Skip to content

Experiment Readout Drafter

Example prompt: "When I mark an experiment as Complete in our experiments Google Sheet, pull the hypothesis and the result numbers, draft a readout in Notion covering what we tested, what we saw, and what I think we should do, and post a short heads-up in #product so the team knows it's ready to review."

The Problem

Experiments get run and then never written up. The team agrees on a hypothesis, the engineers ship the variants, the results come back in a sheet, and then the PM means to write the readout but it sits in the backlog while the next experiment starts. Three months later nobody can remember what the test was meant to prove, which variant won, or whether the change was rolled out — and the next experiment ends up testing something the team already tested.

How GloriaMundo Solves It

We build a workflow triggered by a row in the experiments Google Sheet being marked Complete. An integration step reads the row's hypothesis, variants, success metric, and result numbers, plus any linked Notion notes. An LLM step drafts a readout structured the way the team likes to read them — what we tested, why, what the numbers showed, what we think it means, and a recommended decision (ship, kill, iterate). An integration step writes the draft into the team's Notion "Experiments" database as a new page linked to the sheet row, and a final integration step posts a short heads-up in the #product Slack channel with the page link. Glass Box preview shows the draft and the recommended decision before anything is written or posted.

Example Workflow Steps

  1. Trigger (webhook): Fires when a row's Status column is changed to Complete in the experiments Google Sheet.
  2. Step 1 (integration): Read the row's hypothesis, variants, success metric, sample size, and result numbers.
  3. Step 2 (integration): Pull any linked notes from the team's Notion experiments database for context.
  4. Step 3 (llm): Draft the readout — what we tested, why, what the numbers showed, what it means, recommended decision.
  5. Step 4 (integration): Create a new page in the Notion "Experiments" database with the draft, linked back to the sheet row.
  6. Step 5 (integration): Post a short heads-up in #product Slack with the page link and the recommended decision.

Integrations Used

  • Google Sheets — source of the trigger and the experiment results
  • Notion — where the readout draft is filed and where prior context lives
  • Slack — where the team is notified the readout is ready to review

Who This Is For

Product managers running more than one or two experiments a month where the act of writing them up keeps slipping, and product teams whose institutional memory of "what did we try last quarter" is already patchy. Especially useful where the experiments sheet is the system of record but nobody reads it.

Time & Cost Saved

A reasonable experiment readout takes a PM 45-60 minutes — re-reading the hypothesis, looking up the result numbers, working out what to recommend, writing it up. This workflow turns that into 10-15 minutes of reviewing and editing a draft. The harder-to-measure saving is the readouts that get written at all rather than left in the backlog and forgotten.