Production SchedulingRovaryn Digital9 min read

Job Shop Dispatching Rules: EDD, SPT and Critical Ratio Compared

When three jobs are waiting at the same machine, the rule you use to pick the next one decides who ships late. This guide compares EDD, SPT and critical ratio on one illustrated queue.

Job Shop Dispatching Rules: EDD, SPT and Critical Ratio Compared

Three jobs, one machine, 7:00 AM

At 7:00 AM the horizontal mill is free and three travelers sit in the tray beside it. Job A is a small bracket order the customer has already phoned about; it is due at noon today. Job B is an eight-hour housing run due at 11 AM tomorrow. Job C is a two-hour batch of spacers due around noon tomorrow that nobody has asked about. The machinist looks over and asks the question that opens every job shop's morning: which one first?

Nothing in the shop is broken. The mill works, the operator is certified, the material is staged. What is missing is a rule. Without one, the answer depends on who asked loudest, who walked past last, or which traveler happens to sit on top of the pile. On a day with fifteen jobs waiting in front of six machines, those small coin flips add up to a week of late shipments. This article compares three classic job shop dispatching rules (earliest due date, shortest processing time and critical ratio) on that one small queue, with every hour shown, so you can see what each rule does to the jobs waiting behind it.

What job shop dispatching rules actually decide

Scheduling and dispatching are related but not the same job. Scheduling lays out the plan across machines and days. Dispatching is the local decision made each time a machine becomes free and more than one job is waiting for it. A dispatching rule turns that decision into arithmetic: give every waiting job a priority number, run the lowest (or highest) one, and repeat when the machine frees up again.

The appeal is that these rules are transparent. Anyone on the floor can understand "earliest due date first," and a supervisor can defend it to a customer. There is no hidden weighting. Each rule simply looks at different information: one looks at the due date, one at the run time, and one at both plus the work still ahead of the job.

A rule only ranks the queue, though. It does not create capacity. If the mill has thirteen hours of work in front of it and eight hours in a shift, no ranking changes that, which is why the rule sits on top of a finite-capacity view of the shop, as covered in finite capacity scheduling explained. For the wider picture of how sequencing fits into running a shop, see the complete guide to production scheduling for job shops.

The three rules in plain terms

Earliest due date (EDD). Sort the waiting jobs by due date and run the earliest first. It needs only one piece of data, which is why it is the default in many shops. On a single machine, EDD is the textbook rule for minimizing the largest lateness of any job in the queue. Its blind spot is run time: it will happily start a long job because the due date is near, even when a short job could have cleared the machine and been out the door first.

Shortest processing time (SPT). Run the quickest job first. SPT moves jobs through the machine fast, which lowers the average time a job spends in the queue and keeps work-in-process from piling up. Its blind spot is the due date: it ignores it entirely. A long job can be bumped by shorter arrivals again and again, and it gets late while the machine looks busy and productive.

Critical ratio (CR). Divide the time left until the due date by the work left on the job, then run the lowest ratio first. A ratio of 1.0 means the job is exactly on pace, below 1.0 means it is behind, and above 1.0 means it has slack. Because the work-left figure covers every remaining operation on the traveler, CR is the only one of the three that sees the rest of the route. As a quick worked illustration: a job due in 10 working days with 9 days of work left has a ratio of 1.11. A job due in 10 days with 2 days of work left has a ratio of 5.0. The first needs attention today and the second can wait. The ratio must be recalculated as time passes, and it is only as good as the remaining-work estimates behind it.

A dispatching rule never makes a late job on time. It decides who absorbs the lateness.

One queue, three sequences

Here is the mill queue from the hook, reduced to round numbers. This is an illustration, not shop data. Assume the mill runs continuously, all three jobs are ready now, each has one operation left, and due dates are counted in working hours from 7:00 AM.

Job Run time Due (hours from now) Critical ratio at hour 0
A 3 h 5 5 ÷ 3 = 1.67
B 8 h 12 12 ÷ 8 = 1.50
C 2 h 13 13 ÷ 2 = 6.50

The three jobs add up to 13 hours of machine time. Now apply each rule and add up the finish times.

