What I Learned Running a Machine Vision Distribution Business for 25 Years

Part one of two. This article covers building and running the business. looks at how the technology and the market changed over the same period, and what to ask a supplier today.

The short version

• I started Pinnacle Vision in 1997 with one product line and no customers, merged it into Firstsight Vision in 2000, sold a majority stake to Stemmer Imaging in 2004, and left the business in 2024.
• Almost none of the lessons that lasted were about technology. They were about qualification, ownership, quality products, suppliers and people.
• Nearly every serious project failure I was involved in came from a risk sitting outside my own scope of supply.
• Nobody gives away a proper feasibility study for free. If they do, they will skimp on it, and you both lose.
• What separates the distributors that thrive from those that merely survive is not size. It is whether they can demonstrate value, and whether they invest in the people who deliver it.

How does someone end up running a distribution business by accident?

Not by planning it, in my case.

In 1997 I was writing to American imaging companies looking for a job as a European sales manager. One of them, Coreco, wrote back to say they were not looking for an employee. They were looking for a distributor. That reply is the reason Pinnacle Vision existed, and it is the reason I spent the next twenty five years on the distribution side of this industry rather than the manufacturing side.

I mention Coreco for a second reason. Coreco went on to acquire Imaging Technology, then Dalsa acquired Coreco, then Teledyne acquired Dalsa. It is one of the most successful buy and build strategies our industry has produced, and it shaped my business twice over without anyone asking my opinion.

This article is written for the people who buy machine vision: OEMs, systems integrators, and to a lesser extent end users. My aim is to set out what I learned building one of these businesses, so you have a better idea of what to look for when you choose one.

Lesson 1: Is a smaller share of something larger worth more than all of something small?

The first genuinely big decision I made was to give away half of my company.

By 2000, Coreco had acquired Imaging Technology, which happened to be the primary product line of another UK distributor, Vortex Vision, run by David Hearn. I had a choice. I could have taken on that product line outright and kept every share I owned. Instead we merged the two businesses to form Firstsight Vision.

The logic was straightforward once I stopped thinking about equity. The two companies were a similar size and, oddly, we rarely came across each other in the field. Vortex had a strong team, and strong performers in machine vision were not plentiful at the time. Merging made us the largest machine vision supplier in the UK overnight, leapfrogging everyone above us.

I made the same call again in 2004, when Stemmer Imaging bought a 51 percent share in the business. It was Stemmer's first acquisition, branded internally as Stemmer goes European, and I joined the senior leadership team while continuing to run the UK operation. At the time the whole Stemmer group was turning over around thirty million pounds. By the time it was sold in 2017 it was closer to a hundred and thirty million and had become by some distance the largest value added provider in Europe.

What I was actually looking at in both cases was reach and experience: how do I get in front of more customers, and how do I improve what those customers get when they deal with us. Framed that way, the equity question answers itself. I was not weighing up what I would lose. I was weighing up what the customer would gain, and whether the business could deliver it.

Twice the decision was the same, and twice the scarce resource was not the product line. It was the people.

“Twice I chose a smaller share of something larger. Both times the scarce resource was not the product line. It was the people.”

If you are buying vision: when a supplier of yours is acquired, the question is not what happened to the logo. It is whether the engineers you rely on gained capability or lost autonomy.

Lesson 2: What happens when a business has two leaders and two strategies?

I have made this sound tidier than it was.

The merger nearly killed us. We came out of it with too many bosses and two competing strategies running in parallel, and we went from growth to loss inside a short period. We got to within a few weeks of going out of business. Wilhelm Stemmer, who was a small shareholder in Vortex and known to both David and me, lent us the money to keep the doors open.

We recovered, with some guidance from Stemmer, but it cost redundancies and a lot of goodwill. In hindsight two things were obvious. We should have slimmed down our support operation far earlier than we did. And we had started bidding for development projects that carried no prospect of ongoing product supply, which is revenue with no future attached to it. Busy is not the same as profitable.

That episode is also why the Stemmer investment four years later was not a leap of faith for either side. The loan converted to shares, and by then we had been working together for long enough to know how each other operated.

The broader lesson applies to any acquisition, and I have now been on several sides of this one. The hard part is never the commercial logic. It is aligning the vision of both sides afterwards. Change following an acquisition is significant, and it lands on people who did not choose it. Unless both leadership teams are genuinely pulling in the same direction, and are seen to be, you get exactly what we got: two strategies, a confused organisation, and a business that goes backwards while everyone is being reasonable with each other.

