How herodocu works

By ticker. Enter a US-listed symbol and herodocu finds the company on the SEC registry, takes its most recent annual or quarterly report, and renders the income statement, balance sheet and cash flow on the page. To use an earlier report instead, choose Choose filings and pick from the company's recent reports. Selecting several puts them all in one workbook. Companies listed in the US from abroad are covered too: their annual 20-F or, for many Canadian issuers, 40-F, with figures in the currency they report in.

By PDF. Upload a financial statement PDF under 15 MB — an annual report, interim results, a quarterly statement. herodocu reads the tables, shows them for you to edit, and exports a workbook. On a paid plan you can drop several documents at once and they are processed one after another.

Every run appears under Extraction history, where a ticker result can be viewed again and a PDF workbook re-downloaded for 24 hours.

A long PDF can take a few minutes. You do not have to wait on the page: if you close it, the document still finishes and we email you when it is ready (or if it could not be extracted). In a batch, only the document being extracted at that moment carries on.

The workbook opens on an Overview sheet naming the company, the form, the fiscal period, the reporting currency and the units, with a link to the filing on EDGAR. Then one sheet each for the income statement, balance sheet and cash flow.

Every line item appears in document order, with one column per period as the document presents them. Values are the figures as printed: nothing is recomputed, rescaled or inserted. Printed subtotals are shown in bold, and where a source reference was available it sits in its own column.

A CSV export of the same three statements is available beside the Excel download if your tools prefer flat files. When several filings are exported together, the workbook opens on a Filings index instead, then repeats the three statements per filing. The CSV and both Sheets exports stack the filings one under another, each under a title line naming its form, period and accession number, and each keeping its own period columns — an annual report and a quarterly report can end on the same date, so sharing one column would mix a year's figures with a quarter's.

For Google Sheets there are two routes. Copy for Sheets puts the rows on your clipboard; paste with Ctrl+V (Cmd+V on a Mac) and they arrive as real cells, with the same columns as the CSV. It works the same way in Excel and Numbers, and needs no account linking at all.

Google Sheets instead creates the spreadsheet in your Drive and gives you a link to it. The first use asks Google for permission; we request only the scope that covers files this app creates, so herodocu can reach the sheets it makes for you and nothing else in your Drive. Uploaded PDFs export their tables one under another, each under its own heading.

For an uploaded PDF, every export uses the tables as they stand on the page, so any cell you corrected goes into the workbook, the CSV and the sheet alike. The workbook's arithmetic checks run on your corrected figures, so fixing a mistyped total clears its warning. Corrections are not saved, though: a workbook downloaded again later from Extraction history carries the figures as extracted. When a run holds several documents, all four exports are also offered for the whole run: CSV and Sheets stack each document under a Document: title line, and Excel downloads one workbook per document.

Compare companies takes two to four tickers and puts their latest filings in one table: revenue, gross profit, operating income, net income, cash from operations, capital spending, free cash flow, cash, total assets, liabilities and equity, plus gross, operating, net and free-cash-flow margins. Choose whether to compare annual reports (10-K, 20-F, 40-F) or quarterly ones (10-Q). Each company costs one lookup, exactly as extracting it on its own does, and each run also appears in Extraction history.

Companies do not print the same line items, so each figure is matched to that company's own wording — Apple's revenue line reads Net sales, Microsoft's reads Revenue, and a German filer's net income reads Profit after tax. Hover any figure, or open the Source lines sheet in the Excel export, to see the exact line it came from.

A dash means that company does not print that line at all, not that the figure is zero: banks and insurers report no gross profit or capital spending, for instance. Free cash flow is the one figure we calculate — cash from operations less capital spending — and the export says so.

Two things to watch. Companies choose their own financial year, so an annual comparison lines up the latest annual report of each, whose periods usually end on different dates; every column says which period it covers. And where companies report in different currencies, nothing is converted — we will not invent an exchange rate — so the amounts are not directly comparable while the percentages still are.

After extraction, each printed subtotal is compared against the line items above it. When they do not agree, the subtotal is highlighted and carries a note giving the extracted sum, the printed total and the difference. An Audit Trail sheet lists every such case.

A note is information, not a correction. No figure is ever altered, and no balancing row is ever inserted to make a column add up. Gaps usually come from items the document does not list separately, rounding, or a presentation difference — and occasionally from a genuine extraction miss, which is exactly what the flag is for.

On an uploaded PDF, each statement carries its own label: Adds up when the printed subtotals match the rows above them, Does not add up when one does not, and Not checked when we could not identify the subtotal rows a check needs — a statement we could not verify says so rather than looking verified. Cash flow statements carry no label, because no add-up check applies to them. The checks run on the figures as they stand, so correcting a mistyped total clears its warning, and the Excel, CSV and Sheets exports report exactly what the page shows.

Ticker lookups often need no check at all: where the SEC publishes structured data for a filing, the figures are read directly from it rather than transcribed.

A ticker lookup costs one lookup. A PDF costs one extraction on the free plan, and its page count against your monthly page allowance on a paid plan. Allowances reset on the first of the month, whether you are billed monthly or yearly.

By default everything stops when the allowance runs out and nothing extra is charged. If you would rather keep working, switch on overage under Billing & account and set your own monthly spend cap. You will be emailed when you reach 80% and 100% of an allowance.

Plans can be cancelled at any time and you keep full access until the end of the period you paid for. Receipts for every payment are on the same page. See pricing for the current plans and the refund policy for cancellations.

Two-factor authentication is under Billing & account: turn it on, scan the code with an authenticator app, and from then on signing in — by password, email link or Google — also asks for the 6-digit code the app shows. Admin accounts must use it. API keys keep working without a code. If you lose the phone, email support from your account's address and we'll reset it once we've confirmed it's you.

A promo code is entered on the pricing page before checkout, and the card shows the discounted price before you pay. The discount applies to every renewal for as long as that subscription runs, including after the code itself expires. Each account can use a code once, and it does not carry over if you cancel and subscribe again later.

Text-based PDFs of published financial statements work best: annual reports, interim results, quarterly statements, across fiscal conventions. Scanned PDFs, with no selectable text, are read from images of their pages: they take longer, the figures are more likely to need a check, and on a paid plan they count as more pages than they have, in proportion to the extra processing and at most three times as many. Very long documents are fine; only the pages containing statements matter to the result.

If a run returns no income statement, the document may present it as a statement of comprehensive income or of profit or loss, and herodocu tells you when that has happened. If a statement is missing or looks wrong, the most reliable fix is to upload just the pages containing the statements rather than the whole report.

Extraction failures refund the allowance they took. If a run finishes but the result is not right, you can run the same file again once, within 24 hours, at no charge: choose Re-run free under the result, or upload the identical file again. A trimmed or re-saved PDF is a different file and counts as a new extraction.

If a document consistently fails, send it to support@herodocu.com and we will look at it.

On a paid plan you can create an API key under Billing & account and call the same endpoints the app uses from your own scripts. Your plan's allowances and caps apply to the key exactly as they do in the app, and a key can be revoked at any time.

herodocu is also an MCP server, so Claude, ChatGPT and other assistants can search companies, transcribe a filing's statements and review your extraction history on your behalf.

The developer guide documents every endpoint with examples.