drafty

Research summary template

A tight synthesis of what you found and what it means — generated from your notes, shared as a link stakeholders can react to inline.

What it is
A research summary is a one-to-two-page condensation of a larger study: the question you were answering, the key findings, and the recommended action. It's what a decision-maker reads instead of the full report.

The full report is for the archive. The summary is for the decision — the stakeholder with four minutes before their next call, the engineer who needs to know which finding to act on first. Most research summaries get this backwards: they're structured by method rather than by implication, so the headline insight arrives on page three after the reader has already stopped reading.

Generate it with your agent

Paste this into Claude, Cursor, or any agent — attach your notes, transcripts, or survey data:

claude
❯
Write a research summary from the notes below. Structure it as: 1) The question — one sentence on what we were trying to learn, 2) How we looked — method and sample size in two lines (no methodology deep-dive), 3) What we found — three to five findings, ordered by impact on our decision, each stated as a conclusion not an observation, 4) The one thing that changes what we build — the single most actionable insight, 5) Recommended next step — one clear action with an owner. Keep it under one page. Then publish it to Drafty so I can share a link and collect inline comments — no account needed to reply. [Paste your notes, transcripts, or survey results here]▌

See it on a real one

Live canvas — comment on any elementOpen ↗

What separates a summary from a report

A report documents everything you found. A summary argues for what to do next. The structural difference: a report organises by theme with supporting evidence; a summary leads with the highest-stakes finding and includes only enough evidence to make the recommendation credible. If someone can read your summary and not know what the team should do next, it's an abstract, not a summary.

The five sections that work

  1. The question — one sentence. "We wanted to know whether new users abandon onboarding because the steps are unclear or because they don't yet understand the value." Anchors everything that follows.
  2. Method in two lines — enough to trust the findings. "Eight 45-minute sessions with users who signed up in the last 30 days and hadn't completed onboarding." Not more.
  3. Findings ordered by impact — not by interview order. State each as a conclusion, not an observation: "Users don't reach step 3 because they don't know what step 2 is supposed to produce" — not "we observed confusion in step 2."
  4. The decisive finding — the one insight that, if ignored, will make the next quarter harder. If your findings feel equally weighted, you haven't prioritised yet.
  5. Next step with an owner — "PM to remove the optional phone number field before the next sprint." One action, one person.

Why research summaries get ignored

Findings fail to reach decisions because of delivery, not collection quality. The common failure modes: the headline finding is buried after three pages of method; observations replace conclusions ("users felt confused" vs. "the confirmation screen reads as an error state"); five equal-weight recommendations give no signal about priority.

Structure fixes this, not length. A one-pager with one clear recommended action gets read; a twenty-slide deck gets skimmed to the last slide.

Sharing matters too. A PDF emailed around generates reactions in email, Slack, and the next meeting — scattered and untraceable. A shareable link keeps comments on the exact finding they're about, and your agent can revise the document in place.

Summary vs. full report

Write the summary when you need a decision this week. Write the full report when the finding is contested and the methodology needs to be auditable, or when other researchers will build on your work.

For most rounds of maker or PM research — a usability check, a round of discovery interviews, a quick survey — the summary is the only document the team will actually read.

FAQ

What should a research summary include? The question you were answering, the method and sample in two lines, three to five findings ordered by their impact on a decision, and one recommended next step with an owner. Everything else belongs in the full report.

How long should a research summary be? One page, two at most. If it runs longer, it's a report. An exec or engineer should finish it in under three minutes — if it takes longer, it won't get read.

What is the difference between a research summary and an executive summary? An executive summary is a section of a longer document. A research summary is the standalone document you share instead of the report — for someone who won't read the full thing. The terms get used interchangeably, but the intent differs: an executive summary assumes access to the full report; a research summary doesn't.

Can you use AI to write a research summary? Yes — it's one of the highest-leverage uses of an agent. You supply the raw material (notes, transcripts, survey exports) and direct the prioritisation; the agent handles the structuring and drafting. The risk isn't that the agent gets it wrong — it's that you skip the step of telling it which finding matters most, and end up with a summary that's equally weighted across five points and actionable on none of them.