants. The second error is treating a detected allegation as a finding. Coverage establishes that something is being reported, not that it is substantiated, material or current, and a controversy that was resolved two years ago reads identically to one that is live unless someone checks.

How would you know ESG and controversy monitoring is working?

Check for coverage bias before drawing any conclusion. Companies in markets with a dense free press generate more controversy coverage than companies behaving identically elsewhere, so a naive count ranks press freedom rather than conduct. Normalise against a sector-and-region baseline, and re-derive that baseline periodically — it moves with news cycles more than most people expect.

What to build in the first week

Take twenty holdings and run the check by hand, in the languages of the markets they operate in rather than in English alone. Read everything that comes back and sort it into three piles: genuine incidents, commentary about incidents, and the wrong entity. The proportions will tell you what the automated version has to do, and the third pile in particular will tell you whether name resolution is going to be the hard part — for holdings with common names or local-script renderings it usually is. A week of manual work here determines whether the pipeline is worth building, and for some portfolios the honest answer is that it is not.

What are the alternatives for ESG and controversy monitoring?

ESG data vendors deliver scored, normalised, point-in-time datasets with methodology documents your investment committee can interrogate, which is what most stewardship processes actually require. NGO and regulatory sources are authoritative where news is indicative. This is the early-warning layer between vendor refresh cycles, and it is best used to prompt a question rather than to answer one.

The queries that implement it

• Monitor a market whose language nobody on the team reads — /stories in semantic mode
• Find coverage with a particular tone — /search in hybrid mode
• Watch what is said about a named individual — /stories in hybrid mode

Each recipe makes the case for its parameters. Response fields, failure modes and scheduling cost are in the technical documentation.