Spain’s national statistics institute, the INE, reported in November 2025 that 41.9% of Spanish companies born in 2018 were still trading five years later. Almost six in ten had gone. And the detail I find most telling sits in the small print of that release: the first year takes the biggest bite, with survival rates of 78.5% or lower.

A company rarely closes because it misread the trend of the moment. It closes because something it depended on — a client, a channel, a price — stopped working, and there was no way to replace it in time.

Which is why resilience is not a forecasting exercise. A resilient company does not guess better than the rest. It gets things wrong just as often, and keeps the ability to correct itself when the assumption it started from stops holding up.

Fragility almost always has a name

When a fast-growing company breaks, it is hardly ever a mystery. Three dependencies show up again and again, and none of them is fixed by a headline.

  • A single acquisition channel. While it works it looks like efficiency: the whole budget goes where it converts best. The catch is that the company does not own that channel. An algorithm change or a pricier auction is enough to double acquisition cost in a quarter, with nothing inside the business having changed.
  • A vendor with no costed exit. Depending on someone is not the problem. Not knowing what leaving would cost is. If nobody has put a price and a date on the migration, that is not a vendor: it is a partner with a veto over your margins.
  • Processes that live in one person’s head. They work beautifully until the day that person resigns, falls ill or moves on. That is when you find out it was never a process. It was a habit, well executed.

The same institute, in its central business directory as of 1 January 2025, counts 18.9% of active Spanish companies as twenty years old or more, and 19.9% as under two. Lasting a long time is clearly possible. What almost never happens is lasting without having rebuilt yourself internally several times.

Technology should leave doors open

A good technical decision solves today’s problem without closing off the one three years out. That sounds obvious and gets broken constantly, because the option that shuts doors is usually also the quickest to get running.

We have written about both sides of this: the EU Data Act turns switching provider into a right, and the shutdown of the copper network in Spain showed what it costs to keep a dead technology alive. Freedom to move is paid for up front, or paid for with interest.

What we ask for before signing

Inside the group we have a rule we have tightened over the years: nothing enters the technology core our companies share without written answers to two questions. What happens if that vendor triples its price tomorrow. And how long it would take us to get out, with the name of the person who would do it and the data exportable in a format we can read without asking their permission.

This is not distrust, it is negotiation. When you know what leaving costs, you argue about terms. When you do not, you sign whatever is put in front of you and call it a deal.

Where we set the bar on novelty

The second rule cuts against enthusiasm, which is why it is harder to hold: we do not count adopting a technology as an achievement. We count one specific process working better with it than without it, measured before and after. Adoption is cheap and sets nobody apart. What cannot be copied in a quarter is customer knowledge, your own data, and being inside a process the client no longer wants to touch.

A solid company is not one that holds still. It is one that can move without asking anybody for permission.