Session 6: Facilitator Notes

Duration: 60 min · Deliverable: experiment plan with hypotheses, before/after comparisons, success criteria, rollout

Before the session

• I've reviewed each student's improved repo and have a sense of how much they've shipped.
• I know the canonical DORA definitions cold. If a student asks about "Elite vs High vs Medium" tiers, I can answer.
• The handout (handout.md) is in chat at the start. It's the format we work through together, not homework.
• Everyone has their course logbook (/logbook) open. All of today's work gets recorded there live as we go.

Time budget

Block	Mins	Slides
1. Hypothesis format + DORA intro	10	1–7
2. Revisit baselines	5	8–13
3. Write hypotheses	15	14
4. Compare before/after	15	15–17
5. Rollout strategy + peer review	10	18–21
6. Session 7 preview	5	22–23

Block-by-block

All core work today is done live, together, in session. I demonstrate the format, we draft each part as a group, and students record everything in their logbook (/logbook) as we go. Nothing is assigned as take-home; anything beyond the live drafting is optional polish.

Block 1. Slide 6 (the three-sentence format) is the operational lesson; students will use this format constantly going forward. Have them note the format in their logbook. Reinforce that hypotheses are falsifiable.

Block 2 (baselines). Quick. We open their Session 1 baseline and their Session 3 + 4 deltas together. Lay them out. Some students will discover gaps; handle those in chat, don't slow the session. Cue: record the before/after numbers in the logbook as we go.

Block 3 (workshop, 15 min, live together). We write three hypotheses together in the strict format. Push for absolute numbers, not %. The most common mistake: vague verbs ("improve," "reduce"). Force a number and a time window. Each student records their three hypotheses in the logbook live.

Block 4 (workshop, 15 min, live together). We work through the before/after numbers, then the hypothesis + DORA mapping, as a group. Walk the room; students will pattern-match to each other once they see the first one. Cue: record this in the logbook as we go.

Block 5 (rollout + peer review). Quick on rollout strategies (phased vs big bang). We draft each rollout live and record it in the logbook. Then a peer review: trade logbooks with a neighbor and find one weakness in their proposal. Models real PR review behavior.

Block 6. Preview Session 7. Make sure students know the logbook carries forward: in Session 7 we draft the governance docs together and add them to the logbook, and in Session 8 their proposal comes straight out of it. Continuity, not homework. The templates in workshops/07-standards/starter/ can be skimmed beforehand, optionally.

Anticipated questions

"Some of my improvements are too small to measure individually, what do I report?"

Group them. "Bundle of DevEx fixes saved ~5 min per onboarding × 3 onboardings per month = 15 min/month."

"Can I claim a % when I don't have the baseline number?"

No. Absolute number with a known baseline, or describe the improvement qualitatively. Never invent a baseline.

"What if my change actually made something worse?"

Real engineering. Report it. The proposal in Session 8 has a "risks and what we observed" section for exactly this.

"Is one week enough to run a rollout experiment?"

Usually no. For real metrics, two weeks minimum. We'll cite that on the rollout plan.

If you run short

• Cut block 5 peer review; do it async.
• Compress block 1 to 5 min; skip the DORA tier discussion.
• Never cut blocks 3 and 4. The hypotheses + numbers are the deliverable.

Post-session

In chat: "Everything we drafted today is in your logbook (/logbook). Keep it; in Workshop 8 your proposal comes straight out of it. Optional: reread your hypotheses and tighten any that aren't testable."