Learn

Research workflow: a complete guide

The research problem nobody talks about

Most guides to research focus on finding good sources. That's the easy part. The hard part is what happens after: reading carefully, understanding arguments, connecting ideas across sources, and eventually turning all of that into something you've produced yourself.

The typical failure mode isn't bad sources. It's a pile of saved articles, annotated PDFs, and highlights that never gets synthesised into anything. You read, you mark, you save, and then the research sits in a folder while you stare at a blank page trying to reconstruct what you actually think.

This is the Collector's Fallacy, a term coined by Christian Tietze: the mistake of treating collection as a substitute for learning. Saving an article is not the same as reading it. Highlighting a passage is not the same as understanding it. Annotating a paper is not the same as knowing what to do with it. The collecting feels productive. The synthesis is the actual work.

A good research workflow is designed to prevent this: to ensure that the gap between source and output is bridged by deliberate processing, not just accumulation.

Two modes of research

Research happens in two distinct modes, and recognising which one you're in matters for how you approach it.

Systematic research is structured and comprehensive. You define a question, identify relevant databases or sources, search with controlled terms, filter results, and work through them methodically. This is how literature reviews in academic contexts are typically done. The goal is coverage: you want to be confident you've found the relevant material, not just the material that happened to surface.

Exploratory research is networked and iterative. You start with a question or interest, find one useful source, follow its references and citations, discover adjacent work, and let the research develop organically. This is how most creative and knowledge work research actually happens. The goal is depth and connection, not coverage.

Both modes are legitimate. Systematic research suits formal literature reviews, evidence synthesis, and situations where completeness matters. Exploratory research suits most writing projects, learning new topics, and investigating questions where the boundaries aren't predefined.

Many research projects use both: a systematic pass to establish what the established knowledge is, followed by exploratory reading to develop your own thinking about it.

Step 1: capture without falling into the collector's fallacy

The collection phase is where the collector's fallacy most often strikes. A few principles that help:

Capture the minimum. When you find a potentially useful source, note the title, author, and why you think it's relevant. Don't save the full article until you're ready to read it. A reading list of fifty items you've vetted for relevance is more useful than five hundred bookmarked articles you'll never open.

Separate "interesting" from "relevant." Something can be interesting without being relevant to your current question. The interesting article about a tangentially related topic goes in your general reading list or second brain, not in your active research folder. Be deliberate about the difference.

Keep your reading list visible. An unseen reading list doesn't get read. In Fabric, your reading list, saved articles, and PDFs can all be in one searchable space. In PARA, sources for an active project live in the project folder rather than a separate "to read someday" pile.

Set a reading session, not a collecting session. The collector's fallacy is partly enabled by treating collection as a task that can be completed. Saving twenty articles in an afternoon feels like progress. Schedule reading time separately from collecting time, and hold them distinct.

Step 2: reading and literature notes

Active reading for research is different from reading for pleasure. The goal isn't to enjoy the text or even to understand it fully in the moment. The goal is to extract what's useful for your question and connect it to what you already know.

Read with your research question in mind. Every source should be interrogated from a specific angle: what does this say about my question? What evidence or argument does it provide? What would I need from this to make a case? Reading without a guiding question produces general understanding but rarely useful material.

Annotate as you read. Highlight passages that directly address your question. Write margin notes: reactions, connections to other sources, questions the passage raises. If you're reading a PDF digitally, annotations that are searchable and linkable are more useful than highlights that just mark the text.

Write literature notes. After reading a source (or a significant section), write a brief note: what was the source's main argument? What specific points are relevant to your question? Any direct quotes you want to preserve, with page numbers. This is different from a summary: you're not describing the whole source, you're capturing what matters for your purposes.

Literature notes should be in your own words where possible. Paraphrasing forces you to understand the idea well enough to express it differently, which is different from re-reading a highlighted sentence and assuming you've processed it. This principle comes from the Zettelkasten tradition but applies broadly.

Be honest about what you don't understand. A passage you've highlighted but can't explain in your own words is a gap in your understanding. Note it as a question to resolve, not a piece of evidence to use.

Step 3: synthesis

Synthesis is where most research workflows break down. You have literature notes from ten sources. You need to turn those notes into your own argument or output. The gap between these two states is where people get stuck and stay stuck.

A few approaches that bridge it:

Organise notes by concept, not by source. Instead of "Notes on Source A, Notes on Source B," create concept clusters: "Evidence for X," "Arguments against Y," "Definitions of Z." When you read a new source, you're not adding a new note to the stack: you're adding to (or complicating) existing concepts. The Zettelkasten principle of concept-oriented notes is the formalisation of this habit.

