The Research Organization Checklist (2026): Organise by Use, Not by Source

Article Metadata

By Justkay, Documentary Filmmaker and Founder of Storyflow Published September 2, 2026 · 15 min read · Research

Breadcrumb

Home > Blog > Research > Research Organization Checklist 2026

Table of Contents

1. Why Source-Based Filing Fails
2. Check 1: Capture
3. Check 2: The Note Itself
4. Check 3: Structure by Question
5. Check 4: Retrieval
6. Check 5: The Citation Trail
7. 8 Tools for Organising Research
8. How We Evaluated These Tools
9. Tool Reviews
10. The Two-Tool Answer
11. FAQ: Organising Research
12. The Bottom Line
13. Author
14. Related Reading

Key Takeaways

• File by the question a note answers, not by the source it came from. You retrieve by use.
• Write one line at capture saying what you will use this for. It is the highest-value habit here.
• Separate quotation from paraphrase from your own thought, visibly, or you will misattribute later.
• Zotero for citations. Nothing else here replaces it and everything else pairs with it.
• If retrieval takes longer than re-reading, the system has failed regardless of how tidy it looks.

Why Source-Based Filing Fails

The default structure is a folder per source: a folder for each book, each interview, each set of documents. It feels correct because it matches how the material arrived.

The problem is asymmetry. Material arrives by source and is retrieved by use. Writing a chapter, you do not think "what did the Henderson book say"; you think "what do I have about why the factory closed". If the structure is by source, answering that means recalling which sources discussed it, which is exactly the memory work the notes were supposed to replace.

The consequence compounds. Research you cannot retrieve gets re-read, and re-reading is the single largest time sink in a research-heavy project. People assume the fix is better notes; the fix is usually a different top level.

Organising by use means the top level is your project's questions. Under each, the notes that bear on it, whatever their source. The source lives on the note as metadata rather than as the folder.

Check 1: Capture

Capture fails by friction and by delay. A system that takes ninety seconds to file something means you stop capturing.

• One inbox, not five. Everything lands in the same place first
• Capture takes under thirty seconds or it will not happen consistently
• Capture works offline, in a library, on a phone
• The source is recorded at capture, always, because reconstructing it later costs far more
• A page or timestamp is recorded, not just the source
• Inbox processing happens weekly, not "when I have time"
• Nothing stays in the inbox for more than a week. Either it becomes a note under a question, or it is deleted

Deleting is part of the system. Research that never gets deleted becomes a pile, and a pile is functionally the same as nothing, because searching it costs more than re-finding the material.

Check 2: The Note Itself

Five things per note, and the second is the one that changes everything:

1. The content, in your own words wherever possible.

2. What you will use it for. One line. "Supports the argument that the closure was decided before the announcement." Written at capture, while you still know why you took it.

3. The source, with page or timestamp.

4. The type, visibly marked: is this a direct quotation, a paraphrase, or your own thought?

5. What it connects to, if anything.

Item two is the highest-value habit in this article. Notes without it become a wall of true statements with no indication of why any of them mattered, and reconstructing that is nearly as expensive as re-reading the source.

Item four prevents a specific and serious failure. Three months later, a paraphrase you wrote and a quotation you copied look identical, and the mistake is either plagiarism or a misattributed argument. Mark quotations unambiguously at capture: quotation marks, a prefix, a different field. Whatever you use, use it every time.

Check 3: Structure by Question

• The top level is the questions your project must answer, not the sources
• Between eight and twenty questions for most projects. Fewer is too coarse; more is a taxonomy nobody maintains
• Questions are written as questions, which forces specificity
• Every note sits under at least one question
• A note that fits no question is a signal: either the question list is incomplete, or the note is not for this project
• Questions can change, and restructuring should be cheap, which is an argument for links over folders
• A separate place for material that is definitely relevant and you cannot yet say why

The last item matters. Research produces material whose significance is not yet clear, and forcing it into a question prematurely either distorts it or discards it. A holding area that gets reviewed is better than a false classification.

On tags versus folders: tags scale better because a note usually bears on more than one question. Folders force a single home and produce duplication. If your tool supports links or tags, use them for questions and keep folders for sources.

Check 4: Retrieval

The check nobody runs, and the only one that measures whether the system works.

Pick a question your project has to answer and try to retrieve everything you have on it. Time it.

• If it takes under two minutes, the system works
• If it takes ten, the structure is wrong, usually because it is by source
• If you find yourself re-reading a source to find something you know you read, the system has failed, and no amount of tidying fixes that. Restructure by question instead

Also check:

• Search finds notes by the words you would actually use, not the words the source used
• You can list every note bearing on one question in one action
• You can get from a note back to its source in one step
• Nothing important lives only in your memory of where it is

Check 5: The Citation Trail

• Every source has a full reference captured at first use, not reconstructed later
• Page or timestamp on every quotation
• A reference manager holds the bibliography, not a document
• The citation style is chosen at the start
• Direct quotations are distinguishable from paraphrase, permanently
• You can produce the bibliography without opening the sources
• Archived copies of anything online, since links rot and a dead link in a bibliography is a real problem

Use Zotero, or an equivalent, from day one. Retrofitting a bibliography at the end of a book-length project is days of work and it is the single most avoidable cost in research writing. It is free and it takes an hour to learn.

