Most teams set up their rituals once, at the start of a project, and then live with whatever habits form. Standups settle into the same speaking order within a fortnight. Retros settle into the same format within a month. The team chat’s “question of the day” runs for nine days and quietly dies. This guide plans one quarter of rituals for a seven-person team, using three small tools together, and works out what randomness actually buys you over 13 weeks, including the places where it doesn’t give you the evenness people assume it does.
The quarter we’re planning
The team is seven people: four engineers, a designer, a QA lead and an engineering manager who also attends standup. They run two-week sprints. A 13-week quarter holds six full sprints plus a spare week at the end for planning and cleanup.
Counting the meetings:
- Standups: 13 weeks of 5 working days is 65 days. Drop the six sprint-planning days (the team skips standup on those) and one public holiday, and that leaves 58 standups.
- Retros: six, one at the end of each sprint.
- Question of the day: one per standup day, posted to the team channel before standup, so 58 of those too.
Three tools cover it: the Standup Order Randomizer, the Random Sprint Retro Format Generator and the Question of the Day Generator. Each has a setting that matters over a quarter, and each keeps its memory in one place, which we’ll get to.
Standups: a fresh order, and an honest look at “fair”
Paste the seven names into the Standup Order Randomizer, one per line. The list is saved in your browser, so tomorrow it’s already there. Click Today’s standup order and you get a numbered lineup.
Each name has a small running time next to it, built from Per-person minutes in the Options panel. The default is 2, so the first person shows ~2.0 min, the second ~4.0 min, and the seventh ~14.0 min. That number is when that person should be finished if everyone keeps to time. It is a cue, not a timer; nothing counts down. If the team agreed on 90-second updates, set it to 1.5 (the field moves in half-minute steps), and the last name reads ~10.5 min. Set it to 0 and the times disappear.
On days someone is out, delete their line before clicking. That edit is saved too, so remember to put them back the next morning. It takes five seconds, and it’s the one bit of upkeep the tool asks of you.
Now the arithmetic. A fresh shuffle every day means each person opens standup with probability 1 in 7 on any given day. Over the quarter that’s an average of about 8.3 openings each. But an average is not a promise:
| Over this stretch | What a fresh daily shuffle gives you |
|---|---|
| One day | Each person opens with a 14.3% chance |
| Two days in a row | Same opener both days: 1 in 7 |
| A 10-standup sprint | A given person never opens: 21% |
| A 10-standup sprint | At least one of the seven never opens: about 90% |
| The whole quarter (58) | Typical most-frequent opener: about 12 times |
| The whole quarter (58) | Typical least-frequent opener: about 4 or 5 times |
The last two rows come from simulating 100,000 quarters. Across 57 day-to-day transitions you should also expect the same person to open two days running about 8 times.
That spread is what fair randomness looks like over 58 draws, and it’s worth telling the team before someone notices that they’ve opened three days in a row and asks if the tool has a grudge. What the shuffle removes is the pattern: nobody is always first because they sit nearest the camera, and nobody is always last because they’re the manager. If your team genuinely needs everyone to open exactly as often as everyone else, that’s a rotation, and a fixed rotation brings back the predictability you were trying to lose. The older post on standup order goes into why the habit forms in the first place.
The tool keeps your ten most recent orders under Recent results, which is about two weeks of standups. If you want to check fairness over the whole quarter, copy each day’s result into a running note; the post on auditing picks over time has a method.
Retros: six formats, no repeats
The retro generator draws from 24 formats, from Start / Stop / Continue and Mad / Sad / Glad to Sailboat, Speed Car, Lean Coffee and Retro of Retros. Each draw shows the format’s structure and a short facilitation note.
The setting to use is No repeats until every format has come up. With it ticked, every draw is remembered in your browser, and the tool won’t repeat a format until all 24 have been used. Six retros a quarter means four quarters, a full year, before anything comes round again. Without it, six independent draws from 24 have about a 49% chance of producing at least one repeat inside the quarter, which is roughly a coin flip that the team will hear “didn’t we do this one already?”
Three details make this work better over a quarter:
- The checkbox and the memory are both remembered. Options are saved in your browser, so a box ticked in sprint one is still ticked in sprint two. The list of formats already used stays in the browser either way, and only matters when the box is ticked, so if anyone else uses the tool on that computer, check the box before you draw.
