Series: How to Run a Workshop People Remember (1/10)

I've sat through a lot of bad workshops. When I ask the facilitator afterwards what went wrong, the answer is almost never "my slides were ugly." It's always some version of the same confession: "I built the deck first and figured out the workshop along the way."

The deck is not the workshop. The deck is packaging. If you design the session in Keynote or Google Slides, you'll get a presentation with a bit of discussion bolted on — which is exactly the format people have learned to sleep through.

This series is about the whole job of running a workshop that people actually remember: design, materials, registration, the room itself, measurement, follow-up, and the boring-but-lucrative part — reporting to whoever paid for it. This first part is the foundation: how to design before you build anything.

Start with two sentences

Before you open any tool, write these two sentences on paper:

  1. The outcome: "By the end of this session, participants can ___." One verb, one observable behavior. Not "understand stakeholder mapping" — "can draw a stakeholder map for their own project and name their two riskiest relationships."
  2. The evidence: "I'll know it worked because ___." What will you see or measure that proves the outcome happened? A map on the wall per team? A pre/post score? A commitment written down before people leave?

If you can't finish sentence two, your workshop is a lecture with snacks. Everything else in this series — activities, QR screens, measurement, reporting — exists to make the evidence sentence cheap to collect.

The four-beat arc

Every workshop I've run or admired follows the same rhythm. Steal it:

Beat 1 — Open (10–15%). Low-risk warmup that gets voices into the room. It's not entertainment; it's establishing that speaking here is safe. A ninety-second pair exercise like Two Truths and a Lie does more than thirty minutes of icebreaker bingo.

Beat 2 — Explore (50%). The actual work, done by participants, not by you. This is where tools and props earn their keep — timed exercises, group artifacts, visible outputs. You talk the least in this beat. That's the point.

Beat 3 — Converge (20%). The debrief. This is where learning actually happens, and it's the beat most facilitators cut when they run late. (Part 8 of this series is entirely about this, because it deserves its own article.)

Beat 4 — Close (10%). Each person leaves with one commitment written in their own words. If they leave with nothing written, the workshop evaporates on the drive home.

From arc to run-of-show

Now assign minutes. A run-of-show is the workshop cut into timed blocks with an owner per block: not "team activity," but "Minefield, 20 min, pairs, needs 8 printed cards, debrief 5 min." When something runs over — something always runs over — the run-of-show tells you exactly what to cut without gutting the debrief.

I stopped writing slide outlines years ago and now design every session as a run sheet first. If you want to see what a finished one looks like, the folders on WorkshopSkill each ship with a minute-by-minute run sheet — that's the artifact I'm describing, filed and ready to compare against your own. There's also a dedicated guide there on how to design a workshop that goes deeper on outcomes and timing than this post can.

Where AI and tools fit in design (and where they don't)

I use AI at exactly two points in the design phase, and both are drafting, not deciding:

Sharpening the outcome sentence. Paste your draft outcome into ChatGPT (chatgpt.com) or Claude (claude.ai) with a prompt like:

"Here's my draft learning outcome for a 3-hour workshop: [outcome]. Rewrite it as one observable behavior per participant, list three ways I could see or measure that behavior during the session, and flag anything vague."

It's a superb vagueness detector. It will catch every "understand" and "appreciate" hiding in your sentence. It will not know your participants; you still own the call.

Pressure-testing the arc. Give it your four beats with minutes and ask: "Which beat is underweighted for a group of 25 who don't know each other?" The answers are generic but often embarrassing in a useful way — usually it points out that your Open beat is two minutes shorter than it should be.

For the plan itself, I keep it boring: one page in Notion or any doc with a table — beat, minutes, activity, materials, evidence captured. If the design doesn't fit one page, it won't fit the room.

What's coming in this series

The next nine parts walk the whole pipeline with the tools I actually use — mine and other people's, named and linked, so you can assemble your own stack:

  1. Stealing a run sheet: what a ready-to-run activity folder contains
  2. Registration and check-in that runs itself
  3. Opening the room without forcing an app on everyone
  4. Group work that doesn't collapse into silence
  5. Proving it worked: pre/post measures without the spreadsheet pain
  6. Letting an AI agent do the prep work
  7. The debrief: where learning actually happens
  8. From one-off workshop to monthly retainer (reporting clients read)
  9. The whole stack: a checklist from invitation to follow-up

Tools in this post:

Disclosure: I'm one of the makers of WorkshopSkill, QRCard AI, FormLM and ReportSkills — the tools from my own shop named across this series. Every claim about them matches what the products do today, and links go to the live pages. This series was drafted with AI assistance and reviewed before publishing.