8 Tools for Organising Research

Prices are publicly listed as of early 2026 and change often. Ratings weigh whether the tool supports structuring by question, retrieval speed, citation handling, and cost over a project that runs years.

How We Evaluated These Tools

Four criteria:

1. Can you structure by question rather than by source without fighting the tool?
2. Retrieval speed, timed on the two-minute test above.
3. Citation handling, which is where most general note tools are weak and where the cost of being wrong is highest.
4. Cost over a multi-year project, where one-off purchases and free tools behave very differently from subscriptions.

Tool Reviews

1. Zotero

The correct tool for citations and sources, free, and nothing else in this list replaces it. It captures references from a browser in one click, stores PDFs, handles every citation style, and produces a bibliography from a document without you touching the formatting. If you do one thing from this article, use Zotero from day one.

It is not a note-taking system. Its notes are adequate and you will want them somewhere else, which is why the pairing below exists.

Pricing: free, with paid storage tiers.

Pros: the best citation handling; free; one-click capture; every style; group libraries.

Cons: weak as a note system; dated interface; storage costs money past the free tier.

2. Obsidian

The strongest tool for the notes themselves, and structuring by question is what it is built for: a note per question, links from every note that bears on it, and backlinks showing you the whole set in one view. Local files you own, free, and durable in a way cloud tools are not.

It has a setup cost and a plugin habit, and its citation support requires a plugin and configuration.

Pricing: free for personal use.

Pros: the best linking; files you own permanently; free; excellent search.

Cons: setup and plugin effort; citations need configuring; no built-in collaboration.

3. Scrivener

The right answer when the research and the writing should live together, which for a book-length project is a strong argument. Research in the binder, the draft beside it, split-screen between a source and the paragraph it supports. One-off price across a project that runs years.

Citation handling is weak, so pair it with Zotero, and its search is worse than Obsidian's.

Pricing: around $60 one-off.

Pros: research and writing in one project; excellent for long-form; one-off price.

Cons: weak citations; weaker search and linking; no collaboration.

4. Notion

The most approachable way to structure by question: a database of notes with a question property, filtered views per question, and the source as another field. For someone who does not want to learn a system, this is the shortest path to a correct structure.

Citation handling is essentially absent, and it is slower to search at volume.

Pricing: Free for personal use; Plus $10 per user per month billed annually.

Pros: the easiest correct structure; good writing; usable free tier; collaborative.

Cons: no citation management; slower at large volumes; per-seat for teams.

5. Devonthink

Built for large document archives, and its search and classification across thousands of PDFs is better than anything else here. If your research is mostly documents rather than notes, it earns its price.

Mac-focused, expensive, and more machinery than most projects need.

Pricing: from around $99 one-off.

Pros: unmatched at large document archives; strong search and auto-classification.

Cons: cost; platform-limited; heavy for a note-shaped project.

6. Zotero plus Obsidian

The pairing most serious researchers converge on, and it is free. Zotero holds sources and citations; Obsidian holds notes structured by question, linked to the Zotero entries. It requires a plugin and an afternoon of setup, and after that it is the strongest combination in this article.

Pros: best citations and best notes together; free; files you own.

Cons: setup effort; two tools to learn; no collaboration.

7. Storyflow

Storyflow is our product and it is ranked last, which is the honest placement. It is not a research tool. It has no citation management, no reference capture, no PDF handling, no bibliography and no full-text search across documents, and anyone organising research should use Zotero plus one of the note tools above.

Its narrow use is upstream: laying out the eight to twenty questions your project must answer and seeing whether they cover the ground, which is a spatial problem and the one part of this article that is not about notes. Once the question list exists, the notes belong in Obsidian, Notion or Scrivener.

Storyflow is paid-only during early access; its Free plan launches before the end of 2026, and anyone a paid member invites to a board joins free now.

Pricing: Plus $7.99 per month billed annually ($9.99 monthly); Pro $14; Max $39.

Pros: good for shaping the question list before the notes exist; whole-board AI context.

Cons: no citations, references, PDFs or bibliography. It does not organise research.

8. Google Docs

For a small project with thirty sources, a single well-structured document with headings per question genuinely works, and its search is decent. It has no citation management and no linking, and it stops scaling around the point a project becomes book-length.

Pricing: free.

Pros: free; zero setup; excellent collaboration and commenting.

Cons: no citations or linking; poor at scale; no structure beyond headings.

The Two-Tool Answer

Almost every research project converges on two tools, and knowing that early saves a lot of switching.

One for sources and citations. Zotero, effectively always. It is free, it is the standard, and retrofitting it at the end costs days.

One for notes structured by question. Obsidian if you want links and ownership, Notion if you want approachability, Scrivener if the writing should live with the research.

One rule to make the pair work: the note lives in the note tool and links to the source in Zotero. Never copy the reference into the note by hand, because two copies of a reference diverge and the wrong one ends up in the bibliography.

The Bottom Line

Change the top level from sources to questions and most retrieval problems disappear. Write one line at capture saying what you will use each note for, mark quotations unambiguously and permanently, and process the inbox weekly with deletion as a normal outcome. Use Zotero from day one, pair it with one note tool, and run the two-minute retrieval test occasionally, because a system that looks tidy and cannot be searched is not a system.

Author

Justkay is a documentary filmmaker and the founder of Storyflow. The quotation-marking rule is on this list because he once could not tell, three months later, whether a striking sentence was his or somebody else's.