BOM Optimisation: Reducing Product Cost Without Increasing Engineering Risk

The cheapest line item is not necessarily the lowest-cost engineering decision. A serious BOM optimisation exercise asks how each part contributes to function, assembly, sourcing resilience, qualification, compliance and future change. The target is not minimum purchase price; it is the lowest defensible lifecycle cost for the required product performance.

This distinction becomes more important as products age. A design that was commercially sensible at release can accumulate unnecessary variants, duplicated values, obsolete parts, awkward packaging, expensive custom hardware and supplier dependencies through years of engineering changes. The BOM becomes a historical record of local decisions rather than a deliberately managed architecture.

A structured review can recover substantial value, but only if cost is treated as a systems variable. Removing a 5 pence washer can be a poor saving if it increases field loosening. Consolidating three capacitors into one value can be excellent if electrical margin remains adequate. Replacing a £4 component with a £2 alternative may be negative value if it triggers firmware changes, compliance retest, new programming equipment and years of single-source exposure.

1. Treat the BOM as an engineering model, not a spreadsheet

A bill of materials encodes far more than quantity and part number. Each line carries hidden assumptions:

• what function the component performs;
• which parameters are critical and which are incidental;
• what manufacturing process it requires;
• which suppliers and geographies the product depends on;
• what qualification evidence supports it;
• what regulatory or material declarations are attached to it;
• how it is programmed, calibrated, inspected or tested;
• what other parts must change if it changes.

That is why BOM optimisation should begin with product structure and function. The aim is to understand dependency before changing price.

The ISO 10303 family exists precisely because product data is more than geometry. Current ISO 10303 application modules cover assembly structure, configuration and product information exchange across lifecycle activities. You do not need a full STEP/PLM implementation to apply the principle: the BOM should be traceable to the design definition and should not become an uncontrolled purchasing derivative.

2. Optimise total cost, not quoted unit price

A useful cost model separates the price paid for the part from the cost caused by the part. At product level, a simplified lifecycle cost contribution can be thought of as:

Examples are common:

• A lower-cost connector needs manual keying, increasing assembly labour and error opportunity.
• A cheaper semiconductor is available only on full reels with a high MOQ, increasing cash and obsolescence exposure.
• A “drop-in” alternative has different programming requirements, adding production tooling and software maintenance.
• A low-cost fastener requires a second driver bit and creates another line-side bin.
• A custom bracket costs little per piece but carries expensive drawing, supplier and inspection overhead for a tiny annual volume.

For low-to-medium volume products, transaction and complexity costs can be comparable to component purchase cost. This is where engineering judgement matters more than percentage savings on individual lines.

3. Segment the BOM before trying to rationalise it

Not every part deserves the same optimisation method. A useful first pass classifies components into families such as:

Segmentation prevents a simplistic “remove unique part numbers” target from damaging function. Some uniqueness is waste; some is the deliberate result of incompatible requirements.

4. Rationalise components by function and parameter envelope

Part rationalisation is strongest when the engineering team defines an acceptable parameter envelope and then asks whether multiple BOM lines fit inside it.

Consider passive components. Five resistor lines may differ only because historic designers chose E24 values where the function would accept a common value. That is a genuine consolidation opportunity. But apparently similar ceramic capacitors may differ in dielectric, DC-bias behaviour, temperature coefficient, voltage rating or package stress sensitivity; consolidating on nominal capacitance alone can reduce real circuit margin.

The method should therefore be:

1. Identify the function performed by each candidate line.
2. Define the minimum/maximum electrical, mechanical, environmental and compliance requirements.
3. Identify which characteristics are truly interchangeable.
4. Choose a preferred part that satisfies the combined envelope with margin.
5. Quantify the commercial benefit after MOQ, reel size and inventory effects.
6. Validate representative worst-case applications before release.

This is essentially the inverse of component replacement assessment: instead of finding one alternative for one obsolete line, the engineer defines a family-level specification capable of supporting several applications.

5. Design second sourcing rather than adding arbitrary alternates

A second source only reduces risk if it is genuinely usable. An approved alternative that requires an undocumented firmware branch, different stencil, new test limit or hand modification is not a resilient second source; it is a latent engineering project.

For each strategically important line, classify interchangeability:

• Form-fit-function interchangeable: can be substituted without product or process change.
• Controlled alternate: acceptable with defined BOM effectivity, firmware, test or process conditions.
• Redesign option: technically viable but not suitable for uncontrolled substitution.

