Research & PolicyFree
10-K Analysis for Competitive Intelligence
This year's filing against last year's, paragraph by paragraph, so the risks they added and the ones they moved up the list come out first.
River reads the filing against the one before it. Every paragraph of Item 1A is matched to its counterpart from last year, and the output leads with the differences: risk factors added, risk factors deleted, the ones reworded materially, and the ones that moved position. Then the same treatment on the business description and the discussion of results, with segment numbers and metric definitions compared across periods. Every finding carries the item, the page and the sentence it came from.
Nobody does this because it needs two documents. Every filing summary, every research note and every model built on extracted financials works from one 10-K at a time. The ninety per cent of Item 1A that is copied forward reads exactly like the ten per cent that is new. The SEC has been trying to thin the boilerplate: Item 105 says the presentation of risks that could apply generically to any registrant is discouraged. It survives anyway, and the diff is what separates it from the signal.
Built for the competitive-intelligence analyst, the product leader who keeps losing deals to one company, and the investor reading a filing for what it concedes rather than what it claims. It works on any regular filer, and the same discipline runs through an evidence extraction table for research literature and the filing analysis for a regulator's notice. Sources get kept the way a documented search strategy keeps them, next to the other research tools and workspace packs.
Thirty-one risk factors became thirty-four
A logistics software company files its fiscal 2025 annual report. Item 1A carries 34 risk factors against 31 last year. Twenty-four are word for word identical. Three were tidied. Two were rewritten in a way that changes their meaning, two were dropped, and five are new. The five new ones name customer concentration, energy costs at leased data centres, integration of a June acquisition, a covenant test, and a state privacy regime. Nothing gets added to Item 1A casually, because adding one is an admission.
The most useful change is not in the text at all. Dependence on a limited number of customers sat at position 22 last year and sits at position 3 this year. Item 105 requires the discussion to be organized logically with relevant headings, so the ordering is a choice the company made, reviewed by its counsel. The rewritten text supplies the reason: one customer was 19 per cent of revenue this year, against 11 per cent last year.
Now the numbers, which confirm it. Revenue rose 8 per cent, from 567.0 to 612.4 million dollars. That one customer went from 62.4 million to 116.4 million, a gain of 54.0 million against total growth of 45.4 million. Everything else in the business fell by 8.6 million. The company did not say that. It disclosed the concentration, ordered the risk factor accordingly, and reported the segment lines. Subtraction is what makes it a finding.
How it works
Name the company
The company and the filing, plus where you sit relative to them competitively.
Pull both years
This filing and the one before it, both retrieved in full rather than summarised.
Diff the language
Item 1A matched paragraph to paragraph, then the business description and the results discussion.
Check the numbers
Segment lines, metrics and definitions across periods, tested against what the language changed.
What you get
- A paragraph-level diff of Item 1A, sorted by added, deleted, rewritten and moved
- Position changes in the risk factor ordering, which is a choice the company made deliberately
- Segment revenue and margin across periods, with the arithmetic the filing does not do
- Metric definitions compared year over year, so a redefined series stops passing as growth
- Every finding carrying its item number, its page and the sentence it was drawn from
- The comparison Sheet, built to reopen and extend when the next filing lands
Common questions
Why compare two filings instead of reading the newest one?
Because most of a 10-K is last year's 10-K. Risk factors in particular get carried forward until something forces a change, so the paragraphs that differ are the ones the company decided it had to write. Read one filing and the boilerplate and the disclosure look identical, since they are formatted identically and sit in the same list.
Is a risk factor moving up the list really meaningful?
The order is a choice, made by people who knew it would be read. Companies are required to organise the discussion logically under headings, so position carries intent in a way a single sentence often does not. A risk climbing nineteen places usually arrives with a rewritten paragraph and a number underneath it, and the three together are the finding.
What if the company restated its segments?
Then the prior-year comparison has to be rebuilt before anything is said about growth, and the analysis says so rather than putting two incomparable numbers side by side. Restated segments are common after an acquisition, and they are also the easiest place to lose a declining line inside a growing one.
Does it work on earnings calls and investor decks?
Yes, and the gap between them and the filing is often the point. A deck is marketing and a 10-K is a document with liability attached, so a claim made in one and absent from the other is worth noticing. Transcripts also get diffed against the prior quarter, where the same copy-forward habit applies to prepared remarks.
Can it tell me why a metric moved?
It can tell you when the metric itself moved, which is more often the answer. Definitions of recurring revenue, retention, backlog and adjusted margin change between filings, and a redefined series looks like growth. Each one has to carry a reconciliation to the comparable GAAP measure, so the analysis puts last year's definition beside this year's.
What about a private company or a foreign filer?
Foreign private issuers file annual reports with the same broad structure, so the method transfers with different item numbers. For private companies there is no filing to diff, so a competitive intelligence report records the gap with a source and a date instead of substituting guesswork. Where they hold debt, publish statutory accounts or file in another jurisdiction, those documents work the same way.
What do I get back?
A Doc with the year-over-year language changes, the stated strategy, the named risks and the segment story, each carrying its item and page. A Sheet holds the metric and disclosure comparison across periods, built to extend when the next filing lands. Evidence extraction does the same job for papers.
10-K Analysis for Competitive Intelligence
Fill in the form and your workspace opens with the work already underway.