Prerequisite tree
You know the objective and the obstacles. Each obstacle gets an intermediate objective that overcomes it. This is implementation planning, not another debate.
Topic 02 · It’s Not Luck
The argument is the constraint.
Once delivery moves, a different class of problem shows up. Debt versus features. The big custom deal. “We need more people.” These are conflicts sitting on assumptions. Goldratt’s Thinking Processes are how you surface the assumption instead of compromising.
Comes after Flow. Six sections. Progress stays in this browser.
01
When the pipeline is no longer the problem
Endless debates about what to build, speed versus quality, a strategic account versus a coherent product. These are rarely capacity problems.
Behind a chronic fight is an unresolved conflict. Behind the conflict is at least one assumption that is not necessarily true. Challenge that assumption and you no longer have to pick between two bad options. The conflict evaporates.
02
Five boxes and a conflict
The cloud is the tool you will actually use. A shared objective, two necessary needs, two actions that cannot both be taken.
A · objective
Sustainable delivery of customer value
B · need
Ship features that grow revenue or engagement
C · need
Keep the system maintainable and changeable
D · action
Prioritise only new feature work this quarter
D′ · action
Give significant capacity to tech debt
In order to have “Sustainable delivery of customer value”, we must have “Ship features that grow revenue or engagement”.
Which assumption is doing the damage?
In order to… we must… Challenge the arrow, not the person.
- State the conflict. We must do D, and we must do D′.
- Name the need each action is protecting.
- Name the objective both needs serve.
- Write the assumption under each arrow.
- Challenge the weakest one.
- Find an injection that breaks it and still satisfies both needs.
03
Complaints are not projects
A current reality tree starts from what people are already unhappy about and asks why, until several symptoms share a root.
Pull requests that sit. Calendars full of meetings. A lot of starts and few finishes. Production surprises. Product and engineering frustrated with each other. Treated separately, that is five initiatives. Connected, it is often one policy.
Pull requests sit
Meetings fill the week
Many starts, few finishes
Core problem
Work is started faster than the constraint can finish it
Pick a complaint. Then pick the cause that actually produces it. Wrong arrows are how teams start five projects.
0 of 5 effects connected
04
Does the injection actually work?
A future reality tree starts from the proposed change and traces what it causes — including the harm.
Positive arrows are not enough. One side branch is the negative: the way a sensible rule gets people hurt, or gets quietly abandoned. Trim that branch with a second injection. Do not pretend the harm will not happen.
Injection
A work-in-progress limit of two
Desired effect
Lead time falls, because inventory falls
Negative branch
Urgent work cannot start
Trim
A named expedite, outside the limit
Injection
A work-in-progress limit of two items per person.
- 01Effect not traced yet
- 02Effect not traced yet
- 03Effect not traced yet
05
Only once you are ready to implement
Most of the value is in the cloud and the current reality tree. These two are for the moment you have an injection and need to land it.
Transition tree
The action sequence from here to there, especially when other people have to come with you. It answers the objection in the step, instead of collecting “alignment” in a meeting.
Objective: the limit of two is real on Monday
06
One conflict, half an hour
Pick one chronic argument. Build the cloud with the people in it. Do not boil the ocean into a tree nobody can read.
Primary tool
Evaporating Cloud
What you should leave with
A shared objective, one challenged assumption, a sequence instead of a stalemate.
How to run it
- One conflict. Not the whole backlog of grievances.
- Twenty to thirty minutes. Read each arrow: “In order to… we must…”
- List assumptions. Circle the ones that sound weakest when spoken.
- Injections that break a weak assumption and still protect both needs.
- If it is systemic, a short current reality tree from the complaints in the room.
- A future reality tree, including one negative branch, before you spend a quarter on it.
This week
0/5
Cheat sheet
- Chronic problems
- Usually an unresolved conflict, not a shortage of opinions.
- Assumptions
- Conflicts rest on them. Challenge the right one and the conflict can evaporate.
- Cloud
- Make the dilemma and the arrows visible. Read them aloud.
- Current reality
- Connect symptoms to one core problem. Stop launching a project per complaint.
- Future reality
- Check that the injection produces the effects you want, and trim the harm.
- Win-win
- Prefer an injection that raises throughput and still protects both needs. A compromise that satisfies neither is a failed cloud.
Check yourself
5 questions. No timer.