- A re-roll spends a format. With the box ticked, every draw goes into the memory, including one you reject. That is usually fine. If you draw Retro of Retros in sprint one, before there are three earlier retros to review, re-roll it; it will sit out until the other 23 have had their turn. Just don’t re-roll five times looking for a favourite, because those five are gone for the year.
- When all 24 have come up, the memory clears and the next draw is from the full set again. That draw has a 1 in 24 chance of being the format you ran last.
Draw the format at sprint planning, not five minutes before the retro. Some formats need preparation: Timeline Retro needs someone to sketch the sprint’s dates, Five Whys needs a chosen incident, and Appreciation Retro works much better when people know it’s coming. Two weeks’ notice turns a surprise into something a facilitator can run properly.
If a sprint went badly enough that the draw feels wrong (a production incident, say, when the draw gives you Appreciation Retro), override it. The guide to matching a format to the sprint covers when a deliberate pick beats a random one. The random draw is there to stop the default, not to forbid judgment.
The question of the day: pick a category per sprint
The Question of the Day Generator has three categories of 30 questions each: Work, Fun and Deep. Category set to Any draws from all 90. There’s a No repeats checkbox, and its memory is kept separately for each category setting: Work has its own list of used questions, Fun has its own, and Any has its own.
That separation matters for planning. Here are the options for 58 days:
| Approach | What happens over 58 days |
|---|---|
| Any, No repeats off | About 43 different questions; repeats are close to certain |
| Any, No repeats on | 58 different questions, with 32 left for next quarter |
| Work only, No repeats on | All 30 used by day 30, then the memory clears and they start again |
| Work in sprints 1, 3, 5 and Fun in 2, 4, 6, No repeats on | About 29 of each, no repeats in the quarter |
Any with No repeats on is the simplest. Its catch is that about a third of the Any pool is Deep, questions like “What’s something you believed five years ago that you don’t believe anymore?”, and those land differently in a seven-person channel at 9:15 than in a one-to-one. That’s the argument made in the post on deep questions. Alternating Work and Fun by sprint keeps the channel light, avoids repeats, and gives each sprint a slightly different feel.
One trap: because the memories are separate, a question used under Work can turn up again under Any. Pick one approach for the quarter and stick to it.
One browser holds all the memory
Every bit of state in this plan lives in the browser that ran the tool: the standup roster, the retro formats used, the questions used, and the recent results. Nothing is sent to a server, and nothing syncs between laptops.
For a team that rotates the facilitator role, that’s the main thing to plan around. If the engineering manager runs retro draws in sprint one and the QA lead runs them in sprint two, the QA lead’s browser has no idea what was drawn before. Two ways to handle it:
- One machine owns the rituals. The meeting-room laptop, or whoever is the permanent facilitator, runs all three tools.
- Keep a written log. A pinned message in the team channel with the six retro formats and the date of each. The facilitator ticks No repeats and re-rolls past anything already on the list.
Also note that Clear history in any of the three tools does more than empty Recent results. It also wipes that tool’s no-repeat memory. If someone clears it to tidy the page, the retro generator forgets the formats you’ve used.
The quarter on one page
| When | Ritual | Tool and setting |
|---|---|---|
| Each standup day, 9:00 | Post the question | Question of the Day, Work or Fun by sprint, No repeats on |
| Each standup day, 9:15 | Shuffle order | Standup Order Randomizer, Per-person minutes 1.5 or 2 |
| Sprint planning | Draw next retro’s format | Retro Format Generator, No repeats ticked |
| Sprint end | Run the retro | Format drawn two weeks earlier |
| Quarter end | Review | Six formats noted; decide whether to keep them in the memory |
What a quarter of this actually changes
None of these tools makes a meeting good. A standup with a random order can still run 25 minutes. A Sailboat retro can still produce no actions. A question of the day can still get one reply.
What they change is the small, unnoticed drift. After 58 standups, nobody has been structurally first or last. After six retros, the team has been asked six differently shaped questions about its work instead of one question six times. After 58 days of questions, the channel has had 58 different prompts rather than nine and then silence.
Those are modest wins. They’re also the kind a team only notices by their absence, about a month after the rituals went stale.