Peak Constraints

How do I find the bottleneck in my business?

Describe your business once as a flow: pick a model, set the outcome (for example orders per month), check the steps and enter capacity and load for each step. Peak then shows the step that limits throughput, with confidence and evidence.

1. 01Open Peak Constraints and choose “Describe system”.
2. 02Pick the model that fits best and set the outcome per month.
3. 03For each step, enter how much it can handle (capacity) and how much arrives (load).
4. 04Answer the three questions about wait times and gut feeling. The diagnosis appears right away.

How do I bring values into Peak Constraints via CSV?

In the system editor under “Apply values from CSV”. One row per step, with columns step, capacity, load, throughput, queue, utilization, lost, separated by semicolon, comma or tab. German and English column names both work. Peak matches the rows to your steps by name; it reports unknown steps but doesn't create them.

1. 01Open Peak Constraints → System.
2. 02Paste the CSV into the field or click “Choose file”.
3. 03Click “Apply” and read the message about which steps were updated.
4. 04Click “Save and recalculate” so the diagnosis uses the new values.

How do I work out what more capacity would bring?

On the “Bottleneck” page under “What if”. You choose a step, capacity or demand, and a new value. Peak shows throughput, contribution margin per month and which step limits afterward. Nothing is saved.

How do I create a resolution plan for the bottleneck?

Under “Resolve”, after you've confirmed the bottleneck. The plan has a goal in one sentence, the type of intervention, an owner, a target metric with baseline and target value, a deadline, a verification period, expected impact and risk. You create actions as tasks in Peak Work.

1. 01Open Peak Constraints → Bottleneck and click “Confirm bottleneck”.
2. 02Switch to “Resolve” and fill in the resolution plan.
3. 03Click “Save plan”.
4. 04Create each action with “Create as task”.

How do I report an observation from the team?

On the “Diagnosis” page under “Signals from the team”. Choose the type, write the observation in one sentence and optionally assign it to a step, then click “Report”. Peak weighs it as one piece of evidence among several; on its own, it doesn't make a step the bottleneck.

How is the confidence of a bottleneck hypothesis calculated?

From seven types of evidence with fixed weights: capacity versus load, sensitivity of throughput, causal position in the flow, financial impact, consistency of measurements, trend across several snapshots and human signals. If data is missing, confidence drops and Peak names what's missing. Below 70% it remains a hypothesis.

When is a bottleneck considered resolved?

Not when all tasks are done, but when the target metric reaches the goal, system throughput rises and that holds over the review period. Verification shows each check individually. Peak then tells you where the bottleneck is likely to sit next.

What does “Challenge diagnosis” do?

Challenge mode looks for reasons why Peak could be wrong, such as missing capacity values, only one measurement, estimated values or an old measurement, and deducts points from the confidence for each reason. At the end there's a conclusion. For an adopted bottleneck, Peak records when it was challenged and how the confidence changed.

Can I create experiments for a bottleneck?

Yes, via Peak Experiments. Under “Resolve” → “Diagnostic experiments”, “Create experiment in Peak Experiments” leads to a new experiment related to the bottleneck. Hypothesis, criteria and result live in Peak Experiments; Peak Constraints shows the experiments and their verdict on the bottleneck.

Why can't I adopt a second bottleneck?

Peak Constraints works with one current bottleneck at a time, because throughput is only ever limited at one point. Resolve or discard the current bottleneck, and then you can adopt the next one. The list of previous ones is under “History”.