It is 9:30 AM in a fabrication shop in Ahmedabad. The production manager printed the schedule at 8:00 AM. By 9:30, the schedule is already wrong. A raw material shipment that was supposed to arrive yesterday is stuck in transit. The owner's friend called in a rush order that "absolutely must ship by Friday." The welding machine on Bay 3 tripped a breaker and is down for at least two hours. And the painter just told him that the powder coating oven needs a heating element replacement — it will be ready by afternoon, maybe.
This is not a bad day. This is every day in a job shop that makes custom products. When every order is different — different materials, different operations, different tolerances, different delivery dates — scheduling is not a planning exercise. It is a continuous renegotiation with reality.
The typical response is one of two extremes: either the factory gives up on scheduling entirely and runs on expediting (the loudest customer gets served first), or they invest in MRP software that generates a beautiful schedule which nobody on the floor follows. Both approaches fail. There is a middle path, and it works.
Why traditional MRP scheduling fails for job shops
MRP (Material Requirements Planning) and its successor MRP II were designed for repetitive manufacturing — factories that make the same products in predictable volumes. The scheduling logic assumes:
- Known BOMs that do not change. In a job shop, the BOM is often finalised after the order is placed, sometimes after production has started.
- Stable routings. MRP assumes Operation 10 always goes to Machine A, then Operation 20 goes to Machine B. In a job shop, the routing depends on which machine is available, which fixture is set up, and which operator is on shift.
- Reliable lead times. MRP uses fixed lead times per operation. But in custom manufacturing, the lead time for "welding" could be 2 hours or 20 hours depending on the complexity of the assembly.
- Infinite capacity (in basic MRP). Standard MRP schedules backward from the due date without checking whether the machines can actually handle the load. It might schedule 40 hours of CNC work on a day when you have 16 hours of CNC capacity.
The result: MRP generates a schedule that looks precise on screen and is useless on the floor. The production manager prints it, glances at it, and goes back to managing by WhatsApp and walkaround.
The real scheduling challenge for ETO and job-shop manufacturers
Engineer-to-order (ETO) and job-shop manufacturers face scheduling challenges that are fundamentally different from repetitive manufacturers:
- High mix, low volume. You might run 50-100 different jobs in a month, each with a unique routing.
- Variable processing times. The same operation on the same machine can take 2 hours or 8 hours depending on the job.
- Dependent operations with variable gaps. Welding cannot start until cutting is done, but how long cutting takes depends on the material thickness, the complexity of the profile, and whether the plasma cutter is having a good day.
- Shared resources. The overhead crane, the inspection area, the paint booth, and specific fixtures are shared across jobs and create invisible bottlenecks.
- Rush orders. In Indian manufacturing, rush orders are not exceptions. They are a business model. Handling them without destroying existing commitments is the core scheduling problem.
The "dispatch date backward" method
Instead of scheduling forward from today (which is what most people do intuitively), schedule backward from the dispatch date. This is not a new idea — it is the logic behind MRP. But the implementation for a job shop is different.
How it works
Start with the dispatch date. The customer needs the goods on June 20. Allow 2 days for packing and dispatch logistics. The goods must be ready by June 18.
Work backward through operations. The last operation is inspection/packing (1 day). Before that is painting (1 day including curing). Before that is welding and assembly (3 days). Before that is machining (2 days). Before that is cutting (1 day).
Add buffer between operations. In a job shop, jobs do not flow instantly from one station to the next. There is queue time — waiting for the next machine, waiting for the crane, waiting for the operator. Add a realistic buffer between operations. For most Indian job shops, 0.5-1 day buffer between operations is realistic.
Calculate the latest start date. Working backward: dispatch June 20, ready June 18, inspection June 17, painting June 16, buffer June 15, welding June 12-14, buffer June 11, machining June 9-10, buffer June 8, cutting June 7.
Latest start date for cutting: June 7. If it is currently June 1, you have 6 days of slack. If it is June 10, you are already late on cutting and need to expedite.
The backward schedule as a decision tool
The power of backward scheduling is not in the precision of the dates — those will change. The power is in answering two questions instantly:
Which jobs are at risk right now? Any job whose latest start date for the current operation has passed is at risk. These jobs need attention today.
How much slack does each job have? Jobs with 5+ days of slack can absorb disruptions. Jobs with 0-2 days of slack cannot. When a rush order arrives, you pull capacity from high-slack jobs, not randomly.
This converts scheduling from a rigid plan into a priority system. You are not trying to predict the future. You are trying to make good decisions right now about which jobs to work on next.
