Who is the Value Stream Map for? —— A Five-Step Method for Product Family Classification and Boundary Determination

1. Why Do Many Value Stream Maps Go to Waste?

A company has over forty products. The lean promotion office selected key personnel to conduct on-site observations, measure working hours, and count inventory, spending two weeks to create a "complete plant value stream map": from raw materials entering the factory to finished products being shipped, dozens of processes are densely arranged on a single map. When the map was hung in the meeting room, everyone was satisfied. However, after half a year, none of the improvement points on the map had been implemented—not due to poor execution, but because the map itself lacked guiding significance. With forty products mixed together, the cycle times, inventory levels, and process times were all averaged, making each number on the map "about right" but none of them accurate. The production line manager said, "These numbers don't represent my line," and the sales team said, "This cycle time doesn't apply to my product category."

A value stream map (VSM) is not a plant layout map but a diagnostic tool for a single product family. The first principle of its methodology is: one value stream map should only cover one product family. Different product families have different process flows, batch sizes, changeover methods, and bottleneck equipment. Mixing them on a single map leads to data contamination, and improvement suggestions will not fit well on any specific line. Therefore, the first step in creating a VSM is not to draw the map but to answer two questions: Which product family does this map represent? Where does the boundary start and end?

2. Product Family Classification: Two Steps to Group Dozens of Products into a Few Families

Step One: Conduct a P-Q Analysis (Product-Quantity Analysis). Rank all products by annual production or shipment volume from largest to smallest, and perform a Pareto analysis. Typically, 20% of the products contribute 70% to 80% of the production volume. These "high-volume, narrow-range" products are the primary focus. Products with low volume, which form the long tail, should be set aside to avoid interfering with the main family's classification.

Step Two: Create a Process Similarity Matrix. Place the top-ranked products into a matrix: rows represent products, and columns represent processes or key equipment. Verify the actual process routes for each product. Products that follow the same route, share the same equipment, and require only minor adjustments for changeover are grouped into the same family. Products with significantly different process routes or requiring specialized equipment, even if they have substantial production volume, should be classified into separate families.

The principles for classification are threefold: consistent process routes within the family, shared bottleneck equipment, and significant differences between families. After completing this, you will have one to three main families, plus an "other family" for miscellaneous products. For companies with a wide variety of products in small batches, it is crucial to resist the impulse to fit everything into one map—prefer three clear maps over one that represents nothing accurately.

3. Boundary Determination: From Which Door to Which Door

Once the product family is determined, the next step is to define the boundaries. Value stream boundaries are generally defined from "door to door," and this involves answering three questions.

First Question: Where Does It Start and End? The standard approach is to start from raw materials entering the factory (or being shipped by suppliers) and end with finished products being shipped (or received by customers), incorporating the entire product flow within the factory into the map. A "partial snapshot" that only covers your own workshop and ignores upstream and downstream processes will break the flow, and improvements are likely to become localized optimizations within a department.

Second Question: To What Level Should Material Flow Be Mapped? Material flow should be mapped to the process level, with each process having its own data box, labeled with key data such as cycle time, changeover time, work-in-progress quantity, and first-pass yield. Information flow should be mapped to the order processing and scheduling level, accurately depicting how customer orders are transformed into production instructions.

Third Question: Who Stays Outside the Map? Support processes such as equipment maintenance, human resources, and financial management should not be included in the value stream map. They should only be noted if they significantly impact the flow. Once the boundaries are set, the data collection scope for the entire map is locked, and the criteria must be consistent. When redrawing the current state map in the future, the same boundary rules should be applied to ensure comparability.

4. Three Common Mistakes

Mistake One: One Map for the Entire Plant. As mentioned earlier, average data can mislead decision-making. The judgment criterion is straightforward: if the cycle time of any process on the map cannot be clearly attributed to a specific product category, the map should be split.

Mistake Two: Cutting Boundaries by Department. Some people only map the workshop they are responsible for, ignoring suppliers, customers, and downstream processes. The essence of a value stream is end-to-end flow, and cutting boundaries at departmental walls will always result in localized improvements.

Mistake Three: Arbitrary Boundary Shifts. One time, the map starts from the warehouse, and another time, it starts from the supplier. This inconsistency makes data from different maps incomparable, and improvement effects cannot be verified. The correct approach is to write the boundary rules into the company's mapping standards, fix them, and ensure everyone follows them.

5. Case Study: A Motor Factory with Thirty Products

A motor factory produces over thirty products and ships 80,000 units per month. During the first attempt at creating a VSM, the team chose a line with the "most complete processes" to represent the entire plant, but the improvement suggestions failed on other product lines. Later, they reclassified the products using the two-step method: the P-Q analysis showed that five products accounted for 62% of the production volume. The process matrix further confirmed that these five products followed the same assembly line and shared the same bottleneck equipment, thus forming the main family. The boundary was set from "raw materials entering the warehouse to finished products being shipped," and the information flow was mapped to sales order entry. After completing the main family map, the delivery cycle was reduced from 21 days to 9 days. The same method was then applied to the second family, and within two months, the average delivery cycle for the entire plant was shortened by about 40%. Each family map was drawn separately, ensuring data purity and providing practical guidance for improvements within each family.

6. Conclusion

A value stream map is not better for being more comprehensive but for being more accurate. First, classify the product families, then define the boundaries, ensuring that each map answers one specific question: where is the waste in this product family from order to delivery? An accurate map provides a clear target for improvements.

Before drawing a value stream map, classify the product families and define the boundaries. One map should serve one product family, providing a precise target for improvements.

Knowledge code: 7.1.1

Version: v20260814

Author: Quality Think Tank The Quality Think Tank is dedicated to providing systematic professional knowledge, methodologies, and practical tools to quality management practitioners, helping companies continuously enhance their quality capabilities.