adr.zone

Compare decision-record formats

This table compares the five formats supported by the adr.zone generator. It is not a ranking of products or services—only document shapes.

Attribute comparison (subjective; adjust to your org)
CriterionNygardMADRY-StatementOutcome-FirstISO 42010 Companion
Primary goalOne clear decision, minimal ceremonyRepeatable section layout for search & toolingSingle reviewable sentence + fieldsAsync decision clarity with preserved reasoningConnect decision to architecture description (stakeholders, concerns, views)
Typical lengthShort (≈1–2 pages)Medium (sections expand with options)Very short in the Y sentence; fields add detailShort–medium (top 4 sections very short; lower sections clarify)Medium–long (explicit scope & traceability)
StructureStatus, context, decision, consequencesMADR: YAML meta, drivers, options, outcome + nested consequences/confirmation, per-option pros/cons, more info; or minimal variantOne sentence, then the same content as bulletsStatus, outcome, decision, primary tradeoff, why, decision boundariesSystem, stakeholders, concerns, view/viewpoint, decision, rationale, impact, traceability
Automation & indexBy convention; easiest with plain headingsStrong: tooling ecosystem expects MADR sectionsWeak; good for human scan, not machine indexStrong: structured fields for search, summary, and agent retrievalMedium: structured fields, not a standard file format
Learning curveLowMedium (more sections to learn)Low (once the sentence pattern is known)Low–medium (easy to read; authors learn to separate outcome from decision)Higher (42010 vocabulary)
Best whenYou want Git-first defaults and fast writeYou want linting, indexes, and ADR-Manager-style flowsYou need a line for stand-up, release notes, or a PR descriptionAsync review, distributed teams, decisions needing quick alignmentAudits, multi-stakeholder systems, or tying to architecture views

Which format for which team?

  • Nygard — minimal friction, fast to write. Best for teams that want Git-first defaults and low ceremony.
  • MADR — structured sections including per-option pros/cons; full or minimal template. Best when you need linting, search, or ADR-Manager-style workflows.
  • Y-Statement — a single reviewable sentence. Best for stand-ups, release notes, or PR descriptions where brevity matters.
  • Outcome-First — outcome, decision, and tradeoff visible before the reasoning. Best for async teams that need quick alignment or disagreement.
  • ISO 42010 Companion — explicit stakeholders, concerns, and views. Best for audits, multi-stakeholder systems, or governance alignment.

For the ISO vocabulary (not the same as “ISO 42010 the ADR file”), see ISO / IEEE 42010 and decision records.

Beyond Markdown files: imports & workflow

Teams rarely stop at a raw .md file. The same decision is often reviewed in Git, indexed in a portal, orsynced to a product that tracks ownership and follow-up. That landscape changes quickly, so this site does not list scores or “winner” products.

Representative classes of tools (with examples you can research yourself):

  • CLIs and repo scripts (e.g. Nygard-numbered flows, dotnet-adr) — good when everything stays in the terminal and Git.
  • Editor and GitHub-integrated UIs (e.g. form-based ADR managers) — good when non-git experts write ADRs and still merge via PRs.
  • Developer portals (e.g. Backstage plugins) — good when you need search across services and a single entry point, still backed by Git.
  • Workflow-centric products — any tool that imports or mirrors existing Markdown ADRs can sit beside this model; check whether the source of truth remains a repo you control, and what happens on export.

A living list of specific tools and templates is maintained by the ADR community at adr.github.io/adr-tooling. For more links and background, see our resources page.

Templates · Generator · ADRs and LLMs