Finite versus infinite capacity scheduling
Infinite capacity (the default)
Most spreadsheet-based scheduling is infinite capacity by default. You schedule backward from the due date and assume every machine has unlimited availability. The schedule says "4 hours of CNC turning on June 9" without checking whether the CNC lathe already has 12 hours of work scheduled for June 9.
The result is overloaded days followed by idle days, and a schedule that is impossible to execute as written.
Finite capacity (the reality check)
Finite capacity scheduling respects the actual hours available on each machine. If your CNC lathe is available for 16 hours per day (two shifts), the schedule cannot assign more than 16 hours of work to it on any given day. When the load exceeds capacity, the excess work gets pushed to the next available day.
Why finite capacity matters for job shops:
Consider a 20-machine fabrication shop with these key resources:
| Resource | Available hours/day | Typical load |
|---|---|---|
| CNC plasma cutting table | 16 hours | 12-18 hours (often overloaded) |
| Press brake (bending) | 16 hours | 8-14 hours (variable) |
| MIG welding bays (4 bays) | 64 hours total | 40-70 hours (peaks on assembly jobs) |
| Powder coating booth | 10 hours (limited by curing cycles) | 8-12 hours (frequent bottleneck) |
| Inspection area | 8 hours | 3-6 hours |
With infinite capacity scheduling, you might schedule 22 hours of plasma cutting on a Tuesday. The schedule looks fine on paper. On Tuesday, the plasma operator works 16 hours and the remaining 6 hours push to Wednesday, which was already full. The cascade begins.
With finite capacity scheduling, you see the overload before it happens. You can move work earlier, shift it to an alternate machine, or renegotiate the delivery date with the customer — before you miss it, not after.
Practical finite capacity for Indian SMEs
You do not need expensive APS (Advanced Planning and Scheduling) software to do finite capacity scheduling. You need:
- A list of your key resources (machines/workstations) with daily available hours
- Every active job with its remaining operations and estimated hours per operation
- A visual board (physical or digital) that shows the load on each resource per day
When you load a new job, you check the board. If the CNC lathe is full on Thursday, the job goes to Friday. Simple. The board enforces finite capacity through visual constraint — you cannot stack more hours into a day than the column can hold.
Handling rush orders without wrecking the plan
Rush orders are the reality of Indian manufacturing, especially for job shops and fabrication units that serve OEMs, project companies, and dealers. You cannot refuse them — these customers pay premiums and send repeat business. But you cannot accept them blindly either, or every other customer's delivery suffers.
The rush order protocol
Step 1: Assess the impact. Before saying yes, check your visual schedule. Which machines does this rush order need? What is the current load on those machines? Which existing jobs will be delayed if the rush order jumps the queue?
Step 2: Identify the victims. Determine which specific jobs will be pushed back and by how many days. If Job 47 for Customer B slips from June 15 to June 18, that is a 3-day delay. Is Customer B okay with that? Is Job 47 on the critical path for their project?
Step 3: Negotiate before committing. Call Customer B before accepting the rush order. Explain the situation. Most customers accept a 2-3 day slip if you tell them proactively. What they do not accept is being surprised at the last minute. Proactive communication converts a delivery failure into a schedule adjustment.
Step 4: Price the rush correctly. A rush order costs more than a normal order. It disrupts schedules, causes overtime, and may require expedited material procurement. Factor in:
- Overtime premium: 1.5-2x labour cost for the extra hours
- Schedule disruption: lost efficiency from breaking setups and re-sequencing
- Material expediting: courier charges, premium for small-quantity purchases
- Opportunity cost: the margin lost on delayed jobs
A reasonable rush premium for Indian job shops is 15-30% above the standard price. Customers who genuinely need rush delivery will pay it. Customers who call everything "urgent" will reconsider when they see the premium.
Step 5: Slot it in and re-sequence. Insert the rush order into the schedule, push affected jobs, and communicate the new dates to all affected customers. This re-sequencing takes 15-30 minutes. The alternative — not re-sequencing and hoping for the best — causes 3-5 days of chaos as jobs pile up unpredictably.
Visual scheduling boards — physical and digital
The physical board
A whiteboard with a matrix: rows are machines/workstations, columns are days. Each job gets a magnetic strip or sticky note placed in the appropriate row and column, sized proportionally to the estimated hours.