Rule Sequence Finish times (hours) Late jobs Average finish time
EDD A, B, C 3, 11, 13 None 9.0 h
SPT C, A, B 2, 5, 13 B, by 1 h 6.7 h
CR B, A, C 8, 11, 13 A, by 6 h 10.7 h

Check one row by hand. Under SPT, C runs first and finishes at hour 2, due at 13. A runs next and finishes at hour 5, which is exactly its due time. B runs last and finishes at hour 13, one hour past its due time of 12. The average of 2, 5 and 13 is 6.7.

Three things stand out. First, the machine finishes everything at hour 13 under every rule. The rule changed who was late, not when the work got done. Second, SPT produced the best average finish time and still shipped a job late, because the average hides the one job that mattered. Third, CR did worst here, and it is worth understanding why. At hour 0, B has the lowest ratio because eight hours of work are squeezed into a 12-hour window, so CR starts B. But A is due in five hours with three hours of work, and one long job in front of it is enough to sink it. With a single operation left per job, critical ratio is essentially a slack measure, and it has no downstream route to look at. Its advantage appears when jobs carry several remaining operations, which this queue does not.

Do not generalize from three jobs. Change one due date and the ranking can shift. The point of the exercise is the method: write the queue down, apply each rule by hand, and look at which jobs end up late rather than at the average alone.

Where each rule helps and where it hurts

EDD helps when the shop's problem is late shipments and customers are measuring you against promised dates. It is simple, predictable, and easy to explain. It hurts when due dates are padded or arbitrary, because the rule faithfully sequences by a number that does not reflect real urgency, and when a long job with a near date blocks several short ones.

SPT helps when the queue itself is the problem: crowded benches, too much WIP, travelers lost in the pile. It hurts when long jobs carry the important customers, because they sit and age. Shops that use SPT usually add a guard, such as promoting any job that has waited past a set age.

CR helps on long routings, where a job can look fine at its current machine and still be doomed by the five operations ahead of it. It hurts when remaining-work figures are guesses, since a wrong run-time estimate produces a wrong ratio with a decimal point that looks precise. Already-late jobs also sort to the top automatically and can crowd out jobs that could still be saved.

Real queues carry constraints that none of these job shop dispatching rules sees. A changeover that groups two similar jobs back to back can beat any ranking. A job at the top of the list may need an operator who is not certified on that machine or not on shift. A rush order is simply a manual override of the rule, and it is worth recording as one, with a reason, so the exception does not quietly become the rule. The rush order scheduling guide covers how to weigh that override against the jobs it displaces.

Choosing a rule without a black box

Start from the pain, not from the rule; the best of the job shop dispatching rules is the one you can explain and measure. If customers are calling about late shipments, begin with EDD and measure. If the floor is clogged with half-finished work, try SPT with an age limit. If your jobs run through many operations, move to CR once the routing times are trustworthy. Many shops settle on a hybrid, such as EDD with the shorter job winning ties.

Then measure what the choice does. For context, typical on-time delivery for job shops is 85–95% (User Solutions, 2026), while good on-time delivery is generally 95% or higher (SourceDay, 2025). Those are reference points for where your own number sits; they say nothing about which rule gets you there, and no benchmark can tell you that for your queue. Track on-time delivery before and after a change, over enough weeks to see past one unusual month. The on-time delivery improvement guide and the overview of production scheduling metrics and KPIs show how to set that measurement up.

MachineScheduler follows the same philosophy as the rules above: it is rule-based, with no black box, and you control the schedule. It does not pick the next job for you or optimize the sequence. You decide the order, drag operations on the Gantt board to set it, and read due-date risk as green, amber or red recalculated live. Finite-capacity enforcement and conflict detection are available on Professional and above; Essentials shows the Gantt board but does not enforce capacity or detect conflicts.

Try the rules on paper first

You do not need software to test this. Pull next week's open jobs, list the due date, the run time and the priority for each, and sequence them by EDD, then SPT, then CR. See which jobs go late under each. The Job & Work Order Log with Operation Sequencing is described as a structured register for the jobs and work orders on your floor: one row per job carrying customer, due date, priority and status. It gives you a single place to keep the due dates and priorities that every rule on this page depends on. You can download the Job & Work Order Log with Operation Sequencing and run the comparison on your own queue.

Ready to go beyond the guide?

Most shops are on a live Gantt board within 60 minutes of sign-up, with their existing job list imported from Excel.

Browse scheduling templates for job shops →

Get shop floor scheduling guides in your inbox

Practical articles for production managers — no spam, unsubscribe anytime.

Related articles