Guide

Note Taking App for Researchers and PhD Students: Notes You Can Still Trust in Year Three

A research project outlives the notes most people take for it: by year three you cannot tell where a claim came from or why you made a decision in year one. This page says which tool fits which researcher (and what should stay in your reference manager), then gives a setup for literature notes, a research log and a weekly routine that keeps both usable.

Updated 2026-10-27

5 examples

Workflows that fit how you work

Concrete setups for the situations this kind of work keeps producing.

1.Second-year PhD student with 140 papers and no memory of most of them

When it fits: Use this when your reference manager is full, your PDF highlights are unread, and your supervisor asks what the literature says about something you know you read.

Leila, PhD in public health, year two. 140 papers in Zotero, highlights in most of them, and a literature review chapter due in March 2027.

Setup. Zotero stays the reference manager. Each paper she actually uses gets one literature note, written after reading, never during:

# Okafor 2024, school meal programs and attendance
Citekey: okafor2024meals
Claim (my words): free breakfast raised attendance ~2 pts in rural schools only
Method: diff-in-diff, 312 districts, 2015-2022
Doubt: no control for transport changes
Links: [[School attendance]], [[Diff-in-diff]]

The citekey ties the note back to Zotero. Concept notes like [[School attendance]] collect every paper through backlinks. When her supervisor asked which papers found urban effects, she asked the AI chat and got an answer citing her own literature notes, each one a link she could open and check.

2.Computational researcher reconstructing why a sample was dropped

When it fits: Use this when a reviewer or co-author asks why you made an analysis decision eight months ago and the answer is somewhere in a script comment, a chat message or nowhere.

Mateus, postdoc in computational linguistics. Corpus of 2.1 million posts, a dozen cleaning steps, and a reviewer asking why 14% of accounts were excluded.

Setup. A research log as one dated entry per working day in his daily note, short and factual:

## 2026-10-21
- Dropped accounts with < 20 posts (n = 41,208, 14.2%)
  because: per-account estimates unstable below 20, see [[Threshold test]]
- Tried language-ID v2: worse on code-switched posts, reverted
- [ ] Ask Ana if 20 matches the 2025 paper @due(2026-10-24)

Every exclusion gets a "because" and a link to the note with the evidence. When the reviewer's question arrived, searching "dropped accounts" found the entry, the threshold note and the follow-up with Ana. The reply took twenty minutes instead of a re-analysis.

3.Fieldwork with a paper notebook and strict ethics rules

When it fits: Use this when you take handwritten field notes, your ethics approval limits where participant data can go, and typing up notebooks never happens.

Grace, anthropology PhD, eight months of fieldwork. Handwritten notebooks, pseudonyms for every participant, and an ethics protocol that names where identifiable data may be stored.

Setup. She photographs notebook pages each evening and attaches them to a dated field note. The text is extracted from the photo, so the notebook becomes searchable without typing it up. Before that, she checks the protocol: pages with real names or locations stay in the approved storage only; pages with pseudonyms and observations go into notes. Each field note gets three typed lines:

## Field note, {{date}}
- Setting: market, Saturday, approx. 40 vendors
- Pattern: credit extended only to "regulars" (3rd time seen)
- Links: [[Informal credit]], [[P07]]

She proofreads extracted text against the photo before quoting it. A misread word in a quote is a misquote.

4.Writing the background section and checking the numbers in it

When it fits: Use this when a chapter or paper draft is full of figures you wrote from memory, and you want them checked before your committee or a reviewer does it for you.

Tobias, final-year PhD in energy policy. Background chapter drafted in notes, with claims like "solar was 5% of US electricity in 2023" and "the IRA passed in August 2022", written months apart.

Setup. He keeps each chapter section as a note from the research notes template, where every claim has a source and a "last checked" date. Background verification compares the factual claims in the note against the web and proposes a diff where a figure looks wrong or outdated, which he accepts or dismisses. It flagged two figures that had been revised since he wrote them. It does not replace checking the claim against the paper you cite: he treats every flag as a prompt to reopen the source, not as the answer.

5.Early-career academic with a grant, two courses and four reviews due

When it fits: Use this when research is one job among several and the reading for a grant proposal keeps getting interrupted by teaching and service.

Dr. Nkem Adeyemi, assistant professor of education. A grant due January 15, 2027, two courses, and journal reviews arriving every few weeks.

