The Haystack Syndrome

1. Information is the answer to the question asked, not just data

"Information is not input to the decision process, it is the output of the decision process."

Decision process is crucial. Information systems must embed decision-making procedures to transform data into useful answers. Simply collecting more data is not enough - we need the right process to extract meaningful insights.

Hierarchical structure of information. Lower-level data serves as inputs to generate higher-level information through decision processes. An effective information system should be able to climb this ladder, from basic data to tactical answers to financial bottom line results.

Required data vs. raw data. Focus on capturing the specific data elements needed to derive desired information, rather than amassing all possible data. Accuracy is more important for the critical data points that impact final results than for tangential data.

2. Throughput, Inventory, and Operating Expense are key measurements

"TELL ME HOW YOU MEASURE ME AND I WILL TELL YOU HOW I WILL BEHAVE."

Throughput is paramount. Defined as the rate at which the system generates money through sales, throughput should be the top priority. Unlike inventory and operating expense which are limited by zero, throughput has unlimited potential for improvement.

Inventory is a double-edged sword. While necessary to protect throughput, excess inventory ties up cash and hinders flexibility. The goal is to minimize inventory while still buffering against variability.

Operating expense comes third. While important to control, cutting operating expenses should not come at the expense of throughput or necessary inventory. Focus on eliminating waste rather than arbitrary cost-cutting.

3. Identify and exploit system constraints to maximize performance

"The strength of the chain is determined by the strength of its WEAKEST link."

Constraints limit overall system performance. Every system has at least one constraint - the bottleneck that determines its maximum output. Identifying these constraints is crucial for improvement efforts.

Physical vs. policy constraints. Physical constraints like machine capacity are often easier to identify than policy constraints stemming from outdated rules or metrics. Policy constraints should be eliminated rather than exploited.

Focus improvement efforts on constraints. Since constraints determine overall system performance, efforts to improve non-constraints will not increase throughput. Concentrate on maximizing the efficiency of constraint resources.

4. Subordinate everything else to the system's constraints

"If we don't utilize the constraints properly, the consumption rate will be less than desired."

Align the entire system to the constraint. Ensure non-constraint resources support maximum utilization of constraints, even if it means running them at less than full capacity. This may feel counterintuitive but optimizes overall system performance.

Challenge traditional efficiency metrics. Local efficiency measures often drive behaviors that hurt overall system performance. Develop new performance measures aligned with global optimization.

Buffer constraints against variation. Use time buffers and protective capacity on non-constraints to ensure constraints are never starved of work. This protects throughput at the cost of some additional inventory.

5. Elevate the system's constraints to increase capacity

"Elevate means 'Lift the restriction.'"

Constraint elevation options:

• Add capacity (e.g. overtime, additional shifts)
• Offload work to non-constraints
• Reduce setup/changeover times
• Improve quality/yield
• Outsource constraint operations

Evaluate ROI of constraint investments. Since constraints determine overall system output, investments to increase constraint capacity often have excellent returns compared to non-constraint improvements.

Prepare for constraint shifts. Successfully elevating a constraint will cause the system bottleneck to move. Be ready to identify and exploit the new constraint.

6. Avoid inertia by continually reassessing constraints

"If, in the Previous Steps, a Constraint Has Been Broken, Go Back to Step One, but Do Not Allow Inertia to Cause a System's Constraint."

Constraints evolve over time. As the system changes due to improvement efforts or external factors, the location and nature of constraints will shift. Continuously reassess to avoid optimizing for outdated constraints.

Beware of policy constraints. Often, outdated policies or metrics become the primary constraint on system performance. These can be more difficult to identify than physical constraints.

Challenge assumptions regularly. What was true in the past may no longer hold. Create a culture of questioning the status quo to prevent inertia from limiting improvement.

7. Use buffers to protect against variability and disruptions

"Time buffer is our protection against unknown disturbances."

Types of buffers:

• Shipping buffer: Protects due date performance
• Constraint buffer: Ensures constraint is never starved
• Assembly buffer: Coordinates arrival of components

Buffer management provides valuable data. Tracking buffer consumption highlights problem areas and provides early warning of potential disruptions. Use this data to drive continuous improvement efforts.

Dynamic buffer adjustment. Regularly review and adjust buffer sizes based on actual performance. Reduce buffers where possible to minimize inventory while maintaining protection against variability.

8. Implement dynamic scheduling to optimize resource utilization

"We have to make sure that the amount of blocks that are required to be done at the same point in time will never exceed the number of units available on the resource constraint."

Load leveling is crucial. Smooth out peaks and valleys in resource loading to maximize utilization of constraints and minimize idle time on non-constraints. This often requires shifting work earlier than the due date strictly demands.

Consider capacity when subordinating. Take into account the capacity of non-constraint resources when scheduling, not just lead times. This prevents unrealistic schedules that ignore resource limitations.

Use "drum-buffer-rope" concept. Let the constraint set the pace (drum), protect it with buffers, and use a "rope" to pull material release in line with constraint capacity. This synchronizes the entire system to the constraint.

9. Focus on global optimization rather than local efficiencies

"Local optima do not add up to the optimum of the total."

Suboptimization pitfalls:

• Maximizing machine utilization
• Minimizing labor costs
• Optimizing batch sizes locally
• Focusing on product costs

Throughput accounting vs. cost accounting. Traditional cost accounting often drives behaviors that hurt overall system performance. Throughput accounting aligns decisions with global optimization.

Challenge "common sense" efficiency rules. Many ingrained practices optimize locally at the expense of system performance. Be willing to challenge conventional wisdom when it conflicts with global optimization.

10. Measure performance based on impact on the overall goal

"Local performance measurements should judge the quality of the execution of a plan, and this judgment must be totally separate from judging the plan itself."

Align metrics with the goal. Ensure all performance measures drive behaviors that improve overall system performance, not just local optimization. This often requires developing new, unconventional metrics.

Separate plan quality from execution quality. Don't punish good execution of a flawed plan or reward poor execution of a good plan. This requires clearly distinguishing between planning and execution responsibilities.

Use leading indicators. Develop metrics that provide early warning of potential problems rather than just measuring end results. Buffer management data can be valuable for this purpose.

Last updated: April 22, 2025