What's the Best Note-Taking App for Researchers?

I built my first researcher-grade note system in college on index cards. I migrated to Roam in 2019, to Obsidian in 2020, and have been on Obsidian since. The reason I write about it now — five years in — is that the specific use case of researcher-style note-taking has matured into a clear category-leader, and the gap between #1 and #2 is wide enough to be worth a published verdict.

The use case is not “take notes about your meeting.” It is “build a knowledge graph that grows for a decade and connects to itself in non-obvious ways, with cross-references that survive job changes, OS migrations, and the eventual shutdown of whatever startup made the editor you currently like.”

For that use case, Obsidian is the answer. The reason is local-first markdown plus the most-developed backlinking implementation in the category. The reasons matter individually and they compound.

The criteria that define this category

Researcher-style note-taking has four defining requirements:

1. Durable backlinks. A note’s value compounds as it gets linked from other notes. The graph is the product. If your note app’s backlink implementation degrades over time (links break on rename, links break on file move, links break on moving disks) you are not building a researcher’s notes; you are building a slowly-corroding pile.

2. Local-first storage. The expected useful life of a researcher’s notes is longer than the expected operational life of any cloud-only note-taking startup. Roam was a great product in 2019; the company has since had multiple existential moments. Local-first means your data outlives the editor.

3. Plain-text format. Markdown is the right format because every text editor on Earth can read it, indefinitely, without a vendor’s cooperation. Proprietary block formats (Notion, OneNote) are the format equivalent of a cloud lock-in.

4. Plug-in ecosystem. Researcher-style notes include things like citation management, automated daily notes, image annotation, and templated note structures. The likelihood that any vendor’s first-party software covers your specific researcher workflow is low; the likelihood that a community plug-in covers it is much higher. The plug-in ecosystem is the multiplier.

Obsidian wins all four. Roam wins #1 (it pioneered the model). Logseq wins #1, #2, and #3 (it’s the open-source Roam-style alternative). Notion wins #4 within its own ecosystem but loses #1 and #2 by design. Apple Notes wins #2 and #3 but doesn’t have the cross-reference fidelity to win #1.

Why Obsidian wins on the integration

The reason Obsidian wins as a category — and not just “ties on three of four criteria with Logseq” — is the maturity of the ecosystem. The plug-in count is not the right metric (more plug-ins isn’t intrinsically better). The right metric is whether the specific plug-ins you need for academic work exist, are maintained, and integrate cleanly with each other. For Obsidian, they do. The Citations plug-in handles BibTeX import. The Dataview plug-in lets you query your notes like a database. The Calendar plug-in renders daily notes as a calendar. The Excalidraw plug-in does whiteboard-style drawings inside markdown. None of these are first-party; all of them are mature, maintained, and free.

How I tested

I am the test. I have run a researcher-grade note system in Obsidian for five years (~14,000 notes, ~38,000 internal links). I have also rebuilt portions of my system in Roam, Logseq, and Notion as a comparison exercise — most recently in February 2026, when I rebuilt my last 500 daily notes in each platform and compared the result.

The Roam rebuild produced a system that worked but cost $13/month for what I could do for $0 in Obsidian (or $4/month with Obsidian Sync). The Logseq rebuild produced a system that was 80% as smooth as Obsidian’s and would not have been a regression — Logseq is genuinely good. The Notion rebuild produced a system that was clearly Notion: gorgeous-looking, but where the daily-notes flow required a database with templates that produced friction every morning. Notion is not a researcher tool.

The case against Obsidian

The collaboration gap is the strongest case against Obsidian for a specific reader: someone whose research workflow involves co-editing notes with a team in real time. Obsidian was built for individuals; teams have to adopt different tooling (often a Git repo, sometimes a Notion supplement) for the collaborative parts of their work. If real-time co-editing is your dominant workflow, Notion or even Google Docs is a better answer than Obsidian.

For everyone whose research notes are an individual project — most academics, most independent researchers, most knowledge workers building their own systems — Obsidian is the right answer.

Why the runners-up didn’t win

Notion is the team-collaboration knowledge base. It is excellent for what it is. For individual researcher-grade note-taking, the database-block model and the cloud-only storage are structural mismatches.

Roam Research pioneered the model and remains a credible product. The cloud-only storage and the 3–10x pricing are the structural mismatches. Roam is the right answer if you specifically prefer Roam’s UI.

Logseq is the open-source Roam-style alternative and the close-second to Obsidian. If you are committed to open-source-only tooling, Logseq is the answer.

Apple Notes is the right answer for quick capture. It is not a researcher’s note system.

What to do next

If you don’t have a researcher’s note system and you think you should: install Obsidian, create a vault, write a note, link it to another note, and then keep going for two months. The system is not the editor; the system is the daily practice. The editor matters because it gets out of your way and lets the daily practice work.

If you have an existing system in Roam, Notion, or Apple Notes and you’re considering migrating: the migration cost is real. Roam’s export to markdown is decent but not perfect. Notion’s export is a mess of HTML and CSV. Apple Notes’ export is non-existent. Build a 90-day plan, expect to lose ~5% of links in any migration, and have a fallback plan if the new system doesn’t fit.

If you’re already on Obsidian and reading this for confirmation that you picked right: you did. Don’t re-pick. Build the system.