Setup. One desk for research, with the proposal as a project note and the reading as an annotated bibliography: full citation, two-sentence summary, one sentence on how it supports the significance section. Deadlines live as tasks in the notes they belong to:

- [ ] Draft significance section @due(2026-12-01)
- [ ] Review for J. Ed. Research, MS 26-0412 @due(2026-11-10)

Today's daily note shows carried-over tasks above the editor, so twenty spare minutes between classes go to the next unchecked item instead of deciding what to do.

Which tool fits which researcher

Honest version first: most researchers need two tools, and the notes app is the second one.

References, citations and PDF annotation. A reference manager. Zotero is free and open source; the Better Notes add-on extends its notes if you want everything in one place. No notes app on this list replaces it.
Local Markdown files, links and plugins. Obsidian, free for personal use. Community plugins connect it to Zotero. Good if you enjoy configuring your tools.
Freeform pages with sketches, PDFs and clippings. OneNote, free and cross-platform.
Asking questions of a fixed set of PDFs. Gemini Notebook (formerly NotebookLM) grounds its answers in documents you upload. Strong for a single reading set, less suited to years of your own evolving notes.
Markdown notes with links, AI answers cited from your own notes, and verification of factual claims. Scribelet. Web and iOS, free plan with daily AI limits, Pro at $8 a month. It is not a reference manager, does not annotate PDFs and does not format citations. "Citations" in its chat means links to your notes, not a bibliography.

The three problems researchers have with notes

Claims lose their source. A sentence in your notes says "attendance effects are rural only". Is that Okafor, your inference, or a seminar comment? Without a source on every claim, notes become a liability at writing time.

Year-one notes are unusable in year three. They were written before you knew what the thesis was about. Folders built around your first framing do not fit the final one. Links to concepts age better than folders, because concepts survive reframing.

Decisions are not recorded. Exclusion criteria, coding scheme changes, the reason you dropped a variable. Reviewers ask about these, and memory does not hold them for eighteen months. A research log with one line of "because" per decision is the cheapest insurance in a PhD.

The setup (one afternoon)

One desk per project or thesis.
Literature notes: claim, method, doubt, citekey, links. Read literature notes for the reasoning.
Concept notes: often one line each. Their value is the backlinks.
A research log in the daily note. Dated, factual, with a because for every decision.
A project note per chapter or paper, linking the literature and log entries it relies on. Build the review itself from the literature review template.

The weekly routine (30 minutes)

Ten minutes: turn the week's reading into literature notes. Anything you highlighted but cannot summarize in one sentence gets reread, not noted.
Ten minutes: reread the week's log. Make sure every decision has a because and a link to its evidence.
Five minutes: link new notes to concepts, and add a concept note for anything that came up three times.
Five minutes: review verification suggestions or stale flags on notes with figures, and update Last checked dates.

Data, ethics and where notes should not go

Your ethics approval and your institution's data policy outrank any app's feature list. Identifiable participant data usually belongs in the storage your protocol names. Notes can hold pseudonymized observations, analysis decisions and literature. Before using any AI feature on research material, check whether your institution has rules on sending data to AI providers. Bringing your own API key is optional in Scribelet and lets you choose the provider.

Setup tips

Write the claim in your own words
A literature note that copies the abstract is a second abstract. One sentence of claim in your words, one of method, one of doubt. If you cannot write the doubt, you have not finished reading.
Keep the citekey in every literature note
The reference manager owns the citation; the note owns your thinking. A citekey line joins them, so a note can always be traced to the exact source.
Give every analysis decision a because
"Dropped accounts with fewer than 20 posts because estimates were unstable" is a sentence your methods section and your reviewers will need. "Dropped small accounts" is not.
Concept notes before topic folders
A note per concept ([[Diff-in-diff]], [[Informal credit]]) collects every paper and field note that mentions it through backlinks. Folders force a paper into one topic; links let it sit in four.
Date what you checked
Add Last checked: YYYY-MM-DD to notes with figures or claims about the state of a field. In year three you will want to know which numbers are from year one.

