Current limitations
Stated plainly, because finding these out by surprise is worse.
Logo coverage
About half of companies have a logo. The rest show initials, because the product declines to guess rather than risk showing a different business's mark.
Ordinary-word company names
A company whose name is an everyday word, such as “Factory”, “Trolley”, “Granite” or “Backbone”, can collect news that is not about it, because matching by name cannot always tell the difference.
This is much better than it was. The product now recognises when a company's name is an ordinary word and asks for corroboration before attaching a story: a term from the company's sector, a funding round or other corporate event, or a business outlet that would plausibly be writing about it. Across the companies worst affected, that removes the large majority of unrelated articles.
It does not remove all of them, and the remaining fallback is deliberate: without it, real coverage of your companies in the Financial Times and Bloomberg would be dropped alongside the noise. If you see an obviously unrelated article, report it through support and it can be corrected.
Companies with similar names
Where two real companies share the start of a name, such as Solvera and Solvera Group, or Odyssey and Odyssey Logistics, the product can only tell them apart once both are tracked. Until the second one exists as a company record, its news attaches to the first.
The fix is usually to add the namesake
From that point the two separate, and existing signals can be moved across on request through support. Use the namesake's full name: matching reads the name on the record, so “Odyssey Investment Partners” separates cleanly where “Odyssey Investment” is a weaker handle on the same company.
Was this article helpful?