The part of XBRL every vendor throws away
Alongside the numbers, filings ship a calculation linkbase — the company's own declaration of which line rolls into which subtotal, with signs. It makes adjustments propagate correctly, it ties out arithmetic automatically, and it exposes comparability gaps invisible in flattened data: Danaher's filed graph computes a gross-profit subtotal and Thermo Fisher's does not.
Vendors flatten filings into a standard template. The template is what makes their data comparable, and it’s also what destroys the most useful thing in the file.
Every filing carries a calculation linkbase: the company’s own declaration of the arithmetic — which concept rolls into which subtotal, with what sign. It is filed under the same rules as the numbers themselves, and it is thrown away by every product that maps filings onto a fixed chart of accounts.
Three things it buys you.
Adjustments propagate correctly. Add back a charge and you can walk the parent chain to see which subtotals genuinely move — EBIT, pre-tax, net income — instead of guessing.
Arithmetic ties out. Every parent is Σ (weight × child). Evaluate them all and you get
a per-filing correctness check for free. Discrepancies are findings, not bugs.
Comparability gaps become visible. Danaher’s graph computes a gross-profit subtotal. Thermo Fisher’s routes revenue through a costs-and-expenses block and never computes one. Their gross margins were never the same measurement — and you cannot see that in flattened data, only in the graph the companies filed.