Team & agility
What Scrum team size — and how it feels inside
AIIntroduction — the invisible metric for Scrum success
According to the Scrum Guide, Scrum teams should have between 3 and 9 members.
That sounds like a harmless guideline — until you have worked in all three orders of magnitude.
I have seen teams so small that in the daily you already knew what the others were going to say — and teams so large that the daily traded more small talk than progress.
I have seen how team size influences everything: from the mood in the retro to the likelihood that a story gets finished on time.
And it does not hang on the number alone. Whether people are T-shaped (broadly capable with one core competence) or pure specialists makes the difference between a team being flexible and resilient —
or sinking into chaos with every absence.
Let us dive into three everyday scenes:
- A small 3-person team
- A solid 6-person team
- A large, globally distributed 9-person team
The characters are the same — only the team size changes.
Everyone works remotely, using Jira for tickets, Slack for chat and Teams for meetings.
The characters
AI- Lena — product owner, structured, loves clear sprint goals
- Tom — senior dev, T-shaped, brings humour and experience
- Priya — QA engineer, finds every bug, always
- Max — frontend dev, creative, sometimes chaotic
- Sara — backend dev, specialised in databases, analytical
- Diego — DevOps, loves automation, hates manual steps
- Jin — junior dev, motivated, learns fast
- Nadia — backend dev, calm, precise, loves clean code
- Omar — data engineer, analytical, rather quiet, but on target in meetings
Scenario 1: the 3-person team — closely bound, but fragile
Team members: Lena, Tom, Priya
Just before the daily
AI08:57.
Slack blinks. Priya sends a GIF of a cat staring desperately at the clock.
"Daily in 3 min — and my tests are still red 🙈", she writes.
Tom is already in the Teams call. Camera on, coffee in hand.
Lena joins: "Morning you two — short and snappy today, right?"
In the 3-person team it is always snappy. Everyone knows what the others are doing.
Sometimes that can be almost too much closeness — you know the blockers, but also every frustration of the others.
How the sprint went
AIDay 3: A story escalates. An external API provider has changed its interface.
Tom has to rebuild everything.
Priya stops her tests to help — her own ticket is on ice.
The sprint goal? Already at risk.
Day 5: Lena moves two stories to "next sprint".
Nobody is surprised. But it still feels like a defeat.
Sprint Planning
It takes 20 minutes.
Lena: "We have 4 tickets we could start — but please only one at a time."
Tom and Priya nod. They know: if something escalates again, the sprint is done for.
How the 3-person team feels:
Close coordination, lightning-fast communication — but zero buffer.
One illness or one blocker is enough to topple the sprint goal.
Scenario 2: the 6-person team — the golden middle
AITeam members: Lena, Tom, Priya, Max, Sara, Diego
Just before the daily
08:59.
Max writes in Slack: "I'll be 2 min late — build is still running."
Sara: "No stress, we are not machines 😉."
The daily now takes 15 minutes.
Half the team no longer automatically knows what the others did. The daily is now the place to find that out.
How the sprint went
Day 4: Two stories are stuck.
Sara is waiting on a frontend API from Max. Max started working on another story in parallel yesterday.
"Why?", Lena asks in the chat.
"Because I felt like it at that moment… and the other thing was blocked", Max answers.
Diego is quietly optimising the CI/CD pipeline in the background — his work does not appear in any sprint goal.
Sprint Planning
AIThis time it takes 45 minutes.
There is more discussion.
Tom: "The sprint goal is too broad."
Lena: "Then tell me what we leave out."
Nobody wants to decide — in the end all the stories stay in.
How the 6-person team feels:
Enough buffer to absorb absences.
But communication is no longer automatic; it has to be organised actively.
Sprint goals can get watered down if nobody sets clear priorities.
Scenario 3: the 9-person team — globally spread, consensus as the brake
AITeam members: Lena, Tom, Priya, Max, Sara, Diego, Jin, Nadia, Omar
Just before the daily
08:00 in Berlin, 14:00 in Singapore, 22:00 in San Francisco.
Jin sits with headphones in the coworking space, Omar in the half-dark of his home office.
Lena opens the daily: "Okay everyone, we are all here…"
Tom: "Hold on, Nadia is still in another call."
The daily now takes 25 minutes.
There is small talk, plenty of status updates — and still you do not know who is stuck on exactly what.
How the sprint went
Day 5:
New tickets appear in Jira, started in the middle of the sprint.
Lena writes: "Please only start stories when others are finished."
Sara reacts with a neutral emoji — and starts a new ticket anyway.
The sprint goal?
More of a nicely worded headline for the stakeholder report.
Sprint Planning
AIIt takes 1.5 hours.
The time difference makes the discussion sluggish.
Everyone wants to get "their" stories in. Consensus takes forever — and often leads to mediocrity.
How the 9-person team feels:
High specialisation, but sluggish consensus.
Plenty of coordination, little focus.
A real sprint goal? Rarely. Motivation drops.
The turning point — FlightRoom
AIAfter a particularly frustrating retro ("we always end up talking about the same problems anyway…") Lena suggests running the next team workshop with FlightRoom instead.
90 minutes later the Teams call looks completely different:
on the screen a virtual cockpit.
Everyone has a role — captain, co-pilot, flight engineer, tower, ground technician.
The game master's voice sounds calm, but with palpable tension:
"Welcome to the Birgenair 301 scenario.
Back then the flight ended in disaster. Today your crew decides how it turns out."
The take-off goes smoothly. But soon the instruments start showing contradictory readings.
"I see airspeed too low!", Max calls.
"No, that is an error. We have to keep climbing!", Sara answers.
Time is short, the discussion gets heated.
Tom, as captain, demands: "Tower, we need every piece of data you have, now!"
Priya, in the tower, answers short and precise — no long explanations, only clear facts.
Decisions come fast, agreed, without long back and forth.
After the simulation there is quiet. Everyone feels how important open communication, clear calls and mutual trust are.
Lena says: "That showed us how we can work together better in the sprint too. No long discussions, but focused, direct, as a team."
The result after the FlightRoom
Three sprints later:
- Stories get finished instead of just started
- Sprint goals are reachable and motivating
- The mood is noticeably better — even in the large, distributed team
In closing & practical tips
AI- Choose team size deliberately:
3–4 people: lightning fast, but fragile
5–7 people: good balance, needs prioritisation
8–9 people: effective only with strong moderation and clear communication - Foster T-shaped skills:
Specialists matter, but without basic competence in several areas the team collapses at bottlenecks. - Structure remote work deliberately:
Clear meeting times, shared tools, short feedback loops - Prioritise shared goals:
One sprint goal that matters to everyone brings more than 5 individual goals - Use FlightRoom:
Train communication so that even large teams take decisions effectively and support one another