Example layout for a 10-machine fabrication shop:
| Mon 9 Jun | Tue 10 Jun | Wed 11 Jun | Thu 12 Jun | Fri 13 Jun | |
|---|---|---|---|---|---|
| CNC Plasma | Job 42 (6h) / Job 45 (8h) | Job 47 (12h) | Job 47 (4h) / Job 50 (8h) | Job 52 (14h) | Job 52 (2h) / Open |
| Press Brake | Job 42 (4h) | Job 45 (6h) / Job 42 (4h) | Job 47 (8h) | Job 50 (6h) | Job 47 (4h) |
| Welding Bay 1 | Job 38 (16h) | Job 38 (8h) / Job 42 (8h) | Job 42 (12h) | Job 45 (10h) | Job 45 (6h) / Job 47 (8h) |
| Welding Bay 2 | Job 40 (14h) | Job 40 (2h) / Job 44 (12h) | Job 44 (6h) | Open | Job 50 (8h) |
| Powder Coat | Job 36 (5h) / Job 38 (5h) | Job 40 (6h) | Job 42 (4h) | Job 44 (6h) / Job 45 (4h) | Job 38 (6h) |
Advantages of the physical board:
- Everyone on the floor can see it. No login required.
- Moving a job is as simple as moving a magnet.
- The production meeting happens in front of the board — decisions are immediate and visible.
- Overloaded days are visually obvious (too many strips in one column).
Disadvantages:
- Cannot easily share with sales team or management offsite.
- No automatic recalculation when you move a job.
- Difficult to maintain for more than 2 weeks ahead.
- Colour-coding and sizing get messy with 50+ active jobs.
The digital board
A software-based visual scheduler that replicates the physical board on a screen but adds calculation, sharing, and automation.
Advantages of the digital board:
- Accessible from any device — the sales team can check capacity before promising a delivery date.
- Automatic conflict detection — warns you when a resource is overloaded.
- Drag-and-drop re-sequencing with automatic cascade calculation.
- Historical data — shows how your actual schedule compared to the planned schedule.
- Integration with job costing — the schedule drives labour cost allocation.
Disadvantages:
- Requires discipline to keep updated (a digital board with stale data is worse than no board).
- Can feel disconnected from the shop floor if operators do not see it.
- Over-reliance on the software can reduce the production manager's intuitive feel for the floor.
Recommendation for Indian SMEs: Start with a physical board for the first 3-6 months. Get the scheduling discipline in place. Understand your capacity, your bottlenecks, and your typical flow. Then move to a digital board that preserves the same visual logic but adds computational power. Trying to implement digital scheduling without first understanding your scheduling patterns is like using GPS in a city you have never visited — you follow the directions but you do not understand the terrain.
Buffer management — the unsung hero of job-shop scheduling
In a job shop, buffers are not waste. They are insurance. The question is not whether to have buffers, but where to put them and how big to make them.
Where buffers belong
Before the bottleneck. The bottleneck machine must never be starved of work. If your powder coating booth is the constraint, you want a buffer of ready-to-coat jobs waiting in front of it at all times. An empty buffer means the bottleneck sits idle — and every hour of idle time on the bottleneck is an hour of lost factory output.
Before the customer. A shipping buffer at the end of the production process protects your delivery date from last-minute problems. If inspection rejects a part on the day before dispatch, the buffer gives you time to rework or remake it without missing the delivery.
Not everywhere else. Buffers between non-bottleneck operations just increase WIP and lead times without protecting anything. If cutting takes 2 hours and bending takes 1 hour and neither is a bottleneck, you do not need a 2-day buffer between them.
How big should buffers be
The buffer size depends on the variability of the operation it protects. High variability needs a bigger buffer.
A practical rule of thumb for Indian job shops:
| Buffer location | Recommended size | Rationale |
|---|---|---|
| Before bottleneck | 1-2 days of bottleneck capacity | Protects against upstream disruptions |
| Shipping buffer | 2-3 days before dispatch date | Protects against last-minute quality issues |
| Between stable operations | 0.5 day | Minimal buffer for material handling |
| Between variable operations | 1 day | Absorbs cycle time variation |
Monitoring buffer status
Buffer management only works if you monitor the buffers. A green/yellow/red system is practical:
- Green (> 66% of buffer remaining): On track. No action needed.
- Yellow (33-66% of buffer remaining): At risk. Review and consider expediting.
- Red (< 33% of buffer remaining): In danger. Immediate action — overtime, resource reallocation, or customer communication.
If you check buffer status once per day (which takes 10 minutes with a visual board), you catch problems 2-3 days before they become delivery failures. This is the difference between proactive production management and reactive firefighting.
A practical scheduling framework for a 20-machine shop
Here is a complete framework for a fabrication and machining shop with approximately 20 machines, 8-12 operators per shift, and 40-80 active jobs at any time. This is typical of a mid-size Indian job shop in cities like Pune, Rajkot, Coimbatore, or Faridabad.
Daily rhythm
7:45 AM — Pre-shift review (15 minutes) Production manager + shift supervisor in front of the scheduling board. Review:
- Which jobs are scheduled for today on each machine?
