Here's a professional secret that sounds like laziness but is actually discipline: when a company files its new 10-K, experienced analysts don't read it. Not front to back. They read what changed — because a filing is mostly last period's language carried forward, and the new information lives almost entirely in the edits. The skill is knowing where change hides; the leverage is having the comparison done for you. This guide covers both.
1. Numbers first: the moves, with magnitudes
Start with what moved: revenue, margins, cash flow, debt — against the same period last year, as a percentage, not an impression. Two disciplines keep this honest. First, materiality: a 2% wiggle in revenue is weather; a 15% move in gross margin is news. Second, direction of surprise: the question is never "is the number big?" but "is it different from what the last filing led me to expect?" When AnalystBook processes a new filing, it computes these moves and flags the material ones — each traceable to the exact reported figures, so a surprising move is one click from its source.
2. Then the language: edits are confessions
The subtler signal is in the prose. Management rewrites a sentence in the MD&A for a reason — "strong demand" becoming "stabilizing demand" is a forecast wearing a euphemism. Reading for edits by eye means holding two hundred pages in your head; this is exactly the part worth automating. AnalystBook's change brief shows edited passages before and after, flags shifts in tone, and surfaces language that's new this period or quietly gone — so "what did they change in how they talk about China?" takes seconds instead of an evening.
3. Risk factors: watch the door, not the room
The risk-factor section is written by lawyers to be ignored — twenty pages of everything that could ever go wrong. Reading it cold is a poor use of an hour. But the edits to it are among the highest-signal text in the whole filing: a company adds a risk when not disclosing it became the bigger legal risk, and removes one when it stopped being true — or stopped being something they want to say. New risks, disappeared risks, and materially rewritten ones are exactly what the change brief isolates. Read those three lists; skip the other nineteen pages with a clear conscience.
4. Check who's acting, not just who's talking
Filings are what management says; trades are what they do. After the numbers and the language, glance at insider activity in the weeks around the filing — clustered buying after an ugly quarter, or selling into a rosy one, is context no paragraph provides. AnalystBook keeps the insider record on the same company page, next to the filings it should be read against.
5. Ask the question the delta raises
A good delta read usually ends with one specific question — "why did receivables jump while revenue was flat?" That's the moment to ask Ana, the AI analyst: she answers from the filings and the computed record, with citations, and says so when the documents don't support an answer. One grounded question beats an hour of re-reading.
6. Write the delta down — that's the compounding step
Finish by recording what changed and what you make of it, in the notes that live next to the company record. Next quarter, you won't re-derive your view — you'll diff against it. A research process where every filing updates a written thesis is how one person covers many companies without shallowing out.
Make it automatic: follow the companies you care about
All of this assumes you know the filing landed. That's what following is for: followed companies are kept fresh on a faster cycle, so a new filing is fetched, processed, and diffed within hours — and a weekly digest email rolls up the changes across your whole list. The result is a watchlist that behaves like a junior analyst: it reads every new filing against the last one and hands you the differences, every time. If you're new to reading filings themselves, start with how to read a 10-K — the delta workflow builds on it. Start a free trial and follow three companies you know well; the first filing that drops will make the case better than this guide can. For research purposes only; not investment advice.