If you are buying vision: ask who actually makes decisions in the business you are buying from. Not who holds the title. Who decides. And if your supplier has just been acquired, expect a period of change and ask them directly how the two organisations are being aligned.

Lesson 3: Why did my first big customer stay with us for twenty five years?

Because we did something the rest of the channel was not doing.

In the early days most distributors sold hardware and software tools and stopped there. That left an OEM with two options: build an internal development team, or hand the job to a systems integrator. An integrator wants to supply the whole system, hardware included, because that is where their margin sits. That model works well if you want a vision system installed on a factory floor. It works badly for an OEM who wants vision built into a machine they are going to sell themselves, because they end up paying someone else's margin on their own product.

We filled that gap with development services. The customer we won on the back of it was still a customer when I left the business.

The lesson I took from that was to look for what the channel is not doing rather than what it is doing badly. The lesson I did not expect came later, and it is the subject of the next section.

Lesson 4: What does it really cost to give your engineering away?

Software development, and more importantly software maintenance, takes far longer than you think it will. In our early years I am fairly sure we lost money on the development work. We won overall because that customer bought a great deal of hardware. Had we been selling the development on its own, it would have been a loss.

Support was the same story from a different angle. Whenever something went wrong the customer would ask for a person on site, and very often the problem turned out to be theirs, not ours. The fix was not to become less helpful. It was to write clear operating guidelines and put proper support contracts in place so everybody knew who paid for what before the phone rang. Our competition in development work was the large consulting firms, expensive but genuinely good at identifying the difficult parts of a challenge before committing. We learned to work the same way, with staged plans and milestones so the risk reduced as the project went on. It seems obvious written down. At the time it was all new.

Then there was the day I learned it properly.

One of the largest purchasers of machine vision technology in the country was buying from a direct only supplier. We had a solution that was genuinely novel and substantially lower in cost, so I got in touch. They responded by running an open demonstration day, and almost every supplier in the market turned up to pitch. I worked for three days on our demonstration, including overnight before the event, then drove there for a thirty minute slot.

I still believe we had the strongest proposition in the room. They stayed with the incumbent, and used the whole exercise as leverage to negotiate a better price. We had ten minutes with the decision maker, who was several steps removed from the technical detail.

Getting five companies to build an application demonstration on the promise of a large order is an efficient way to run a procurement exercise. It was an expensive way for me to learn about qualification. After that, I would not commit that level of engineering effort without some commitment coming back the other way.

“If you give your engineering away too easily, the other side does not recognise your value. Very often they do not even appreciate it.”

If you are buying vision: if you want a supplier's best engineering thinking, give them something real in return. Access to the actual problem, time with the person who decides, and an honest view of where you are in the process. You will get better work, and you will find out quickly who is serious.

And do not expect a genuine, in depth feasibility study for free. Anyone who agrees to do one for nothing will skimp on it, because they have to. The purpose of a feasibility study is not to win the order. It is to de-risk the project for both sides, which means it only works if both sides have something invested in it. Stage it, and put the biggest risks first, so that the cheapest possible outcome is finding out early that the project should not go ahead.

Lesson 5: Where do vision projects actually fail?

At the interfaces. Almost always at the interfaces, and I do not mean camera interfaces.

We won a significant project to design a vision system that would be integrated into a larger machine. Someone else was responsible for all the mechanical handling. We built our part, and on our own controlled test rig in the lab it worked flawlessly. We had proved the acquisition.

Then it went onto the real machine, and the precision of that machine was nowhere near the precision of our rig. The results were poor. The end customer withheld the final payment, and we did not get the follow on business. We then spent a great deal of time and money trying to engineer around a limitation that was never ours to fix, while two subcontractors pointed at each other.

What we failed to do was identify where the risk actually was. The risk was not in the imaging. It was in whether the mechanical handling could present the part accurately and repeatably enough for any imaging system to work.

This is not really an argument for running one more test. It is an argument for defining the risk points at the start, writing them down, and then verifying each one deliberately. Every project has a handful of assumptions that everything else rests on. Presentation accuracy was ours, and we never wrote it down, so nobody was responsible for proving it.