The approved manufacturer list should preserve that distinction. Purchasing needs to know not merely that two part numbers are “approved”, but under what configuration each one is approved.

For semiconductor alternatives, compare the parameters that matter in the real circuit, including dynamic behaviour, not just headline voltage/current ratings. For mechanical and connector families, include tolerances, plating, mating life, contact system and tooling requirements.

6. Make lifecycle and obsolescence visible at design time

Obsolescence cost is often deferred rather than avoided. A part with excellent current pricing but weak lifecycle position can be expensive if it forces redesign while the product is still commercially active.

IEC 62402:2019 treats obsolescence management as a lifecycle process and explicitly includes strategies to minimise obsolescence during design. That is the right mindset for BOM optimisation.

Useful lifecycle fields include:

• manufacturer lifecycle status;
• years since introduction where known;
• number of credible authorised sources;
• package/technology commonality;
• product-change and discontinuation notification route;
• known regulatory/material changes;
• forecast product life and service obligation;
• last-time-buy implications.

A lifecycle score should not become a rigid algorithm; it is a decision aid. A unique specialist component may be entirely appropriate if it creates essential product value. The aim is to make the dependency visible and managed.

7. MOQ, packaging and lead-time economics belong in the engineering review

Engineering teams often optimise line count without modelling purchasing units. A £0.02 saving is meaningless if the new part is available only in a reel quantity that exceeds several years of demand.

For each proposed consolidation or replacement, model:

• annual usage across all assemblies;
• minimum order quantity and multiple;
• reel/tray/tube packaging;
• lead-time volatility;
• supplier count and authorised-distribution coverage;
• storage/MSL constraints where relevant;
• cash tied in inventory;
• scrap exposure if the design changes.

Sometimes a slightly more expensive common part wins because it turns several slow-moving stocks into one high-turn stock. Sometimes the opposite is true: a proposed common part has a large MOQ and creates more working capital than the existing mixed set.

8. Part-count reduction can save assembly cost — and create new failure concentration

Reducing part count is usually attractive because each part brings procurement, handling, placement, inspection and failure opportunities. IPC-2231 explicitly includes design for manufacturing, assembly, testability, cost and reliability within its DFX scope.

But part-count reduction has diminishing returns and can concentrate risk. Combining two brackets into a complex moulding may remove fasteners but create expensive tooling, long lead time and a single point of supply failure. Replacing two modest regulators with one highly integrated power device may reduce placement count while increasing thermal density and obsolescence impact.

A good review asks what complexity is being removed and what complexity is being relocated.

Part-count opportunities frequently include:

• standardising screw sizes and head types;
• using formed features instead of separate spacers where mechanically justified;
• integrating cable clips or alignment features into moulded/printed parts;
• removing redundant washers where joint engineering allows;
• rationalising connector families and crimp terminals;
• combining low-risk sheet-metal features without creating an unnecessarily difficult part.

9. Compliance data is part of BOM quality

A part number without material/compliance evidence can be commercially unusable in a regulated product. The BOM or linked product data should make it possible to determine which declarations and evidence apply to the released configuration.

IEC 62474:2018+A1:2020 provides a standardised framework for material declarations in the electrotechnical industry. Whether or not a company uses the IEC exchange format, the principle matters: material/substance information should be structured and traceable rather than reconstructed during every customer request or regulatory update.

BOM optimisation should therefore avoid deleting or substituting a part purely on cost when the replacement creates uncertainty around:

• RoHS/REACH or other applicable substance restrictions;
• flammability ratings;
• material grade or colourant;
• UL-recognised systems or component approvals where relevant;
• conflict-mineral or responsible-sourcing declarations;
• country-of-origin/customer contractual requirements.

Commercial optimisation that breaks the compliance evidence chain is not optimisation.

10. Custom mechanical parts often hide the largest avoidable cost

Electronic BOM reviews receive attention because component prices are visible. Custom mechanical parts can hide more structural cost in tolerances, finishes and manufacturing route.

Review each custom part for:

• tolerances tighter than function requires;
• machined features that could be formed, punched or moulded;
• surface finishes applied to non-functional areas;
• unnecessarily exotic materials;
• small radii, deep pockets or thin walls that slow machining;
• sheet-metal bends or hole positions that complicate tooling;
• multiple variants that could be parameterised into one common design;
• cosmetic requirements stronger than the customer's actual need.

Our tolerance stack-up guide explains why the correct response is not simply to widen every tolerance. Relaxation should be based on the functional chain and process capability. A single well-chosen datum or architectural change can save more than a broad drawing simplification.

