How to Run an Evidence-Based Brainstorming Workshop

Most experimentation brainstorming sessions follow the same pattern. Someone books a room, the team gathers, a facilitator asks 'what should we test next?', and the loudest voices win. The ideas that emerge are driven by opinion, recency bias, or whatever a competitor launched last week. Very few are grounded in what the organisation already knows. An evidence-based brainstorming session starts with what you have already learned and works outward from there. The result is a pipeline of ideas that are connected to real findings, aligned to business objectives, and far less likely to duplicate work that has already been done.

Before you read this guide, try something. Pick an area of your product that your team has tested in the past, say the checkout flow or the onboarding experience. Now find every experiment that has been run in that area in the last eighteen months. The hypothesis, the result, the decision that followed, and what was learned. How long did that take you? Did you find all of them? Are you sure? If the answer took more than five minutes, or if you are not confident you found everything, the brainstorming guide below will be useful but the underlying problem is bigger than how you run workshops.

Before the Session

The quality of a brainstorming session is determined before anyone enters the room. Preparation is not optional.

1. Define the scope

Every brainstorming session needs a boundary. Without one, you get a sprawling list of ideas that nobody can prioritise because they span five different product areas and three unrelated objectives.

Set the scope by answering two questions: which business objective does this session serve, and which part of the product or journey are we focusing on? Share the scope with attendees at least 48 hours before the session so they can prepare.

Checkpoint: Can you state your team's current quarterly business objectives right now, without looking them up? If not, where are they documented? If you do not know where they are documented, that is the first problem to solve.

2. Gather the evidence

This is where most teams fall down. Before the session, collect: past experiment results in this area, relevant customer research, current analytics, and active experiments.

Checkpoint: Try assembling this list right now for your scoped area. How many places did you have to look? A spreadsheet, a testing tool, a Confluence page, someone's slide deck, a Slack thread from six months ago? If the answer is more than one place, your team will spend hours on this step every time you run a brainstorming session. If the answer is 'I asked someone who has been here a while', that knowledge leaves when they do.

Package this into a pre-read document and share it with attendees 48 hours before the session alongside the scope. One to two pages.

Checkpoint: Do you have a single view of every experiment currently running across all teams? If someone in your brainstorming session suggests an idea that another team is already testing, would you know?

3. Choose the right people

Six to eight people. Include people who bring different viewpoints: someone close to the data, someone close to the customer, someone who understands technical feasibility, and someone connected to the business objective.

Running the Session

Step 1: Set the context (10 minutes). Restate the scope and walk through the evidence summary.

Step 2: Identify gaps and opportunities (15 minutes). Before generating ideas, ask what the evidence does not cover. Where are the blind spots?

Checkpoint: When your team identifies a gap in your evidence, where do you log it? If the answer is 'someone writes it on a sticky note' or 'we just remember it', those gaps will be forgotten by next week.

Step 3: Generate ideas individually (10 minutes). Silent ideation. Each idea captured with: what are we testing, why do we think this will work, what business objective does this connect to, and where in the journey does this sit.

Step 4: Share and cluster (20 minutes). Share briefly. Group similar ideas. Name each cluster.

Step 5: Discuss and refine (20 minutes). For each cluster: has this been tested before, is this grounded in evidence or a hunch, is it feasible, what is the expected impact?

Checkpoint: When someone in the room says 'I think we tested something like this last year', what happens? Does the team have a way to verify that in the moment? Or does the conversation stall while someone searches through old documents, eventually gives up, and the group moves on without knowing?

Step 6: Prioritise (15 minutes). Using your agreed framework.

Step 7: Capture and assign (10 minutes). Who will write this up as a full experiment plan, and by when?

After the Session

Every idea generated should be logged in your experiment backlog, including those deprioritised or parked.

Checkpoint: Open your experiment backlog right now. Can you see the ideas from your last brainstorming session? All of them, including the ones that were parked? If not, where did they go?

Each idea should be linked to the evidence that inspired it. Track the conversion rate from idea to experiment to learning to decision.

Checkpoint: What percentage of ideas from your last brainstorming session became experiments? What percentage of those experiments had a decision recorded at the end? If you do not know the answer, you are not tracking the loop.

What Good Looks Like

A well-run evidence-based brainstorming session produces a shortlist of experiment ideas where each idea is connected to a business objective, grounded in evidence, free of duplication with past or active experiments, captured in a format that can be turned into an experiment plan, and assigned to an owner with a deadline.

If you worked through the checkpoints in this guide and found yourself reaching for multiple tools, asking colleagues for information that should be documented, or realising that you simply do not know the answers, that is worth paying attention to. The brainstorming method is straightforward. The hard part is having the infrastructure that makes each step possible without heroic effort.