Your Team Agreement is Probably Dead
Let me guess - you created a working agreement during a retro six months ago. It's sitting in Confluence. Nobody looks at it. Here's how to fix that.
The Working Agreement Problem Nobody Talks About
I've seen this pattern at every company with 20+ engineers
You know how it goes. New team forms, or someone reads about working agreements on Medium. Big workshop with sticky notes. Everyone agrees on "respond to PRs within 4 hours" and "update tickets before standup."
Fast forward three months. PRs are sitting for days. Nobody updates Jira. The team agreement might as well not exist.
Why Team Agreements Fail:
No Connection to Reality
You wrote "review PRs in 4 hours" but have no idea what your actual review time is. Can't improve what you don't measure.
Document Dies in Wiki
After the workshop, it goes to Confluence to die. Out of sight, out of mind. Nobody checks it during daily work.
Generic Template Nonsense
You copied some agile working agreement template. "Be respectful." "Communicate openly." Meaningless fluff that doesn't drive behavior.
Zero Accountability
Who's following the agreement? Who's not? Nobody knows. Without visibility, agreements become suggestions.
Working Agreement Examples That Actually Work
From real teams who track their agreements (not just write them)
1. PR Review Agreement (Most Common Bottleneck)
This is where most teams get stuck. Here's what actually works:
• First review within 4 hours: Not "ASAP" - specific time
• Small PRs get priority: Under 200 lines = same day
• Author picks reviewers: Stop auto-assigning to busy seniors
• One approval for non-critical: Two for database/auth changes
• Track it: Current average: 6.2 hours (improving from 8.5)
Reality check: If you can't see your actual review times, this agreement is worthless.
2. Technical Debt Agreement (The One Nobody Makes)
Every team has debt. Few teams agree how to handle it:
• 20% rule: Every sprint includes 20% debt work
• Boy Scout rule: Touch it, improve it (no exceptions)
• Debt gets a number: Track debt work vs feature work
• Monthly debt review: Not "when we have time"
• Visible metrics: Code coverage, complexity trends
In my experience, teams that don't track debt allocation always end up doing less than 10%.
3. Meeting Agreement (Because Meetings Suck)
Stop wasting everyone's time:
• No agenda = cancelled: Send it 24 hours before
• Default to 25 minutes: Not 30 (Parkinson's law)
• Three person rule: More than 3? Must have clear purpose
• Silent start: First 5 minutes reading shared doc
• Action items or it didn't happen: Who does what by when
4. On-Call Agreement (The Burnout Preventer)
Because nobody wants to get paged at 3am for non-critical stuff:
• Define "page-worthy": Customer-facing down = page. Internal tool = wait
• Comp time is real: Weekend incident = Monday morning off
• Handoff requires sync: Not just "good luck, you're on"
• Post-mortem everything: Even if it was "just a blip"
• Track MTTR by person: Some people need more support
How to Make Team Agreements Actually Stick
Hint: It's not about better workshops
Step 1: Start with One Thing That Hurts
Forget comprehensive agreements. Pick your biggest pain point.
For most teams at 20-50 devs, it's PR review time. Your PRs are probably sitting for days. Start there. Make one specific, measurable agreement about reviews.
Step 2: Make It Visible Daily
If you have to go look for your metrics, you've already lost.
Put your agreement tracking where work happens. Slack. Your dashboard. Standup. "Yesterday our average PR time was 5.2 hours. Our agreement is 4."
When people see the number every day, behavior changes. Magic.
Step 3: Celebrate Wins, Don't Punish Misses
This isn't about surveillance. It's about getting better together.
When someone hits the PR review target, call it out. When the team improves week-over-week, celebrate. When you miss? "OK, what's blocking us? How can we help?"
The moment this becomes punitive, it's dead. I've seen it happen.
Step 4: Review and Adjust Monthly
Your agreement isn't carved in stone. It's a living thing.
Every retro, look at the data. Is 4-hour PR review realistic? Maybe 6 is better right now. Did adding the 20% tech debt rule help? What does the data say?
Adjust based on reality, not aspiration.
What Actually Happens When You Track This Stuff
Real numbers from teams that went from wiki agreements to tracked agreements
PR Review Time: 4 days → 6 hours
Just making it visible changed behavior. No process changes, no new tools. People saw the number and started reviewing faster.
Sprint Completion: 60% → 85%
When teams see they're at 60%, they start asking why. Turns out half the stories weren't ready. Fix that, completion jumps.
Tech Debt Work: "Some" → 22%
Before tracking: "We work on debt when we can." After: "We're doing 22%, target is 20%." Suddenly debt work is real.
Team Conflicts: Way Down
Hard to argue about whether someone's pulling their weight when the data is right there. Conversations become about systems, not people.
The uncomfortable truth: Most teams have no idea if they're following their agreements. They wrote "respond to Slack within 2 hours" but nobody knows their actual response time. They agreed on "20% tech debt" but can't tell debt work from feature work.
Without measurement, agreements are just wishes.
Dead Simple Team Agreement Template
Start here. Add more only when the basics work.
The Only 5 Agreements You Need to Start:
1. Code Review Agreement
"We review PRs within [X] hours during working hours"
Track: Average PR review time, PRs waiting 24+ hours
2. Meeting Agreement
"Meetings have agendas 24 hours before or they're cancelled"
Track: Meeting time per week, meetings with clear outcomes
3. Tech Debt Agreement
"We spend [X]% of our time on debt/refactoring"
Track: Actual % of commits tagged as debt work
4. Communication Agreement
"Core hours are [X] to [Y] for sync work"
Track: Response time, after-hours messages
5. Sprint Agreement
"We commit to what we can finish, not what we want to start"
Track: Sprint completion rate, carry-over stories
Pro Tip from Someone Who's Done This Wrong:
Don't create 20 agreements in your first workshop. You'll track none of them.
Start with ONE that really hurts. Get that working. Add another. Five good agreements beat twenty ignored ones.
And seriously - if you're not going to measure it, don't bother writing it down.