Runner-Repeater-Stranger

Runner-repeater-stranger is one of the more useful classification tools in lean and one of the most underused in small shops. Most shops treat every part the same: same MRP process, same scheduling logic, same expediting attention. The RRS framework says that is a mistake. Parts with very different demand profiles deserve very different scheduling strategies. Sorting them into three buckets is the first step toward giving each part the right pull mechanism.

"A part ordered weekly does not belong in the same system as a part ordered once a quarter. Sort first, then schedule."

How runner-repeater-stranger works

The classification has three buckets, sorted by demand frequency and stability. Runners are the parts that get ordered constantly with predictable volume. Think of them as the parts you make every week, in roughly the same quantity. They are the easy candidates for replenishment pull with kanban signals and supermarkets, because the demand is steady enough that stocked inventory makes sense.

Repeaters are parts that get ordered regularly but with lumpier volume. Maybe they get ordered monthly, or every few weeks, or in clusters around a customer's release schedule. Repeaters do not justify constant stocking but they do justify a predictable production cadence. The scheduling approach is usually a planned cycle through the repeater list at a known interval, often a heijunka box that runs through repeaters in rotation.

Strangers are sporadic. They get ordered rarely or with no pattern. One-off custom parts, prototypes, parts for customers who only call once or twice a year. Strangers should not be in any kanban system. The right approach is usually sequenced pull or build-to-order, with capacity reserved on the schedule for whatever stranger orders come in.

The classification is empirical. Each shop sets its own cutoffs based on its volume profile. A high-volume contract manufacturer might define a runner as anything ordered weekly with under 15 percent variation. A small job shop might define a runner as anything ordered monthly with under 30 percent variation. The exact numbers matter less than having a consistent rule that the team applies to the part list and revisits quarterly.

The output of the classification is a different production strategy for each class. Runners get kanban, supermarkets, and steady-pace production. Repeaters get a scheduled rotation, with their own kanban racks but cycled less often. Strangers get one-at-a-time scheduling, with reserved capacity slots. The whole shop runs on three different pull mechanisms in parallel, each sized for the demand profile it serves.

Where runner-repeater-stranger fits on the shop floor

Picture a small contract machine shop running about 200 part numbers across maybe 40 active customers. Without RRS classification, every part goes through the same MRP process, gets scheduled the same way, and competes for the same expediting attention. The shop runs about a four-week lead time across the board, has roughly $300,000 in WIP, and the owner cannot tell which parts are profitable.

The owner sorts the 200 parts. About 25 of them turn out to be runners: ordered weekly with predictable volume. Another 80 are repeaters: ordered monthly or every few weeks with steadier patterns. The remaining 95 are strangers: one-off custom work or rarely-ordered specialty parts. The shop installs kanban supermarkets for the 25 runners, with two-bin systems at the cell that consumes them. The 80 repeaters get a weekly cycle schedule through a heijunka rotation. The 95 strangers get sequenced into reserved capacity slots as orders come in.

Within a quarter, the runners and repeaters run on autopilot. The expediter only has to focus on the strangers, which are the actual one-off work that needs attention. Lead time on runners drops to under a week. Lead time on repeaters drops to two weeks. Strangers still take four weeks but the floor is no longer chasing them through chaos. The shop is calmer, the customers are happier, and the owner can see which classes of parts are profitable.

Common mistakes with runner-repeater-stranger

• Treating boundaries as fixed. Demand shifts. A part that was a runner last year may be a repeater this year. Revisit the classifications quarterly.
• Classifying once and never revisiting. A stranger that becomes a regular order should move into the repeater bucket and get the matching scheduling treatment.
• Treating strangers like runners. Strangers do not need kanban. Trying to stock them creates excess inventory of parts that may never sell again.
• Sorting by revenue instead of demand frequency. RRS is about how often a part is ordered, not how much money it brings in. Mixing those criteria muddies the classification.
• Skipping the strangers. The bucket that needs the most active management is strangers. Leaving them to MRP guarantees they get treated badly.

Runner-repeater-stranger and related Lean tools

Runner-repeater-stranger classification drives different mechanisms for different demand profiles. Runners typically get replenishment pull through kanban supermarkets. Repeaters get cycled through heijunka on a known cadence, with the cycle interval set by every part every interval. Strangers usually run on sequenced pull or build-to-order. The classification is most useful in shops running mixed-model production where the variant mix needs to be sequenced thoughtfully.