--- title: "Research report template – Display.dev" description: "An HTML research report – TL;DR, findings with confidence levels, a reliability-rated source sheet, and recommendations. Publish it to a URL in one click, no account." --- # Research report template An HTML research report – TL;DR, findings with confidence levels, a reliability-rated source sheet, and recommendations. Publish it to a URL in one click, no account. - Format: HTML - Detail page: https://display.dev/templates/research-report - Raw template file: https://display.dev/template-sources/research-report.html - One-click anonymous publish (no account): POST the file to https://api.display.dev/v1/public/artifacts as multipart/form-data (`file` field) ## Agent prompt Write a research report on this question on display.dev. Fetch https://display.dev/template-sources/research-report.html as the starting point and keep its structure: TL;DR, numbered findings each with a stated confidence level, a source sheet rating each source's coverage and reliability, recommendations tied to findings. Fill it with the actual research – cite sources, state confidence honestly, keep the file self-contained (inline CSS, no external requests). Publish it with the display.dev MCP publish tool if connected. Otherwise POST the file to https://api.display.dev/v1/public/artifacts as multipart/form-data (`file` field, optional `name`) – no account needed. You'll get back `previewUrl` (the link to share) and `claimUrl` (one-shot link to take ownership). Reply with both. ## What this template is for Evidence, not vibes. Every finding carries a stated confidence level, and the source sheet rates each source's coverage and reliability – so a reader can tell a replay-backed number from a nine-person interview theme. Recommendations reference the findings they rest on. This is the structure that keeps agent-led research honest: an agent fills it fast, and the confidence and source fields force it to show its work. ## How teams use it - **The output format for research agents.** Send the prompt with a question; the agent returns this report with its sources rated. Weak confidence levels tell you exactly where to dig before deciding. - **A decision record.** Publish behind your company's Google or Microsoft sign-in and the report becomes the citable reference for why the roadmap moved – with the evidence attached. - **Reviewed in comments.** Domain experts challenge a finding inline; the revision publishes to the same URL. ## Template source ````html Research report – onboarding drop-off

Research report

Why new users drop off before their first project

14–21 July 2026 · sources: session replays (n=214), churn interviews (n=9), funnel data (30 days) · author: research agent, reviewed by Sara L.

TL;DR Most drop-off happens at the empty workspace, not at signup. Users who create anything within 10 minutes retain at 3× the rate of those who don't – and the single biggest blocker is not knowing what a good first project looks like.

Findings

1. The empty state is the cliff

62% of users who churn in week one never created a project. Replays show a median 47 seconds on the empty workspace before exit – most never open the create dialog. Confidence: high (replays + funnel)

2. Examples beat instructions

Users who opened the sample project were 2.4× likelier to create their own within the session. Only 18% found it – it sits behind an unlabeled icon. Confidence: high (funnel)

3. "Set up later" is a myth

Of users who skipped workspace setup intending to return, 9% ever did. Interviewees described the skip as "quietly deciding no". Confidence: medium (interviews, n=9)

Source sheet

SourceCoverageReliability
Session replays, 14–21 Jul214 new-user sessionsHigh – full sample of the week
Churn interviews9 users, churned within 14 daysMedium – self-selected responders
Funnel data30 days, all signupsHigh – instrumented events

Recommendations

````