In the app

How to do this in Scribelet

Create one desk for the project and paste the research notes and literature review templates into notes, then save each as a template from the "..." menu. Research Notes is also one of the five built-in templates.
Link papers to concepts with [[Concept]]. The backlinks panel on each concept note lists every paper and log entry that touched it.
Attach PDFs and photos of notebook pages to notes. A vision model extracts the text and a caption, so the content is searchable and usable by the AI. Annotation stays in your PDF reader.
Ask the AI chat questions across your notes, such as "which papers I read found urban effects", and get answers with citations to your own notes. The free plan includes AI with daily limits; BYOK is optional.
Turn on background verification (Pro) for notes with factual claims. It compares them against the web and proposes diffs you accept or dismiss.
Import an Obsidian vault if you already have one, and export everything to Markdown whenever you want.

Start with the quick start if you are new to Scribelet.

Frequently asked questions

What is the best note-taking app for researchers?

Usually two tools: a reference manager (Zotero is free) for citations and PDF annotation, and a notes app for thinking. For the notes side, Obsidian suits researchers who want local Markdown files and plugins, including Zotero plugins. OneNote suits freeform pages with PDFs and sketches. Scribelet suits researchers who want AI answers cited from their own notes and verification of factual claims. Check current pricing on each vendor's page.

Do I still need Zotero if I use a notes app?

Yes. Scribelet and most notes apps are not reference managers: they do not format citations or build bibliographies. Keep Zotero (or Mendeley or EndNote) for references, and put a citekey in each literature note to connect the two. The literature notes article covers the handoff.

How should a PhD student organize research notes?

One literature note per paper you use, one concept note per idea that recurs, a dated research log, and a project note per chapter or paper. Link instead of filing into folders. The for students page covers the coursework years if you are still taking classes.

What is a second brain for researchers?

A set of notes that answers questions you will ask later: what did this paper actually claim, which papers disagree, why did I drop that sample. It needs claims in your own words, links between ideas and a log of decisions. The software matters less than writing those three things down.

Can I use an AI notes app with participant data?

Check your ethics approval and institution's data policy first; they decide where identifiable data may go. Scribelet encrypts notes at rest with per-desk keys and lets you bring your own API key, but that does not replace your protocol. Many researchers keep identifiable data in approved storage and only pseudonymized notes in a notes app.

Start the notes your year-three self will thank you for

Create a project desk, write literature notes with a citekey and a doubt, and keep a dated research log. When your supervisor asks what the literature says, ask your notes and check the cited answers.

Start your research desk

or get the app

Download on theApp Store

Free plan, no credit card. See pricing

Research

Research Notes Template

Research notes go wrong in two ways: you cannot tell where a claim came from, or you cannot tell whether it is still true. This template ties every finding to a source and a confidence level, keeps observation apart from interpretation, and records when you last checked.

Research

Literature Review Template

A literature review is an argument about what is known, built from sources, not a list of summaries. This template takes you from scope and search, through a source matrix, to one section per theme with agreement, disagreement and gaps. Pick the example organized the way your sources group.

Research

Annotated Bibliography Template

Each entry is a full citation in your assigned style followed by a short paragraph that summarizes the source, judges it, and says where it fits in your argument. Copy the template, then use the example closest to your course and citation style.

Research

Lab Notebook Examples and Template

A good lab notebook entry lets someone else repeat your work and see what you saw: what you meant to do, what you actually did, the raw numbers, the calculations, and what you will do next. Below are five complete entries from different kinds of labs and the template they share.

Article

Literature notes

From the Scribelet blog

Priya, second-year economics. Four lectures a day, slides posted afterwards, a problem set every Friday. Her problem was never capture; it was that 60 pages of notes a month were never opened again until finals.
Recommendation. Typed notes in one note per lecture from the lecture notes template, titled `ECON 201, L14, 2026-10-22`, tagged `#econ201`. During the lecture she writes fragments and marks anything she did not follow with `??`. Within 24 hours she spends ten minutes turning the `??` marks into questions at the top of the note and links each concept that came up before with `[[Elasticity]]`. The weekly review is a Sunday pass through the week's notes asking the AI chat "what are the open ?? questions in #econ201 this week" and answering them from the textbook. By week 12 the questions list is her revision guide.

Guide

Best Note-Taking App for Students: Setup, Templates and a Weekly Routine

The app matters less than the system. This page tells you which app fits which kind of student in one honest section, then spends the rest on the setup and the weekly routine that keep lecture notes, readings and lab work usable by exam week.