The True Cost of Machine Vision: Beyond the Hardware
9 MIN READ 18 August 2026By Calum Browne
When you budget for a machine vision system, the number on the purchase order is the figure that gets approved. It can also be the smallest number you will spend on that system over its lifetime.
A production vision system typically operates for 5 to 10 years. During that time, you will spend money on software licences, support contracts, spare parts, training for new operators, hardware replacements and the engineering time required to adapt the system as your production changes. These ongoing costs are predictable, budgetable and in many cases avoidable if you make the right decisions at the specification stage.
This article breaks down the costs that appear after the installation is complete, so you can budget for the full lifecycle rather than being surprised by them later.
The cost iceberg
The purchase price of a machine vision system is the visible tip. Below the surface sit six categories of ongoing cost that most buyers do not budget for until they appear on an invoice.
| Cost Category | Typical Annual Cost | What Drives It |
| Software maintenance | £500 to £3,000 | Commercial vision software platforms often carry an annual maintenance fee for updates, bug fixes and new features. Some entry-level packages update for free, Zebra Aurora Focus among them but even then the engineering time to apply an update and retest the system is a real cost. Without maintenance you keep the version you bought and lose access to improvements and security patches. |
| Support contracts | £1,000 to £5,000 | Guaranteed response times for production-critical systems. Free standard support covers most needs, but if your line cannot tolerate a 48-hour response time, a paid SLA with defined response and resolution commitments is worth the investment. |
| Spare parts | £500 to £2,000 (provision) | A spare camera, a spare lighting unit, key cables and any other component whose failure would stop the line. This is not a repair cost. It is a buffer cost. The spare parts sit on your shelf and only cost you when you need them but not having them costs you a production line. |
| Training | £500 to £2,000 per event | Training new operators when staff change. Upskilling engineers when the system is upgraded. Refresher training when performance drifts because the team has forgotten how to optimise the system. This is not a one-off cost. It recurs every time personnel change. |
| Hardware replacement | Variable | LED lighting degrades over time (typically 3 to 5 years of continuous use before noticeable drop in intensity). Optical surfaces accumulate contamination. Cameras have long lifespans but are not immortal. PCs need replacing when operating systems reach end-of-life or when processing demands increase. |
| Adaptation and upgrades | Variable | Production changes with new product variants, faster line speeds, additional inspection points, new regulatory requirements. Each change requires engineering time to modify the system. The question is whether this engineering is done by your team (if trained) or by an external supplier (at their day rate). |
What this looks like over 5 years: a worked example
Consider a mid-complexity label verification system with four cameras, custom software and PLC integration. The system costs £40,000 to purchase, install and commission. Here is what the next five years can typically look like:
| Year 1 | Year 2 | Year 3 | Year 4 | Year 5 | |
| Software maintenance | £1,500 | £1,500 | £1,500 | £1,500 | £1,500 |
| Spare parts (provision) | £1,500 | - | - | £500 | - |
| Training (new operators) | - | £1,000 | - | £1,000 | - |
| Lighting replacement | - | - | £800 | - | - |
| Adaptation (new variant) | - | - | £6,000 | - | £6,000 |
| PC replacement | - | - | - | - | £2,500 |
| Annual total | £3,000 | £2,500 | £8,300 | £3,000 | £10,000 |
5-year total cost of ownership: £40,000 to purchase, plus about £26,800 in ongoing costs, gives roughly £66,800.
The ongoing costs add around two thirds to the original purchase price over five years and that is for a system looked after in-house. Where an integrator provides a paid support contract, or where the line changes often and needs regular adaptation, the ongoing total can approach or exceed the purchase price within three to four years.
This is not a reason to avoid machine vision. The ROI on a well-specified system still justifies the investment many times over. The point is that the purchase price alone does not tell you what the system will actually cost. Budgeting for the full lifecycle prevents unpleasant surprises and allows you to plan for the ongoing investment rather than reacting to it.
Five decisions at purchase that reduce your ongoing costs
The biggest lever on total cost of ownership is not negotiating the annual support fee. It is making the right decisions when the system is specified and purchased.
1. Choose open, standard hardware
Cameras that comply with GeniCam standards (GigE Vision, USB3 Vision, CoaXPress) can be replaced with cameras from any GeniCam-compliant manufacturer. If a camera reaches end-of-life in year 6, you can replace it with a current model from the same or a different manufacturer without a major rebuild of the software. Proprietary cameras from closed ecosystems lock you into one supplier's product roadmap and pricing for the life of the system.
2. Invest in training from day one
The cost of training your team to operate and maintain the system is a fraction of the cost of paying an external supplier every time something needs adjusting. A two-day training course costs a few thousand pounds. Five years of external support calls for routine changes costs significantly more and each call takes longer because the external engineer does not know your production environment as well as your own team does.
