reviews · one team · offline + online
Start · Stop · Continue.folder
One verb per card, behaviors never people, and no card survives without an owner and a date. The smallest retrospective with real discipline.
// kit v2: a stage-ready HTML deck (presenter drawer, run-of-show timer, built-in QR codes) · poker-spec print-and-cut cards with backs and blanks · an A3 activity poster · manuals in A4 and US Letter with bookmarks · three license tiers (CC BY free / single event / organization).
The start-stop-continue retrospective, rebuilt as a complete workshop folder: printable verb cards (one per person per pile, with the card rules printed on each), a silent-writing round, live clustering, dot-voting and an owner-plus-date round that turns the loudest pain into one auditable commitment per pile.
The props are poker-spec print cards (63.5 × 88.9 mm, bleed and cut guides included) — START, STOP, CONTINUE plus a timing card — sized so one sheet covers a team. The run sheet carries the CONTINUE cap that moves honesty to the hard piles, and the rescues for pre-blamed cards, camp votes and vague commitments.
the run sheet
// this is the guide file, word for word — read it here, run from it here. the .md at the bottom of the page is the same text, offline.
Folder:
start-stop-continue· Category: reviews · one team of 4–8 · 30 min · offline + online · print props · assessment-typeLicense: CC BY 4.0
The retrospective that fits in one wall: every card is exactly one verb — start something, stop something, keep something. The structure is small; the discipline is real: no card survives without an owner and a date.
Objectives
By the end of the activity, the team will have:
- Seen its own working patterns <mark>sorted into three honest piles</mark>.
- Turned the loudest pain into one action with one owner and one date.
- Practiced the retro discipline: cards are behaviors, never people.
Setup
Space: one wall or whiteboard; online, a shared board with three columns.
Materials:
| Item | Source |
|---|---|
| 1 × START card per person | start-stop-continue-prop-cards.pdf (card 1) |
| 1 × STOP card per person | start-stop-continue-prop-cards.pdf (card 2) |
| 1 × CONTINUE card per person | start-stop-continue-prop-cards.pdf (card 3) |
| 1 × timing card | start-stop-continue-prop-cards.pdf (card 4) — for you |
The card rule — say it out loud: "One behavior per card. Never a person's name. 'Stop cc-ing the whole team' — not 'stop cc-ing Dana'."
Grouping rule: one real team, 4–8 people, including its lead. Larger teams split into working pairs first, then merge boards.
Minute-by-minute script
| Time | Phase | Facilitator does | Participants do |
|---|---|---|---|
| 0–3 | Frame + card rule | State the rule. Deal the three cards per person | Read cards, think |
| 3–11 | Silent writing (8 min) | No talking while writing — quality over speed | One behavior per card, all three verbs |
| 11–18 | The wall (7 min) | Invite reads; cluster duplicates live | Post cards, hear echoes, merge |
| 18–24 | Vote (6 min) | Three dots each: what matters most this cycle | Dot-vote across all piles |
| 24–30 | Owner + date (6 min) | "Top card per pile gets an owner and a date. Say them out loud." | Commit aloud; photograph the wall |
The wall photo and the three commitments are the retro's artifacts — see start-stop-continue-debrief.md for the 10-minute close.
Known failure modes & rescues
- Cards arrive pre-blamed. "Stop being late to standup" — aimed at one person everyone can name. Rescue: rewrite live, with the room's permission: "the behavior is arriving unprepared — stop that." Behaviors only, every time.
- Everything lands in CONTINUE. Praise is safe; START and STOP stay empty. Rescue: cap CONTINUE at three cards per person before the wall goes up — scarcity moves honesty to the hard piles.
- The vote splits into camps. Dots pile along team lines; the retro becomes a referendum. Rescue: silent dot-voting only, and re-cluster the top five into one shared shortlist before the owner round.
Variations
- Sprint-length version: 20 minutes, one verb per person per sprint — rotate verbs weekly.
- Remote: shared board, cards as stickies, dot-vote with emoji, photo becomes an exported PNG.
- Paired version: for low-trust teams, write cards in pairs first — every card carries two names, which halves the exposure.
the debrief sheet
// the debrief questions, in full — the .md version prints on one page for the room.
8 questions, 3 failure modes, one rule: the wall photo is the contract.
Use with
start-stop-continue-guide.md. Runs 10 min after the owner round.
The 8 questions
1. The pile sizes.
"Look at the wall: which pile is biggest, and what does the shape say about this team right now?"
→ An all-CONTINUE wall is a team at ease or a team afraid. Ask which.