The industry has since produced a genuinely useful framework for this. The VDI/VDE/VDMA 2632 guideline series sets out how to approach the specification and the acceptance of a machine vision system, including how to write a requirements specification that both supplier and user can work from, and how to structure an acceptance test. I would encourage anyone specifying a system to read it. It is the closest thing we have to an agreed way of having this conversation, and it exists precisely because so many projects have failed the way ours did.

Alongside that, agree clear acceptance criteria backed by a validated set of test samples, and use the same set twice: once in the lab, and again on the first machine. If the samples change between the two, you have not validated anything. You have run two separate experiments and hoped.

“We proved the vision system worked. What we never proved was that the machine carrying the part could hold it still enough.”

If you are buying vision: list the risk points before anyone quotes, agree who verifies each one, and be clear about who owns the whole chain. If nobody owns it, the gap between two suppliers is where your project will fail.

Lesson 6: When is vision the wrong answer?

More often than the industry likes to admit.

I have had senior people come to me wanting to talk about how vision could help their industry without being able to tell me what problem they were trying to solve. You can inspect almost anything. That does not mean you should. If the process is stable and there is no defect problem, you do not need a vision system, and anyone who sells you one on that basis has done you a disservice.

The most common mistake I see from customers new to vision is assuming it amounts to pointing a camera at something and running a test. The environment is critical. Lighting, presentation, vibration, ambient conditions, part variation. Those decide whether a system works, far more than the camera specification does.

The most common mistake I have seen from value added distributors, including my own business, is making a recommendation without understanding the full picture, and then owning the blame when the system underperforms. The related failure is letting a customer buy the cheapest option without ever explaining what that option cannot do.

Qualification protects both sides. It is not a sales technique.

Lesson 7: What actually makes a customer stay for fifteen years?

This is the shortest lesson in the article and the one I would defend hardest.

Say what you are going to do, and do what you say. Be honest when something goes wrong. Own the problem and go the extra mile to fix it.

Things always go wrong. Every long relationship I had with a customer included at least one moment where something failed badly. Turning up and doing the right thing at that moment is what created the loyalty. Nothing in a marketing programme substitutes for it.

“Say what you are going to do, and do what you say. When it goes wrong, own it. Everything else is decoration.”

Lesson 8: Are good machine vision people a commodity?

No, and treating them as one is the most expensive mistake a supplier in this industry can make.

I was fortunate with hiring. Many people stayed a long time, and several left and came back when the grass turned out not to be greener on the other side. One in particular stands out. He was a PhD student presenting at a conference I was attending. I was impressed enough to talk to him there, and he joined us as soon as he finished.

He stayed a few years. He was deeply technical and also very personable, which is a rarer combination than it should be. He moved on to advanced development work in military applications, then started his own consultancy. We used him. He used our equipment. When someone needed his particular expertise we referred them to him. One of those referrals was a fledgling opportunity at the time, and it took off. That company became one of our largest customers. His technology was so central to what they did that they eventually acquired his business, and he is now technical director of a company with one of the most impressive growth paths I have seen.

The moral is not simply to hire good people. It is to accept they may not be with you for life, to take an interest in their career rather than resenting the departure, and to stay in contact. Good things come back around.

I have also made poor hiring decisions. Get the wrong person into a small technical team and they are disruptive out of all proportion to their role. The real cost is not their salary. It is the good people who become unhappy and start looking.

Machine vision experts take years to build. You cannot recruit your way out of a gap quickly, which is precisely why culture matters commercially and not just morally.

What separates the companies that thrive from those that merely survive?

This is the question I get asked most often, and after twenty five years I think the answer comes down to four things.

They add value, and they can demonstrate that they add value. Businesses that simply move boxes do not succeed in machine vision. The product is too complex and the buying decision is too consultative. If a supplier cannot show you why a particular product is the right one, rather than just confirming it is available, you are dealing with a box shifter.

They have progressive leadership with a can do attitude. The distributor I took my first product line from in 1997 is still trading today at broadly the same turnover as it had nearly thirty years ago. Nothing went wrong. Nothing changed either.

They know their product line properly, and they invest in internal training. Product knowledge is not something you absorb passively. It has to be funded.

They have a nurturing but results oriented culture. These two words are usually presented as opposites and they are not. Experienced machine vision people take a long time to develop and are not a commodity, so you want them to stay. At the same time you have to be honest about who is underperforming, train them properly, build them up, and if that genuinely does not work, let them go. Carrying weak performers is not kindness. It is what makes strong performers leave.

