Production Scheduling for Engineering SMEs – Stop Guessing Delivery Dates

A customer asks whether a batch of precision-machined components can arrive next Friday. Sales checks the order book and says yes. Production then discovers the milling centre is already overloaded, the specialist operator is on leave and the material will not arrive until Thursday.

The delivery promise was made without a dependable view of capacity. For engineering SMEs, scheduling can improve without an expensive factory-wide system.

Why Promised Delivery Dates Slip

Each job competes for machines, skilled people, materials, tooling and inspection time. One change can invalidate the plan.

Research by Land and Gaalman, published in Production Planning & Control in 2009, examined seven make-to-order SMEs. The researchers identified weak capacity visibility before accepting orders and delays during engineering as recurring causes of planning problems. These case findings are not representative of all UK manufacturers.

More recent research published through the US National Institute of Standards and Technology (NIST) in October 2020 examines how integrated enterprise information can support dynamic scheduling when demand and operations change. Decisions depend on reliable information.

The Capacity Check Every Order Needs

Before confirming a date, compare the new order with finite capacity: the hours that genuinely remain after existing commitments, planned maintenance, staffing and realistic operating assumptions. A theoretical 40-hour machine week is not automatically 40 hours of productive capacity.

A Simple Example

Suppose a machining centre has 37.5 scheduled hours next week. Planned maintenance and agreed downtime account for 5.5 hours, leaving 32 hours of usable capacity. Existing released jobs require 27 hours and a proposed order needs 12 hours. Demand would be 39 hours: a seven-hour overload.

The planner now has an informed conversation: move an agreed job, use a qualified alternative resource, approve overtime, subcontract suitable work or offer a later delivery date. Each option requires approval.

This illustrative calculation excludes setup, yield and staffing effects. One bottleneck can determine the completion date.

Five Inputs That Make a Schedule Trustworthy

1. Approved Order Priorities and Due Dates

Use one authoritative order record. Distinguish requested dates from dates the business has actually committed to. Record who may change priority and why, so urgent requests do not silently displace existing customers.

2. Realistic Routing and Operation Times

Identify each production step, its qualified resources and estimated setup and run times. Compare estimates with actuals and correct persistent errors. A dashboard cannot fix unrealistic times.

3. Available Machines and Skills

Account for shift patterns, maintenance, holidays, specialist qualifications and shared tools. Two nominally identical machines may not be interchangeable for a particular component or tolerance.

4. Material and Engineering Readiness

A job is not ready simply because an order exists. Confirm approved drawings, material availability, programmes, tooling and any necessary customer clarification before release. A clear “not ready” status prevents false capacity assumptions.

5. Actual Progress and Exceptions

Operators or supervisors should record meaningful progress at agreed stages, not spend their shift updating software. Capture completion, delays and reasons consistently. Avoid inventing real-time status.

Practical Scheduling Rules for a Busy Job Shop

Protect the bottleneck. Identify the machine or specialist skill that limits throughput. Keep it supplied with approved, ready work rather than flooding every work centre with unfinished jobs.

Separate planned from released work. A tentative job may appear in a forward-looking capacity view, but only authorised work should enter production. Use a clear release gate for materials, engineering and quality prerequisites.

Define when to re-plan. A breakdown, urgent order or material delay may justify a new schedule; every minor change does not. Set a decision owner, an agreed planning horizon and a communication process for affected customers.

Measure schedule stability. Track how frequently promised dates move, which changes were avoidable and whether the plan reflects what happened on the shop floor. Avoid optimising machine utilisation at the expense of overall delivery performance.

What Software Is Actually Necessary?

Start with the system you already have. An ERP may contain orders, routings and inventory but lack an effective visual capacity view. A focused planning application can sometimes use supported interfaces to display workloads and exceptions without replacing the ERP.

Options range from a controlled planning board with disciplined data entry to a finite-capacity scheduler, or a custom workflow that connects approved orders, resources and progress updates. Choose according to constraints and vendor support.

Ficode’s API integration services can help connect approved business-system data, while its data management services address the consistency needed for reliable reporting. Where standard software cannot accommodate a distinctive workflow, bespoke software development may be appropriate.

Do not connect new scheduling software directly to machine controls without qualified operational-technology assessment. Advisory planning is different from machine control.

An Honest Business-Case Example

Consider a hypothetical engineering business that conducts eight substantial schedule revisions per week. Each requires 25 minutes of chasing information and rebuilding the plan, across 46 working weeks.

8 revisions × 25 minutes × 46 weeks = approximately 153 hours per year.

If a better planning workflow reduced that administrative effort by half, it would release about 77 hours of staff capacity. At an illustrative fully loaded labour rate of £35 per hour, that is around £2,680 of capacity value, not necessarily a cash saving.

Potential benefits include fewer late commitments and less expediting. Measure these separately; compare them with integration, licensing, training and support costs.

A Six-Week Production Planning Pilot

Week 1: Baseline One Product Family

Select a manageable group of make-to-order jobs. Record on-time delivery, schedule changes, queue time, the most constrained resource and how long planners spend preparing updates.

Week 2: Clean the Essentials

Confirm work-centre calendars, routing steps, estimated durations, open orders and material readiness. Agree one owner for correcting data and one for approving delivery commitments.

Weeks 3–4: Build a Visible Capacity View

Use a supported export, integration or small application to show load against available hours. Include a readiness flag and an exception queue. Test what happens when a job is delayed, cancelled or moved to another resource.

Week 5: Run in Parallel

Compare the pilot schedule with the current planning method. Let experienced planners challenge unrealistic sequences. Keep the established process as the operational fallback while testing.

Week 6: Decide Using Evidence

Review date-promise accuracy, planner time, overdue work and the quality of exception handling. Expand only if the new approach improves decisions without creating new operational risk. Six weeks may not prove sustained delivery improvements.

Where AI Fits and Where It Does Not

AI or optimisation algorithms may eventually help test alternative sequences, estimate job durations or identify orders at risk. But bad routing times and missing material data do not become trustworthy because an algorithm uses them.

Start with explicit rules, validated capacity and planner oversight. If advanced scheduling is later justified, compare it with a transparent non-AI baseline. Keep commercial commitments and safety-critical decisions under authorised human control.

How Ficode Can Help

Ficode helps businesses examine operational workflows, integrate existing applications and develop tailored software around measurable business needs. A digital transformation review can establish whether the first improvement should be cleaner data, a shared capacity dashboard, ERP integration or a more substantial planning application.

Are You Promising Delivery Dates Without Seeing Your True Capacity?

Discuss a Production Planning & Capacity Visibility Review with Ficode.

Tell us which resource causes the most scheduling conflicts, what software you use and how you currently agree delivery dates. We can explore a proportionate first step without assuming you need to replace your ERP.

Discuss your manufacturing scheduling challenge with Ficode

For practical manufacturing software, data and AI insights, follow Ficode Technologies on LinkedIn.

Frequently Asked Questions

Do We Need a Manufacturing Execution System to Improve Scheduling?

Not necessarily. A controlled capacity view may be enough initially. The right system depends on how many resources, constraints and shop-floor updates must be managed.

Is Finite-Capacity Scheduling the Same as AI?

No. Finite-capacity scheduling considers available resources and constraints. It can use straightforward rules, specialist optimisation or other methods without generative AI.

What Should We Measure First?

Begin with the accuracy of promised dates, schedule changes, bottleneck loading and planner time. Agree definitions before collecting results.