- Any material shortages flagged by the store?
- Any machine maintenance issues from last night?
- Any new rush orders received since yesterday?
Adjustments are made on the spot. Jobs are re-sequenced if needed. The shift starts with a clear, agreed plan.
12:00 PM — Midday check (10 minutes) Quick walkthrough or dashboard review. Are jobs progressing as planned? Any downtime events? Any quality issues? This is the early-warning check — if a morning delay has cascading implications for afternoon jobs, you catch it now, not at 5 PM.
5:00 PM — End-of-shift update (15 minutes) Supervisor updates the board with actual progress. Jobs completed are removed. Jobs in progress are noted with remaining hours. Any issues (material hold, machine down, quality rejection) are flagged for the next shift.
Weekly: Friday capacity review (30 minutes) Production manager reviews the next 2-week schedule. Checks:
- Is any resource overloaded in the coming week?
- Are there material deliveries that might be late?
- Which jobs have the least buffer (most at risk)?
- Any confirmed rush orders to slot in?
This weekly cadence prevents the "surprise Monday" — arriving at the start of the week to discover that the coming week is impossible.
Job prioritisation rules
When multiple jobs compete for the same machine, use this priority sequence:
- Buffer status red — jobs whose delivery buffer is critically low. These go first regardless of other factors.
- Rush orders with confirmed premium — jobs the customer is paying extra for expedited delivery.
- Highest revenue per machine hour — when two jobs have similar buffer status, prioritise the one that generates more revenue per hour on the bottleneck machine.
- First-come, first-served — when all else is equal, the job that was released to the floor first goes first.
This priority system is simple enough to apply in real time. The production manager does not need to run an optimisation algorithm. They need to ask four questions in order.
Handling the unpredictable
Even with the best framework, job-shop scheduling requires daily adaptation. Here are the common disruptions and their responses:
Machine breakdown: Immediately check which jobs are affected. If the job can move to an alternate machine (even at lower efficiency), move it. If not, recalculate the delivery date and inform the customer the same day. A customer who hears about a 2-day delay on the day it happens is far more understanding than a customer who discovers it when the delivery does not show up.
Material delay: If the material is for a job with large buffer, wait. If the buffer is yellow or red, source the material from an alternate supplier — even at a higher price. The cost of a single late-delivery penalty or a lost customer relationship far exceeds the premium on emergency material.
Operator absence: Cross-train your operators. Every critical machine should have at least 2 operators who can run it competently. A single-point dependency (one operator for one machine) is a scheduling risk that compounds every monsoon, every festival season, and every cricket match.
Quality rejection requiring rework: Slot the rework into the next available window on the relevant machine. Do not queue it behind 3 other jobs — rework on a nearly-complete job is higher priority than starting a new job, because the nearly-complete job has consumed resources across multiple operations and delaying it wastes all of that invested time.
Connecting scheduling to quoting
The schedule is not just a production tool. It is a sales tool.
When a customer asks for a delivery date, the typical Indian job-shop response is a guess based on gut feel: "4-5 weeks." Sometimes the guess is right. Often it is not.
With a visual schedule and finite capacity data, the sales team can answer precisely: "Our CNC plasma is fully loaded until June 15. Welding has availability from June 12. Powder coating has a 3-day queue. Earliest realistic dispatch is June 25, with a buffer."
This precision does two things:
Wins trust. Customers prefer a realistic date over an optimistic one. A manufacturer who says "4 weeks, guaranteed" and delivers on time gets more repeat business than one who says "2 weeks" and delivers in 5.
Enables premium pricing. When you can show a customer that slotting their order in ahead of the queue requires overtime and re-sequencing, the rush premium is justified by data, not by negotiation.
The best job shops in India use their scheduling data as a quoting input. The delivery date is not a promise — it is a calculation. And calculations can be trusted.
What to do this week
- List your machines and their available hours per day.
- List your top 10 active jobs with their remaining operations and due dates.
- Schedule backward from each due date. Which jobs are already at risk?
- Set up a visual board — physical whiteboard with magnetic strips or sticky notes.
- Run a 15-minute morning meeting in front of the board for one week.
- At the end of the week, count how many times the plan changed and why. That pattern tells you where your scheduling system needs the most strength.
Scheduling in a job shop is not about creating the perfect plan. It is about making better decisions faster when the plan inevitably changes. The framework above does not eliminate disruptions. It gives you a structure to respond to them without chaos.
QuoteERP connects your quoting directly to your production schedule — so the delivery date you promise the customer is the date your shop floor can actually deliver. No more guesswork. No more broken promises. See the scheduling module in action →