Unlearn or expire: Why unlearning speed matters more than learning speed

What is the one habit that got you here that is quietly killing you right now?

I put that question to a room full of founders at the EU-Startups Summit in Malta this May, and I could see it land. Every founder in that room had a habit they were proud of – a personal superpower that closed the early deals, fixed the product, and willed the company into existence. In most cases, that same habit was already working against them.

Most founders treat scaling as accumulation: more knowledge, more people, more process. My experience points the other way. The expertise that made you successful yesterday is the very thing that will kill your company tomorrow.

Scaling demands the painful, constant work of dismantling what worked yesterday. It means unlearning your best habits – which makes your unlearning velocity your most durable competitive advantage.

Across every company I have worked at, the same pattern repeated: the habit we were proudest of became the thing we had to let go of.

Facebook was built around B2C growth and direct performance advertising. The founder was deeply involved in every product detail. Once past €1 billion in revenue, that model hit a wall.

The company had to build a genuine B2B product and sales organisation – a completely different set of capabilities – while simultaneously navigating the platform shift from desktop to mobile, one of the most radical transitions in its history. How the product was developed, how it was sold, and how the organisation was structured all had to be rebuilt.

Airbnb’s early model ran on scrappy local field teams, which were essential for building trust in a platform nobody believed in yet. It worked brilliantly, until it stopped scaling. The company centralised for global efficiency, then later had to rebuild local sales capability as it expanded to serve professional hosts.

Two reversals, each requiring the leadership team to walk away from the approach that had demonstrably worked.

Deliveroo felt like joining a different company every few months, because it was. We went from survival mode, with a small group making decisions on pure instinct after the company came close to bankruptcy, to hypergrowth of 2x in 18 months, which required a decentralised organisation that trusted local teams to drive growth.

Then, as the market demanded profitability, we had to unlearn hypergrowth itself in favour of operational efficiency and sustainable growth.

Three companies, one lesson: at every stage, be ready to abandon the thing that made you.

Between €9 million and €26 million ARR, the founder is the bottleneck. At around 150 people you can no longer know every employee, and the tribal knowledge and informal alignment that powered the early years simply break.

There are too many deals for you to be the hero in all of them. The unlearning here is dismantling the “founder-as-the-OS” model – passing the baton on actual decision rights rather than just delegating tasks.

You have to replace the chaotic energy of the hustle with a Minimum Viable Operating System: regular planning cycles, clear KPIs, and systems that work when you aren’t in the room.

When you scale to €26 million and €87 million ARR, the danger goes underwater. Your dashboard can look perfectly healthy while the company drowns in siloed teams, misaligned roles, and decisions that take weeks instead of hours.

The visible metrics sit above the waterline; the real problems such as decision rights, internal systems, culture sit below it. The unlearning here is resisting the instinct to fix things by changing targets or pushing the team harder.

When a team breaks at this stage, the cause is almost always below the waterline. Fixing the business requires two distinct operational shifts. On one hand, explicitly defining not just what you will do but also what you will stop doing is essential for execution velocity.

On the other, leaders need to reduce systemic friction, eliminating the “meeting tax,” clarifying role boundaries, and simplifying decision-making so teams can execute more effectively.

From €87 million toward the €1 billion mark, you are reading the map from orbit. You can no longer see individual campaigns or product features, and if you are still trying to, you are neglecting the horizon.

The job comes down to three pillars: vision and strategy, culture as infrastructure, and capital allocation. At Deliveroo, that meant managing our international markets as a portfolio – exiting some entirely, doubling down on others – and then getting out of the way of the local teams delivering the work.

The unlearning here is giving up your grip on how the work gets done, and trusting your people to execute.

AI shrinks your margin for error

Founders have always had to unlearn, and companies have always had to pivot, whether they like it or not. What AI changes is the clock.

AI lets competitors replicate your features, code, and content in weeks rather than years. You cannot win a feature war when AI creates parity overnight. The moats that remain are the ones a model cannot copy quickly: trust, proprietary infrastructure, human service, and genuine network effects.

Every transition I described above played out over years. Before AI, you had time to course-correct. The same cycles today are compressed into months.

HR software tools are increasingly commoditised, and AI will only accelerate that. That makes it more important to focus on the areas that are harder to automate, including compliance, expert judgment, and situations where local context matters.

You can automate payroll and time off, but a sensitive termination in Italy may still require someone who understands the business and the local rules. The challenge is knowing when to walk away from the feature race and focus instead on the areas where judgment and trust still matter.

In an AI world, tools become commodities, while human judgment and local expertise remain harder to replicate.

Four questions before you go back to work

Unlearning is hard because it asks you to walk away from the successes that define your identity. AI only makes it harder and more necessary. So before you close this tab, ask yourself:

The companies that reach €1 billion and continue to scale share one trait: they dismantled their old habits fastest and built new ones in their place. The companies that survive the next decade will be the fastest unlearners.