3. Use software that your team can maintain
Flowchart-based platforms such as Zebra Aurora Design Assistant let an engineer build and change vision applications visually rather than in code. Your own team can learn one of these platforms and then handle routine changes in-house, adding a product variant, adjusting a threshold, modifying a pass or fail rule, without calling the original developer back each time. Fully custom code that only its author understands does the opposite. It ties you to one person and bills you every time production moves on.
4. Specify the right lighting first time
Lighting that is specified properly at the outset gives you consistent images that do not need constant software tuning. You can under-specify in two ways. You can choose a light that only just works for the application, so image quality is marginal from day one, or you can choose a cheap light whose components degrade quickly, so a system that worked at installation drifts within a couple of years. LED output falls over time, and many industrial lights are rated around 50,000 hours, roughly five years of continuous use, so a weak choice here shows up as re-tuning thresholds, chasing ambient changes and eventually replacing the lighting altogether. Getting lighting right is measured in hundreds of pounds. Getting it wrong is measured in years of engineering time.
5. Budget for spare parts at purchase
Ordering a spare camera and a spare lighting unit at the time of the original purchase costs far less than sourcing them urgently when a component fails. The spare parts sit on a shelf and cost you nothing until the day a camera fails at 3pm on a Friday. At that point, the £1,000 you spent on the spare camera saves you an entire weekend of production downtime while you wait for a replacement to be shipped. This is the highest-ROI purchase in any vision system and it is the one that most buyers skip.
The open-source cost trap
Open-source software such as OpenCV is free to download. It is not free to run in production. Any capable engineer can learn it, just as they would learn a commercial library, so the issue is not really who can write the code. The issue is what happens when the system stops working at two in the morning. With a commercial platform you have a vendor to call, documented behaviour to check against and a support contract that puts someone on the hook to help. With open-source you have none of that. You carry the whole risk yourself.
For research and prototyping, open-source tools are excellent. For a production line that has to run reliably for years, the annual cost of a commercial licence is almost always less than the cost of carrying that risk alone. The licence buys you updates, bug fixes and a route to expert help when something breaks. That is not really a cost. It is insurance against the day the system goes down and the person who wrote it is on holiday, or has left.
When the cheaper system costs more
Here is the calculation that changes how people think about machine vision purchasing:
| System A (lower purchase) | System B (higher purchase) | |
| Purchase price | £30,000 | £45,000 |
| Software | Open-source (free) | Commercial licence (£1,500/year) |
| Support | None included | Paid SLA (£2,000/year) |
| Internal maintenance | 1 day/month (£6,000/year) | 2 hours/month (£1,200/year) |
| Training | None (developer maintains) | Operators trained (£2,000 one-off) |
| Adaptation (per change) | £3,000 (external developer) | £500 (internal, Zebra Aurora) |
| 5-year TCO | £30k + £30k ongoing = £60,000 | £47k + £25k ongoing = £72,000* |
System B looks more expensive over five years. That comparison assumes the in-house developer stays the whole time and the open-source code stays maintainable. If that developer leaves in year two, replacing them or bringing in a contractor to decipher bespoke code will wipe out the difference and then some. The lower-risk option is usually the one with commercial software, trained operators and a support contract behind it.
What to include in your TCO calculation
When building a business case for a machine vision system, include these ongoing costs alongside the capital expenditure. Your finance team will appreciate the transparency and the project is less likely to be challenged later when ongoing costs appear that were not in the original approval.
Annual fixed costs: software maintenance, support contract (if applicable).
Periodic costs: spare parts provision (year one, then a top-up every three to four years), lighting replacement (every three to five years) and PC replacement (every five to seven years). A Windows industrial PC tends to reach end of life when its operating system support ends rather than when the hardware fails, so a Linux-based system will often run longer before it needs replacing.
Variable costs: training (every time operators change), adaptation engineering (every time production changes) and unplanned repairs. Adaptation is the one that catches people out. A simple threshold tweak is minor but a genuinely new product variant that needs fresh lighting work, new reference images and algorithm tuning can run from £8,000 to £15,000 depending on how different it is.
Internal costs: your own team's time spent operating, maintaining and troubleshooting the system. This is real cost even if it does not appear on a supplier invoice.
Our Budget Planning Guide includes a worksheet that covers both capital and ongoing costs in a single document. Download it to build a complete TCO estimate for your project.
Plan the full lifecycle, not just the purchase
A machine vision system that is budgeted properly for its full lifecycle will not produce financial surprises. The ongoing costs are predictable, manageable and in many cases reducible through the right decisions at purchase. The five decisions above (open hardware, training, maintainable software, proper lighting and spare parts) collectively reduce ongoing costs by 30% to 50% compared to a system where these are afterthoughts.
If you are building a business case for a machine vision investment and want to understand the full lifecycle cost, our engineering team can help you model the TCO for your specific application and production environment.
Download the Budget Planning Guide
Get in touch: info@clearview-imaging.com | +44 (0)1844 217270
Related: Machine vision pricing | Engineering services pricing | Budget Planning Guide