Marcel Duijvestijn

The through-line

Everything Marcel Duijvestijn says circles a single tension: in software the ground under your feet keeps moving, so you have to keep climbing to higher-value territory, but you can only climb as fast as your organisation and your customers will actually follow. On the market side that means treating growth as an obligation rather than an ambition, "In Saas is groei geen keuze. Want als wij niet groeien, ja, dan komt er iemand anders voorbij en die druk je de markt uit" [1], and accepting that yesterday's differentiator is today's utility. On the people side it means the discipline of looking backwards, "Ik kan zelf wel al 10 stappen vooruit zijn, maar als ik dan omkijk en er is niemand met me meegelopen, dan heeft het niet gewerkt" [1].

The same pattern shows up in how he thinks about customers. Automation only works when the person being automated still feels in charge, and product moves only work when the underlying platform stays stable for people who depend on it. His hardest lessons, a failed acquisition and years of learning to let go, both come from underestimating how much slower reality moves than the plan [1].

On why invoice recognition is a commodity

Scanning and recognising invoices is finished as a competitive field. "Als je je nog wil onderscheiden op herkenning, dan gaat er iets niet goed. Dat is water uit de kraan, dat moet gewoon werken" [1]. This is not a fashionable position arrived at late; Blue10 had its own neural network for invoice recognition back in 2017, "Wij hadden al een eigen neuraal netwerk in 2017 voor de herkenning van facturen" [1]. The value has moved up the chain into the approval process and, increasingly, into data-driven insight and spend management [1].

That shift changes what a company like Blue10 fundamentally is. As it becomes more of a data company, one option on the table is to launch a cheap scan-and-recognise offering under a separate label with no purpose other than harvesting more invoice data points [1]. The commodity, in other words, is worth running at low margin if it feeds the thing that is not a commodity.

The adjacent opportunity he points to is spend visibility. Software is on track to become the number two expense category after personnel in Dutch SMBs within three years, and companies routinely overpay by hundreds of thousands of euros through inactive users and unused licences [1]. Accountants, by contrast, feel little pressure to automate, because clients need them regardless of how efficiently they work and the inefficiency is absorbed by the business paying the bill [1].

On automation people will accept

Automation fails when it removes the sense of control. Even when matching is fully automatic, users insist on being able to see afterwards what the system did on their behalf [1]. The model he uses is the intermediate state of a car that mostly drives itself: "Eigenlijk zit hier nu op de autopilot van Tesla. Dus je gaat laten zien wat de bandbreedtes zijn en maar je geeft mensen nog het gevoel dat ze zelf aan het stuur zitten. En we gaan langzaam maar zeker toe naar autonoom" [1]. Show the bandwidths, leave the hands on the wheel, move to autonomy gradually.

Underneath that sits an unglamorous truth about product quality: "De performance is eigenlijk de belangrijkste feature en die wordt pas gemist als die er niet is" [1]. It never wins a demo and it loses everything when absent.

On escaping the feature comparison

In a head-to-head sales battle fought on feature lists, the prospect will always find one thing you do not have. The way out is not to win the list but to stop being on it, by solving a bigger problem through multiple products so you are no longer directly comparable [1]. Blue10 made that move at roughly €5M ARR, deliberately earlier than Jason Lemkin's advice to wait until €10M [1]. The failure mode he names from experience is the opposite: "Wij wilden gaan upsellen, maar we hadden eigenlijk niks te upsellen" [1].

Growth itself is steady rather than explosive, 15 to 25% year over year on the back of that multi-product strategy [1]. And the American canon translates only so far. Lemkin and Jacco van der Kooij are directionally right for the Dutch market, but the absolute numbers usually need halving or quartering [1].

On the acquisition he calls the most expensive MBA

The acquired contract management technology turned out to be a startup proof of concept, which is a fundamentally different object from a product built for customers who are used to a stable platform. It had to be rebuilt, and a 2.5-year payback plan stretched to roughly five or six years [1]. His verdict is affectionate and unsparing: "Nieuws en ik noemen het onze duurste MBA die ooit iemand gaat doen" [1].

On partners, onboarding and churn

SaaS vendors and traditional implementation partners want opposite things. The vendor wants the fastest possible self-service onboarding; the partner wants billable hours, and that incentive quietly pushes products toward complexity [1]. Blue10 stopped partner implementations in 2017 and now uses partners only as lead generators [1]. Onboarding was also given a price, because free onboarding was simply not valued, and a more intensive onboarding measurably reduces churn among young customers with low adoption [1].

On the integration side, the company built a repeatable 12-week pipeline for connecting new accounting systems, first tested with Exact Bouw, with the aim of shipping one new integration per quarter in 2026 so that vertical and industry focus becomes possible [1].

On strategic consistency

Switching costs inside an organisation are enormous, which is why holding a consistent strategic line pays off over time. He contrasts Exact's repeated pivots with AFAS's steady course and its financial results [1]. The counterpart to that consistency is where new products are built: development of something genuinely new has to sit outside the regular organisational and management line, founder mode rather than manager mode in Paul Graham's terms, or it does not survive [1].

On leading change rather than announcing it

His biggest learning as CEO is socio-emotional: leading change means bringing people with you, and being ten steps ahead counts for nothing if you turn round and find you walked alone [1]. Practically that has meant learning to trust, to let go, and to stop reacting to everything immediately [1]. He is emphatic that a leadership development programme for the management team should start earlier than feels necessary, because once people understand each other's drives and behaviour the team becomes dramatically more effective [1].

Takeaways

• Treat invoice recognition as plumbing, not differentiation; the defensible value sits in the approval process, spend management and data-driven insight [1].
• Automate the work but keep the audit trail visible: users accept automatic matching only if they can see afterwards what was done for them [1].
• To escape a losing feature comparison, become multi-product and solve a bigger problem; Blue10 did this at ~€5M ARR rather than waiting for Lemkin's €10M [1].
• Charge for onboarding, because free onboarding is not valued, and use intensified onboarding to cut churn among young low-adoption customers [1].
• Do not let implementation partners own onboarding; their billable-hours incentive pulls the product toward complexity [1].
• Apply US SaaS playbooks for direction but halve or quarter the absolute numbers for the Dutch market [1].
• Build new products outside the normal management line, in founder mode, and hold one strategic course over years because internal switching costs are enormous [1].
• Start leadership development for the management team earlier than you think you need to; mutual understanding of drives and behaviour is what makes the team effective [1].