Factory automation / Article

Build vs Buy: When a Tailor-Made Machine Beats Catalogue Equipment

A custom automation machine is engineered from your requirements. A catalogue machine is designed to work acceptably for many companies at once. That difference is invisible until your part is unusual, and then it decides everything. Standard equipment breaks in predictable places: awkward geometry, soft or sticky material, mixed SKUs, tolerances tighter than the machine guarantees, and interfaces to equipment that was never meant to talk to anything. Custom costs more up front and takes longer to arrive, so it is only justified when one of those five signals is real. This article shows you where the line sits, what the premium buys, and the hybrid path that is frequently cheaper than either option.

The numbers at a glance

8 to 14 wks
Typical lead time for a single custom cell, with the part and process frozen
4 to 7 mo
Typical lead time for a special-purpose machine
6
Drivers that set custom machine price, none of them a list price
85%
OEE widely treated as a world-class long-term target

01 / The definition

Built from requirements, not adapted from a catalogue

The clearest description we have seen of the difference is about where the design begins. Unlike standard equipment, custom machines are built from requirements rather than adapted from a catalog, integrating mechanical, electrical, controls and software into a single system designed to meet defined performance, throughput and quality targets, and custom is the right answer when standard equipment cannot meet your tolerance, cycle time or integration requirements, when the process must be documented and validated for a regulated environment, or when the production capability you are building is itself a competitive advantage.

The counterpart description is equally useful. Off-the-shelf automation is designed to work well enough across many companies, while custom automation is engineered to deliver repeatable performance in your real production environment. Nothing in that is a criticism of catalogue equipment. Well enough is a genuine achievement at a low price, and for an ordinary part it is the correct purchase.

The trouble starts when a plant buys a machine designed around the average part and then spends the next five years staffing the gap between the average part and their part. That gap does not appear on the quotation. It appears as an operator who straightens every unit before it enters the machine, a rework bench that was never in the plan, and a changeover that takes half a shift.

A useful test

Ask what the machine assumes about your part. Every catalogue machine assumes something: that the product is rigid, that it is square, that it presents the same face every time, that it does not stick to itself, that it arrives within a millimetre. Write those assumptions down and check each one against your worst SKU. Wherever an assumption fails, either a person or a custom module will end up filling the hole.

02 / Where catalogue breaks

Standard machines fail in predictable places

After enough projects the failure modes repeat. These are the five we see most, and any one of them is enough to make a build worth pricing.

01Geometry the feeder cannot orient  Asymmetric parts, near-symmetric parts that must go one way, nested or tangling shapes, parts with no flat reference face. Feeding and presentation is where most standard-equipment projects actually fail, not at the process itself.
02Material that misbehaves  Flexible film, soft foam, sticky or oily surfaces, thin electronics that flex under vacuum, parts that scuff if a gripper touches the wrong face. Catalogue tooling assumes rigid and dry.
03Mix the machine was not built for  A standard machine handles one product family. A plant running many variants pays for that mismatch at every changeover, in time and in the person who has to do it correctly.
04Tolerance or cycle beyond the guarantee  When your requirement sits outside what a catalogue machine specifies, the honest answer from the supplier is that it may work. That is not a specification you can plan capacity on.
05Interfaces that do not exist  Older equipment, a proprietary machine, a line with no common handshake. Integration work is where a cheap machine quietly becomes an expensive one.

Notice that four of the five are about the part and the line rather than about the process step everyone is focused on. That is the recurring lesson of build versus buy: the process is rarely the hard part, and the feeding, presentation, changeover and handoff around it usually are.

There is a sixth case that is strategic rather than technical. If the way you make the product is the reason customers buy it, a machine that any competitor can order from the same catalogue does not protect anything.

