# The Haystack Syndrome **Eliyahu M. Goldratt** | [[Numbers]]  --- > "Information is not input to the decision process—it is the output." Goldratt obliterates the distinction you thought existed. Information isn't raw data. It's **the answer to the question asked**. Companies are drowning in data but starved of information because they've confused inputs with outputs. > "Measurements are a direct result of the chosen goal. There is no way that we can select a set of measurements before the goal is defined." Most organisations measure everything and control nothing. The goal is to make more money, now and in the future. Three questions: How much money is generated? How much is captured? How much do we spend? That's it. Everything else is noise. > "Changing a constraint changes everything." In the cost world, tweaking one variable barely moves the needle. In the throughput world, shifting the constraint rewrites the entire system. You cannot optimise locally and expect global performance. Constraints are the 0.1% of variables that determine 99.9% of results. This book strips cost accounting down to its bones and shows you why it's killing your business. It's not gentle. Goldratt has no patience for allocation games or "product cost" fantasies. If you want to manage for throughput rather than efficiency theatre, start here. --- ## Core Frameworks ### [[Data vs Information]] **Data** = any string of characters describing reality. **Information** = the answer to the question asked. This isn't semantic hair-splitting—it changes everything. Information is the *output* of a decision process, not the input. "What is data and what is information is in the eyes of the beholder." The same string of characters can be data for one person, information for another. Most companies collect oceans of data but never convert it into actionable information. Key reframe: an effective information system doesn't start with data—it starts with the question, then builds the decision procedure, *then* sucks the required data from a data system. ### [[Throughput Accounting]] Replace cost accounting with three global measurements: 1. **Throughput (T)** – Money generated through sales (selling price minus materials) 2. **Inventory (I)** – Money the system invests in things it intends to sell 3. **Operating Expense (OE)** – Money spent converting inventory into throughput > "When is the only point in time that we add value to the company? Only when we sell, not a minute before!! The whole concept of adding value to a product is a distorted local optimum." Value is realised at the moment of sale, not during production. "Product cost," "product margin," "product profit"—delete from your vocabulary. Net profit exists only at the company level, never at the product level. Evaluate decisions by their impact on T, I, and OE—not by phantom allocations. > "Tell me how you measure me, and I will tell you how I will behave. If you measure me in an illogical way, do not complain about illogical behaviour." ### [[Theory of Constraints]] The core discipline for managing constraints—five focusing steps: 1. **Identify** the constraint (the weakest link) 2. **Exploit** it (squeeze maximum value without major investment) 3. **Subordinate** everything else to it (align non-constraints to serve the constraint) 4. **Elevate** (break) the constraint if needed 5. If it shifts, **repeat**—do not allow inertia to cause a system constraint **Most constraints are policy-driven, not physical.** The real bottleneck isn't the machine—it's the rule you refuse to question. > "In the throughput world, constraints are the essential classification, replacing the role that products played in the cost world." Manage by constraints, not by products. ### [[Local vs Global Optimisation]] > "Local optima do not add up to the optimum of the total." In independent systems, the Pareto principle holds: 20% of variables drive 80% of results. In dependent systems with variability, it's the 0.1-99.9 rule: a tiny fraction determines almost everything. **Every local efficiency that doesn't improve the constraint is waste.** If the market is the constraint, exploit = 100% on-time delivery (not 99%, one hundred). The work ethic trap: "If a worker does not have anything to do, find him something to do!" This is the opposite of subordination. Non-constraints should have *protective capacity*, not be kept busy. --- ## Key Insights **An information system must have three modules:** Schedule (realistic plans based on constraints and buffers), Control (track deviations and quantify Murphy), and What If (simulate future scenarios and decisions). > "In order to identify even the CURRENT constraints, we need to simulate FUTURE actions." Scheduling isn't optional—it's fundamental to constraint identification. Information systems ≠ data systems. Data systems answer straightforward questions; information systems answer questions requiring decision procedures. The decision procedure must be embedded in the information system, not bolted on afterward. **Start from market constraints (customer orders), not internal capacity.** Subordinate non-constraints: ignore their own limitations, focus on serving the constraint. Use **time buffers**, not inventory buffers, to protect constraints from Murphy. Buffer size is a judgement call—must be set by those responsible for overall performance. **Buffer management** = monitor buffer consumption to detect and control disturbances. Changing a constraint changes everything. Go back to step one. Look at the system as if you've never seen it before. It *is* a new system. > "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." Two types of deviations: **Not doing what was supposed to be done** (impacts throughput) and **Doing what was not supposed to be done** (inflates inventory). Measure deviation impact in **dollar-days** (financial impact × time). > "Try to measure by three or more non-financial measurements, and you have basically lost all control." Non-financial measurements = anarchy. You cannot compare apples, oranges, and bananas. Every measurement must tie to the goal: making more money. **Cost accounting is the enemy.** "Product cost" is a mathematical phantom—you never pay money *to a product*. Cost accounting promotes local efficiency at the expense of global effectiveness. Allocation games obscure the real question: What is the impact on T, I, and OE? Dr. Ohno (inventor of Kanban): "Cost accounting was the one thing I had to fight all my life. It was not enough to chase out the cost accountants from the plants; the problem was to chase cost accounting from my people's minds." When evaluating whether to drop a product: What is the impact on **total throughput**? What is the impact on **total operating expense**? (How many people will be laid off? Specify names.) If reduction in throughput < reduction in OE, drop it. Otherwise, you're jeopardising the goal. --- ## Connects To - [[Playing to Win]] - Both emphasise that the goal must be defined before measurements make sense; strategy precedes metrics - [[The Fifth Discipline]] - Local optimisation destroys global performance; systems thinking requires seeing interdependencies - [[7 Powers]] - Constraints determine competitive performance; focus on the weakest link, not across-the-board improvements - [[Better, Simpler Strategy]] - Simplify to the vital few measurements (T, I, OE) rather than tracking everything - [[Dead Companies Walking]] - Companies die because they optimise locally (cost per unit) while destroying global performance (throughput) --- ## Final Thought You are swimming in data but have no information. Information isn't what you collect—it's the answer to the question you're asking. And most companies never define the question properly because they've never defined the goal. The goal is to make more money, now and in the future. That requires three measurements—throughput, inventory, operating expense—and nothing else. Cost accounting, with its phantom allocations and "product cost" fantasies, is the enemy. It encourages local optimisation (keep everyone busy) at the expense of global performance (exploit the constraint). **Changing a constraint changes everything.** In the cost world, tweaking one variable barely registers. In the throughput world, the constraint is the 0.1% that determines 99.9% of results. Move the constraint and the entire system rewrites itself. This is why companies that optimise locally—chasing efficiency at every workstation—destroy value. They're polishing non-constraints while the constraint sits idle. Use time buffers, not inventory buffers. Schedule from the constraint backward. Subordinate non-constraints to serve the constraint, and accept that they *must* have protective capacity. Keeping everyone busy 100% of the time is a recipe for chaos. This isn't a book about information systems. It's about rethinking how you see your business. Stop managing by products. Start managing by constraints. Stop collecting data. Start answering questions. And for the love of throughput, stop using cost accounting.