60.93 percent of 2,186 investor pages found, and a fiscal year end on 8 times as many companies

The drawing

The problem

Every company website hides the same pages in different places: investor relations, pricing, careers, trust centers, terms. A buyer who needs one of them across a list of companies either clicks through each site or pays someone to. We wanted an actor that does it, and we wanted every number in its listing to be measured, with its sample size stated.

What we built

The Page Finder and Extractor on Apify takes a domain and a page type, renders the site when it has to, finds the page, and extracts its fields. We measured it on a population of 2,207 listed company domains before publishing, and hand read random samples against the stored pages.

What it produced

• The investor relations page found for 60.93 percent of 2,186 companies, against 60.1 percent for our previous approach.
• A fiscal year end on 16.15 percent of companies, against 2.0 percent before, about 8 times as many.
• 96.7 percent precision on the located investor pages, on a hand read sample of 30.
• 97.6 percent agreement with official filings within 7 days on high confidence rows, on a sample of 41.
• A row for 100 percent of inputs. A page we could not read returns as undetermined, not as missing.

What runs now

The actor went public on August 16, 2026, with 50 ready made tasks, 20 of them published on the Store, and an MCP server on npm and the Official MCP Registry.

Your scenario

If your list needs one page from every company on it, try the Page Finder and Extractor on Apify with your page type, or send us the list and we'll tell you which page types it reads best.