03 / The decision

Five questions, one answer

QuestionPoints to cataloguePoints to customWhy it decides
Part geometryRegular, rigid, easy to orientAsymmetric, flexible, tanglingFeeding is where standard equipment fails first
Product mixOne family, stableMany variants, frequent changeoverChangeover cost repeats every schedule change
Volume and cycleWithin published ratesBeyond guaranteed cycle or toleranceA maybe is not a capacity plan
IntegrationStandard interfaces existLegacy or proprietary equipmentInterface work is the hidden cost line
Process lifeUnchanged for the asset lifeProduct likely to evolveAdaptability is what the premium buys
Competitive valueCommodity stepThe capability is the advantageCatalogue capability is available to everyone

Scroll the table sideways on a phone

Score your station honestly. Mostly left, buy the catalogue machine and do not let anyone talk you out of it. Mostly right, price a build. A genuine split usually means the hybrid path in section six, which is where a lot of good projects end up.

One more filter worth applying, from a builder that puts it well: equipment that is mission-critical is often fully customised even when the operation itself is simple, so that it is built to the highest integrity and reliability standards that fit the budget, and metrics stipulated in customer contracts or insurance policies are most reliably achieved with custom-engineered equipment. If the station stopping means your customer's line stops, reliability is worth more than list price.

04 / The premium

What you pay, and how long you wait

Be clear-eyed about the trade. Customised equipment incurs a cost and lead time premium, and is far more adaptable to changing business and technical needs. Both halves of that sentence are true, and a business case that only quotes one half is not a business case.

On price, there is no list. Custom automation price is set by six drivers: part complexity, throughput, number of processes integrated, inspection depth, changeover needs, and controls and safety, so any quote should be treated as scope-specific rather than a list price. This is why two quotations for "the same machine" can differ by a factor of two. They are usually not the same machine.

On time, there are published shapes. A single cell usually takes 8 to 14 weeks, a special-purpose machine 4 to 7 months, and a full integrated line 9 to 14 months, and those timelines hold only when the part and process are frozen, because a design that keeps changing resets the schedule. A simpler test fixture can be quicker: a straightforward automated test fixture might be completed in 8 to 12 weeks.

That freeze condition is the single biggest risk in a build project, and it sits on your side of the table rather than the builder's. If engineering changes the part in month three, the fixtures, the grippers and sometimes the whole presentation strategy change with it. We would rather delay a project start by three weeks to freeze the design than deliver a machine that was correct for a version of the part you no longer make.

Compare the right things. Not purchase price against purchase price, but cost per good part across the asset's life, including scrap, rework, the manual steps a catalogue machine leaves behind, and changeover time multiplied by how often the schedule changes.

05 / Mix and changeover

The hidden bill arrives every changeover

If your plant runs many variants, changeover is where the real money is, and it is the argument for custom that is most often left out of a business case. Manual changeover is one of manufacturing's most persistent inefficiencies, and the fix is not effort. Recipe-based control systems, servo motors and MES integration are cutting changeover times from hours to minutes, allowing plants to say yes to smaller orders and customised products that were once unprofitable.

Custom builders design to that expectation. Machines routinely handle 5 to 20 or more variants with quick-change tooling, recipe-based changeover and servo-driven adjustments, with changeover targets under 5 minutes as standard and many systems changing over automatically with no operator intervention. Flexibility, not exotic capability, is one of the main reasons manufacturers choose custom over standard.

The gain shows up in OEE, because changeover is availability loss. Automation guidance notes that product or tooling changeover that previously took up to 3 hours can be reduced to around 15 minutes with modern equipment, against a world-class OEE target widely put at 85 percent. Two hours and forty-five minutes recovered, multiplied by how many changeovers you run a week, is often a larger number than the labour saving that the project was justified on.

So put changeover in the model explicitly: minutes per changeover, changeovers per week, and the value of an hour of that line. If your mix is high and your changeovers are manual, that single line frequently decides build versus buy on its own.

06 / The hybrid

Keep the base machine. Close the specific gap

The most common good outcome is neither pure build nor pure buy. When a standard machine does the core process well and fails at the edges, the sensible move is to keep it and engineer the edges. When a standard machine gets you most of the way, adding custom modules such as a vision station, a new fixture or a robot to tend it is often cheaper than a full new build, and this hybrid retrofit path keeps a proven base machine in service while closing the specific gap you have.