The companies I watched fail rarely failed dramatically. Some of the strong original distributors in our industry were acquired, and interestingly that often created a gap that new competitors, sometimes led by people who left after the acquisition, moved straight into. Others simply bobbed along as lifestyle businesses, where most of the money was taken out and very little was reinvested. Those companies stay small and regional, and they are vulnerable the moment anything changes around them.

Consolidation has raised the bar. As manufacturers consolidate, weaker distributors lose the good lines, and the strong ones get offered dual channel access. As the technical difference between brands narrows, your story matters more. You have to be able to explain in thirty seconds why someone should buy from you rather than from anyone else. A lot of companies in this industry cannot do that, and it shows.

Is scale a friend or an enemy?

Any company that is not trying to grow is in dangerous territory, so I am firmly in favour of growth. What growth requires is different skills at different stages, which is where a lot of owner managed businesses come unstuck.

But scale itself is neutral. Culture is what makes it a friend or an enemy, and I saw both models from the inside.

Stemmer under Wilhelm Stemmer was a family oriented business, and the owner put a great deal back into the employees. The culture was good, and the structure we agreed, operating separately while sharing resources under a service agreement, gave us years of accelerated growth and a very good sounding board for ideas. Some of those ideas, including the tech forum and the machine vision handbook, actually started in the UK and were rolled out from there. If I am honest, the style could be a little too relaxed, with each leader largely choosing their own priorities. The answer always came back that we were growing well above the market average, so we did not need to change.

After the sale to Primepulse in 2017 things pivoted. With external investors, the push for efficiency and growth became a far bigger focus. A lot genuinely did need modernising and it was good to see. But the family feel was lost, and many people who were not used to that kind of pressure left.

Having seen both at close range, I think there is a middle ground, and it is where I now prefer to spend my time: companies with a strong care ethic and a growth strategy that employees can share in. If the focus is entirely on EBITDA, staff become secondary, they do not get rewarded properly, and they leave with low morale. Who decides where the money is spent, and how employees are engaged, is most of the story.

“Any company that is not trying to grow is in dangerous territory. But growth without a motivated team is just a bigger version of the same problem.”

What would I do differently?

Very little on principle and quite a lot on execution.

I would keep the ethics, the culture and the commitment to a strong, well understood product line exactly as they were. Those were never the problem.

I would apply a harder commercial focus in the early years. And I would introduce process earlier as we grew, so that the experience a customer had was consistent regardless of who they happened to speak to. We got there eventually. We should have got there sooner.

Where does this leave you?

If you buy machine vision, almost everything above is really one question asked eight different ways: does this supplier own the outcome, or do they own the invoice?

That distinction does not appear on a datasheet, and it is not the same thing as size. Some of the best engineering support I have seen came from small teams, and some of the most frustrating projects I watched involved very large organisations. What it comes down to is whether the people in front of you can tell you why a product is right, whether they will tell you when it is wrong, and whether they will still be engaged when something fails at three o'clock on a Friday.

The second part of this article looks at the other half of the picture: how the technology and the market changed over the same twenty five years, what I called wrong, and the questions I would put to a supplier today. [LINK: read part two]

A closing note on Clearview

I have deliberately kept this article free of company promotion, because the lessons stand on their own and you can judge any supplier against them, including the ones I used to run.

It is fair to say why I chose to work with Clearview, though, because it follows directly from the sections above. When I look at a machine vision business now I look for the same combination: a genuine care ethic alongside real growth ambition, engineering depth that is funded rather than claimed, and people willing to tell a customer when vision is the wrong answer. That combination is less common than it should be. Where I find it, I am happy to be associated with it.

Measure them against the eight lessons above. Measure everyone against them.

About the Author

Mark Williamson has worked in machine vision since 1989. He founded Pinnacle Vision in 1997, which merged with Vortex Vision in 2000 to form Firstsight Vision, later acquired by Stemmer Imaging, where he served as Managing Director of the UK business and as a member of the group senior leadership team until 2024. He was Chairman of the UK Industrial Vision Association from 2006 to 2016 and sat on the VDMA Machine Vision Board from 2015 to 2024, chairing it for the final three years. He now consults independently with companies across the UK and European vision market, including Clearview. He can be reached via LinkedIn.