11. Separate engineering definition from manufacturing effectivity

A BOM change has two dimensions: what the design permits and when production is allowed to use it. Confusing those questions creates traceability problems.

A robust change record should define:

• old and new part/revision;
• reason for change;
• affected assemblies and documents;
• interchangeability classification;
• validation evidence;
• stock disposition;
• serial/date/lot effectivity where required;
• supplier and production instructions;
• any firmware, test or tooling dependency.

This is where the engineering BOM, manufacturing BOM and actual build record need coherent relationships. Modern product-data standards such as ISO 10303 include explicit concepts for assembly structure, configuration and change because product definition is fundamentally relational.

Commonality is valuable, but maximum commonality is not always optimal

Component commonality reduces line count, purchasing transactions, setup changes and inventory fragmentation. Those benefits are real. There is, however, a point at which commonality begins to concentrate rather than reduce risk.

If one regulator, connector family or polymer grade is used across every product, its failure or discontinuation has a much larger enterprise impact. The optimisation problem is therefore not simply “minimise unique part numbers”. It is closer to a portfolio-design problem: reduce unnecessary diversity while retaining enough independence that one supplier or technology event does not stop the whole range.

A practical way to reason about this is to distinguish intentional commonality from accidental monoculture. Intentional commonality uses a preferred component because its parameter envelope, lifecycle position and supplier base make it suitable across applications. Accidental monoculture occurs when one part gradually becomes universal without the combined demand, failure consequence and replacement difficulty ever being reviewed.

For high-leverage common parts, include an enterprise-impact question in the review: if this part became unavailable tomorrow, how many products, customers and production lines would be affected? A part that saves £10,000 per year but creates a single point of failure across £5 million of annual product revenue deserves a different mitigation plan from a low-value unique cosmetic item.

BOM data quality determines how much optimisation is possible

Before analytical optimisation, normalise the underlying data. Duplicate manufacturer numbers entered with different punctuation, descriptions such as “CAP 100N” versus “0.1uF 50V X7R”, inconsistent units and obsolete internal part numbers can make the same physical component appear as several unrelated lines.

A useful clean-up pass establishes:

• one controlled internal identifier per engineering item;
• normalised manufacturer and manufacturer-part-number fields;
• structured value, tolerance, rating, package and technology attributes where practical;
• approved-source relationships rather than free-text alternates;
• lifecycle and compliance status tied to the exact manufacturer part;
• quantity-per and reference-designator information traceable to the released design;
• clear separation between engineering notes and purchasing notes.

This is not clerical work. Poor master data masks rationalisation opportunities and can create false ones. If two descriptions look similar but one represents a safety-approved variant or a different dielectric, automated grouping can become dangerous. Data cleaning should preserve technical semantics, not merely text consistency.

12. Prioritise changes using risk-adjusted return

Not every technically valid saving deserves implementation. A simple decision model can prevent small annual savings from consuming disproportionate engineering time.

Suppose a replacement saves £0.30 per unit on a 5,000-unit annual product: £1,500 per year. If engineering, testing and document change cost £2,500 and obsolete stock write-off is £1,000, the change does not recover its direct implementation cost until well into year three. If the product has only two years of remaining sales, the “30 pence saving” is negative value.

Conversely, a £0.03 passive rationalisation used 500,000 times per year can justify careful validation quickly. The economics must be scaled by usage and remaining lifecycle.

A disciplined BOM optimisation process

1. Freeze a baseline BOM and product revision.
2. Clean duplicate descriptions, units and manufacturer-part mapping.
3. Segment lines by technology, function, value and lifecycle risk.
4. Identify high-spend, high-count, high-risk and low-turn targets.
5. Define engineering parameter envelopes before proposing consolidation.
6. Review second-source and lifecycle opportunities.
7. Include custom mechanical tolerance/process review.
8. Model MOQ, inventory and assembly effects, not just unit price.
9. Rank opportunities by risk-adjusted return.
10. Validate, release and record effectivity through controlled engineering change.

Standards and technical references

• IEC 62402:2019 — Obsolescence management.
• IEC 62474:2018+A1:2020 — Material declaration for products of and for the electrotechnical industry.
• IPC-2231 DFX Guidelines.
• ISO 10303-1:2024 — Product data representation and exchange: overview and fundamental principles.
• NIST — Contextualizing Manufacturing Data for Lifecycle Decision Making.