Write synthesis notes early. Don't wait until you've finished reading everything before you start synthesising. After reading a few sources, write a short note on what they collectively say about your question. What do they agree on? Where do they diverge? What seems uncertain or contested? These synthesis notes evolve as you read more, but starting early prevents the pile-of-sources problem from developing in the first place.

Use your research question as an organising structure. If your question has sub-questions or components, organise your notes under those. "What do we know about X?" "What's contested about Y?" "What evidence exists for Z?" Notes filed under questions are closer to output than notes filed under sources.

Talk it through. One of the most underrated synthesis techniques: explain your current understanding of the research to someone who isn't familiar with it, or to an AI assistant. The act of explanation forces you to identify what you actually know versus what you've stored without processing, and it often reveals the gaps and the through-line simultaneously.

Write before you're ready. The most reliable way through the synthesis bottleneck is to start writing even when you don't feel fully prepared. Write a rough, bad version of what you think the argument is. Then read it, identify where it's wrong or weak, and return to your notes to fix those specific gaps. Writing from an incomplete draft is more productive than collecting more sources in the hope that clarity will arrive.

Step 4: from notes to output

By the time you have good literature notes and some synthesis notes, the outline often writes itself. The synthesis notes, organised by concept and question, become the sections. The literature notes provide the evidence and specifics for each section.

A few things that help this final stage:

Build an outline from your synthesis notes, not from scratch. Don't start with a blank outline. Start with your concept clusters and synthesis notes and let the structure emerge from what you actually have. The outline is a reorganisation of existing material, not a new document.

Let your argument drive the selection of sources. The sources are there to support and challenge your argument, not the other way around. If a source is interesting but doesn't contribute to the argument you're making, it can wait for another piece.

Keep a "parking lot" for material that doesn't fit. Good research produces more than any single output can use. A note for "things I found but didn't use this time" prevents you from either stuffing the output with extraneous material or losing things you might need later.

Where research notes live

In a PARA setup: active research for a current project lives in the project folder. Research on an ongoing topic or interest lives in Resources. Completed projects with their research archive in Archives.

In a Zettelkasten: literature notes from sources have their own permanent home alongside your permanent idea notes. The two types of notes are kept distinct: literature notes document your encounter with a source, permanent notes develop your own ideas from it.

In Fabric: PDFs, web clips, literature notes, and synthesis notes all live in the same searchable library. When you're writing, searching by concept rather than by source title surfaces relevant material from across your reading without requiring you to remember which source covered what.

Frequently asked questions

How do I avoid going down rabbit holes?

By keeping your research question visible and asking it of every source you encounter: does this address my question? If yes, read it. If no, decide quickly whether to save it for later (it's actually interesting and might be useful for something else) or skip it. The rabbit hole usually starts when interesting becomes confused with relevant.

How many sources is enough?

For academic work, the field and the type of output determine this, and your supervisor or the submission guidelines will often specify. For other types of research, the practical answer is: enough that you've encountered the main positions, arguments, and evidence, and enough that new sources are no longer changing what you think. Diminishing returns on new sources is usually a signal that you're ready to synthesise and write.

What's the difference between research notes and book notes?

Book notes are focused on a single source and what you take from it. Research notes are organised around a question and what multiple sources collectively say about it. Book notes are about your encounter with a specific text; research notes are about developing your understanding of a topic. The two overlap but serve different purposes.

How do I handle sources I've collected but never processed?

Declare research bankruptcy on the ones that are old or no longer relevant: archive them rather than letting them sit as a permanent source of guilt. For the ones that are still potentially useful, schedule a specific block of time to read and take notes, not to organise. If you can't find time to read them, they weren't actually that important to your question.

Can AI help with research synthesis?

Yes, with appropriate care. AI assistants can help surface connections across your notes, answer questions about material you've already captured, and push back on arguments you're developing. The AI assistant in Fabric can answer questions about your own library of sources: "What do these papers say about working memory in ADHD?" or "What are the main objections to X argument?" This is different from using AI to generate the research itself, which bypasses the understanding you need to produce your own argument.

How do I handle research across a long project?

Maintain a running synthesis note that you update as you read new sources. Each time you read something, you don't just add a new literature note: you update the synthesis. What has changed in your understanding? What's now confirmed, complicated, or challenged? The synthesis note grows more refined over time rather than the pile of individual notes growing larger.