01Presentation module  A custom feeder, fixture or orientation stage in front of a standard machine. Fixes the most common failure without touching the process.
02Inspection station  A vision station that verifies before or after the standard process, so defects are caught rather than discovered by your customer. This is where machine vision earns its place on a line.
03Tending cell  A robot loading and unloading a machine that was designed for a person, which is the fastest way to add a shift without adding headcount.
04Control layer retrofit  Sequencing, interlocks and data from machines that currently work in isolation, so the line behaves as a line.

The hybrid path also de-risks the decision. You find out whether the gap was really where you thought it was for the price of one module rather than one machine, and if the answer changes the plan you have lost weeks rather than months.

It has a limit. If the base machine's core process cannot hold your tolerance or your cycle, no amount of custom modules around it will fix that, and continuing to add modules to a machine that is wrong at its centre is the most expensive way to arrive at a build.

07 / De-risking the build

Prove it on your parts before it ships

A custom machine is a promise until it has run your product. The formal mechanism for turning that promise into something you can hold a supplier to is acceptance testing, and it is standard practice in this industry: a formal factory acceptance test at the builder's facility and a site acceptance test at yours, with acceptance criteria documented in the purchase order, and guaranteed cycle time, throughput, uptime and quality metrics.

This is exactly how we work. CODETRACE builds tailor-made machines and proves them on your parts in our workshop before they ship, and we mean the real parts: the heavy variant, the flimsy one, the batch that came back from a customer, the units your operators quietly set aside because they never run well. A machine that has only ever handled good parts has not been tested, it has been demonstrated.

01Freeze the part and the process  Before design starts. This one decision protects the schedule more than any other.
02Write acceptance criteria into the PO  Cycle time, throughput, uptime, quality, measured how and over what run length. Agreed before build, not argued after delivery.
03Send real parts, including the bad ones  The awkward variants are the specification. Sending only good samples guarantees a machine that only handles good samples.
04Witness the FAT  Watch it run to the agreed criteria in the workshop. Problems found there cost days. The same problems found on your floor cost production.
05Own the controls  Source code, documented sequence, named faults and an I/O list, so the machine can be maintained and modified by someone other than the builder.

That last point is covered in detail in our article on what a PLC does and why the control layer decides whether your line scales. A tailor-made machine with a locked, undocumented control program is a custom machine you do not fully own.

08 / On your floor

We will tell you when to buy the catalogue machine

CODETRACE builds tailor-made machines and turnkey lines, which makes the next sentence worth saying plainly: if your part is ordinary and your process is stable, buy the standard machine. It is cheaper, it arrives sooner, spare parts sit on a shelf, and any competent engineer can support it. We have told customers this and then quoted them for something else entirely, usually a smaller module that fixed the actual problem.

When a build is the right answer, the sequence is the same every time. We study the station: the part in its worst state, the cycle time you need, the mix you run, what sits either side of it, and the floor and services you actually have. Then we design, build and run the machine on your parts in our workshop, with every actuator sequenced, monitored and interlocked from one control layer, before it ships. Then we commission on your floor and train the people who will run it.

Local build and local support are part of the argument. A machine built down the road gets tuned when your product changes and fixed when it stops, rather than waiting for a visit from another time zone. CODETRACE integrates on site from Shah Alam in Selangor and Batu Kawan in Penang, and is a member of the NVIDIA Inception program.

If you are still weighing it up, start with the station rather than the machine. Our guides on choosing between a cobot and an industrial robot and on the palletizing payback model both start from the same place, which is the work rather than the equipment.

A catalogue machine builds a catalogue part. Check which one you make.

FAQ / Custom automation machines

Questions, answered.

01

What is a custom automation machine?

A machine engineered for one specific process, product and environment rather than adapted from a catalogue. The distinction is where the design starts. Custom machines are built from your requirements: your part geometry, your tolerance, your cycle time, your floor and your existing equipment. Catalogue machines are designed to work acceptably for many companies at once, which is exactly what makes them cheap and exactly what makes them fail on an unusual part. Custom is the right answer when standard equipment cannot meet your tolerance, cycle time or integration requirements, or when the production capability itself is a competitive advantage.

02

When does custom automation beat off-the-shelf equipment?