2. The hardest card.
"Which card was hardest to write — the START, the STOP, or the CONTINUE? Why that one?"
→ STOP cards are usually hardest; asking why surfaces the fear behind the behavior.
3. The echo cluster.
"Which cluster gathered the most duplicates? What is it like to know five people wrote the same STOP?"
→ Echoes are the team's quiet consensus finally made visible.
4. The card nobody posted.
"Was there a card you wrote and pocketed? No need to read it."
→ Heads nodding is enough. The pocketed cards tell you the retro's safety ceiling.
5. The vote and the silence.
"Which high-vote card got discussed the least? What's underneath it?"
→ Sometimes the room votes for a card precisely to avoid talking about it.
6. The daily version.
"Between this retro and the next one, where will the wall's three verbs show up without a formal retro?"
→ Transfer question. Expect: 1:1s, channel threads, sprint planning.
7. The three commitments.
"Read the three owner-plus-date cards aloud. Do all three still feel real, right now?"
→ One may wobble under the room's eyes — renegotiate it on the spot, honestly.
8. The photograph.
"Who has the wall photo? When does it get pulled up next — before or after the next retro?"
→ Closing question. The photo is the contract; schedule when it gets audited.
3 known failure modes of the debrief itself
- The retro becomes a venting session. STOP cards get debated one by one, time dies, nothing is owned.
Rescue: the vote already chose — debate belongs to the top card per pile, nowhere else.
- The lead answers everything. The most senior voice interprets every cluster.
Rescue: hand the interpretation to the card-writers: "author reads, room listens" as a rule.
- Commitments are vague to be safe. "Improve sync" — owner, date, no verb anyone can audit.
Rescue: the one-line test: can someone who missed the retro tell if it happened? If not, rewrite.
If you are running the pre/post measure (start-stop-continue-assessment.json), note these observables:
| Observable | Points to |
|---|---|
| STOP pile size vs. CONTINUE pile size | Honesty level |
| Duplicate clusters form without prompting | Shared pain surfacing |
| All three commitments pass the one-line test | Commitment quality |
The report your team receives is a self-comparison across cycles, never a ranking against other teams — keep it that way.
who it's for · what it leaves behind
Who is the Start·Stop·Continue retrospective for?
Any working team of 4–8 that ends a cycle: sprint ends, project phases, quarterly reviews. It works for agile and non-agile rooms alike — the three verbs need no framework — and runs fully remote on a shared board. Bring the team lead in; a retro the lead skips is a retro the team performs.
What results does it produce?
The wall photograph and three auditable commitments — one per verb, each with an owner and a date that survives the meeting. A pre/post team-level delta from the assessment preset measures honesty, behavior-over-person framing and commitment quality. It measures the team's review habits, not facilitation skill.
Can my AI agent use this folder?
Yes. Every folder ships a SKILL.md skill file: frontmatter for matching, plus runnable prep commands — fetch the props, pull the debrief, build the live pre/post measure with formlm-cli. The three-step workflow (discover, install, run) lives on the skills page, every folder ships a one-command zip bundle, and the machine index is skills.json.
Is it really free?
Yes — CC BY 4.0, no account, no email wall inside a folder. Download it, run it, copy it, rebrand it for your own workshop. Attribution keeps everything free.
make it live
The folder's paper pre/post measure is built for rooms with nothing but markers. If you want the digital version — the same 6 items as a live form, each participant's before/after delta computed automatically — two ways:
for AI agents
$ formlm-cli smart generate --input "Pre/post measure for the Start-Stop-Continue retrospective: 6 Likert items on honesty, behavior-over-person framing and commitment quality — the same 6 items before and after the activity, so each participant's delta is computed automatically"
Step 4 of the agent workflow: npm install -g @formlm/cli, log in once, and the prep above runs itself. The form and its data live in your own workspace.
— or, with a mouse —
Open the free pre/post training quiz template on FormLM
and paste the 6 items from start-stop-continue-assessment.json — the template already runs the before/after comparison and scores it.
// optional, not required — the paper version in this folder is complete on its own. free account on their side.
put it on the wall
The folder ships a single-file HTML deck that presents full-screen from one laptop. To run the activity on the room's big screen — submissions land on the wall in real time, the countdown and the spotlight stay in your hands, and the whole room joins with one scan — the scan-and-project companion is QRCard-AI:
See the trainer live-screen mode on QRCard-AI — one QR code on the wall, participants join and follow from their own seats.
// optional, not required — the deck in this folder already runs full-screen on a single laptop.
take this folder with you
// every file downloads as a plain file. no account, no email wall inside a folder.