Five signals, and one is usually enough. Your part has geometry or material behaviour that standard feeders and grippers cannot handle reliably. Your tolerance or cycle time sits outside what catalogue equipment guarantees. The machine has to integrate with existing or older equipment that has no standard interface. Your product mix demands changeovers a fixed machine cannot do. Or the process itself is the thing your customers buy from you, in which case a machine anyone can order does not protect anything. If none of the five apply, buy the catalogue machine.

03

How long does a custom automation machine take to build?

Published guidance puts a single cell at roughly 8 to 14 weeks, a special-purpose machine at 4 to 7 months, and a full integrated line at 9 to 14 months, with the caveat that those timelines hold only when the part and the process are frozen, because a design that keeps changing resets the schedule. Treat that last point as the real risk. Most late custom machines are late because the specification moved, not because the builder was slow.

04

Is custom automation more expensive than catalogue equipment?

On purchase price, usually yes. Custom equipment carries a cost and lead time premium and is more adaptable in return. The comparison that matters is not price against price but cost per good part over the machine's life, including the rework, the scrap, the manual steps a catalogue machine leaves behind, and what a changeover costs you every time the schedule changes. Custom pricing is driven by part complexity, throughput, the number of processes integrated, inspection depth, changeover requirements, and controls and safety scope, so treat any quotation as scope-specific rather than a list price.

05

Can I combine standard equipment with custom modules?

Yes, and it is often the cheapest correct answer. When a standard machine gets you most of the way, adding custom modules such as a vision station, a new fixture or a robot to tend it is frequently cheaper than a full new build, and it keeps a proven base machine in service while closing the specific gap you have. We recommend this hybrid path whenever the catalogue machine's core process is sound and the failure is at the edges, in feeding, presentation, inspection or handoff.

06

How do I reduce the risk of a custom machine project?

Freeze the part and the process before design starts, write acceptance criteria into the purchase order, and insist that the machine is proven on your real parts before it ships. Formal factory acceptance testing at the builder's facility followed by site acceptance testing at yours, against agreed criteria for cycle time, throughput, uptime and quality, is standard practice for a reason. Include the rejects and the awkward variants in that testing, because a machine that only ever ran good parts has not been tested.

07

When should I just buy the catalogue machine?

When your part is ordinary, your volumes are stable, the process will not change over the life of the equipment, and a standard interface exists to whatever sits either side of it. Catalogue equipment is cheaper, faster to get, easier to support and comes with spare parts on a shelf. The mistake is not buying standard equipment, it is buying standard equipment for a non-standard part and then paying for the difference in people, scrap and rework every shift afterwards.

Sources / Every figure in this article

Where the numbers come from

Lead times, price drivers and changeover figures come from companies that design and build automation equipment, so they are labelled vendor and should be read as a shape rather than as a quotation. The OEE benchmark is a widely used industry target rather than a measurement of your plant. The five failure modes in section two are our own observation from projects, labelled as first-party.

01Custom automation equipment buyer guide  Six price drivers; lead times of 8 to 14 weeks, 4 to 7 months and 9 to 14 months; the frozen-design condition; the hybrid retrofit path.Vendor
02Guide to custom machine design  Built from requirements rather than adapted from a catalogue; when custom is the right answer; 8 to 12 weeks for a straightforward test fixture.Vendor
03Custom automation build versus buy explainer  Off-the-shelf equipment is designed to work well enough across many companies.Vendor
04Off-the-shelf versus customised systems comparison  Cost and lead time premium against adaptability; mission-critical and contractual-metric reasoning.Vendor
05Custom automation and special purpose machines  5 to 20+ variants with quick-change tooling and recipe changeover; sub-5-minute changeover targets; FAT and SAT with documented acceptance criteria.Vendor
06Automating changeover for mass customisation  Recipe-based control, servo motors and MES integration cutting changeover from hours to minutes.Trade press
07Flexible automation and OEE  Changeover reduced from up to 3 hours to around 15 minutes; 85 percent as a world-class OEE target.Vendor
08CODETRACE first-party: tailor-made machines built and proven on customer parts in our workshop before shipping, with every actuator sequenced, monitored and interlocked from one control layer.First-party

Send us the part that breaks standard equipment. We will tell you if it needs a build.

Talk to our engineers
Scroll to Top