All posts

The best stack is the one you can maintain

Choosing technology is more than a technical decision: it's a business decision. A note on trends, continuity and projects that end up orphaned.

Wooden blocks scattered across a table.

A while back, a relative called me with a problem I’d seen before. They had a working website, but nobody at their company knew how to touch it. The original developer was no longer available. There was no access to the repository. No original files. All that was left was the deployment: a folder of minified files, a couple of credentials and a database nobody dared to look at. The site was still online, but in practice it was dead.

This scene sums up a decision we make all the time without realizing it. Choosing a stack isn’t just choosing how we’re going to build something. We’re also deciding how that project will live beyond us. And that puts a responsibility on the decision that is often ignored.

The question that matters

Every time I start a project, someone asks me what technology I’m going to use. It sounds like a technical question, but it’s really a business question. The answer determines the cost of maintenance, how fast you can change something, whether you can find help, and the risk of being tied to a single person.

For a long time, developers chose out of comfort (and a little out of selfishness). We pushed clients toward the language or framework we knew best, even when it wasn’t the best fit for the project. It was understandable: working with what you know is faster and safer. And if that language or framework was obscure enough that only a handful of developers could use it, that client was ours forever.

Today, with AI assistance, that excuse matters less. Language is no longer such a high barrier. I can walk into a project written in something I haven’t mastered, understand it and maintain it with help. That doesn’t mean experience in a language is worthless: it means it’s no longer the only variable. And it changes the conversation: it’s no longer about what I know how to use, but about what the project needs. More than ever.

At the end of the day, our job is to deliver a solution to the client.

Every so often a new stack shows up promising to solve everything. The docs are beautiful, the examples are fast, the demo looks incredible. It’s easy to get excited.

The problem arrives six months later, when you have to update, migrate or explain to someone else how it works. That’s when the real cost appears: learning time, dependencies that break, version changes, technical debt. Sometimes it’s worth paying. Sometimes it isn’t.

My rule is pretty practical: if the new technology lets me do something I couldn’t do before, or do it much better, I’ll evaluate it. Novelty has a cost you don’t always see in the demo.

The orphaned project

Let’s go back to that first case. It isn’t rare. I’ve seen sites built in unpopular languages, with frameworks almost nobody uses, no documentation and no access to the code. The project works until something breaks. Then the search begins: someone who understands that stack, who’s willing to work on something they didn’t build, who can reconstruct the context from scratch.

Sometimes it’s possible. Sometimes the cost is so high that starting over makes more sense. And the client didn’t just lose the site: they lost the investment, the time and the ability to update without depending on someone who wanted a client “forever” without realizing they wouldn’t always be available for that client.

The question I always ask myself is this: who’s going to maintain this when I can’t? If the answer is “nobody,” the stack is a problem, no matter how modern it is.

The advantage of being common

WordPress has plenty of flaws, but it has one virtue that almost never gets recognized: it’s common. You lift a rock and there’s a developer who knows how to work on it. There’s documentation, tutorials, plugins, hosting, people who know it. If today’s developer isn’t around, anyone can maintain a WordPress tomorrow.

That doesn’t mean WordPress is always the answer. It means you have to value the community, the documentation and the job market as part of the technical decision. A popular stack isn’t automatically good, but a stack without a community is a riskier bet.

The question isn’t just “is this better?” but “is this sustainable?” Do people use it? Is there documentation? Can you find someone else to take over if you need to? How much does it cost to migrate?

It’s not just the language

The stack also includes hosting, the CMS, integrations, external services, the way you deploy. Every dependency you add is one more thing that can break. Every plugin is a pending update. Every external service is an account you have to pay for and an API that can change.

It’s not about living in fear, but about choosing with intention. If a tool doesn’t bring a clear benefit, it doesn’t go in. If it does, it needs documentation and an exit plan.

On this site, for example, I use Astro to generate static HTML, Markdown for content and a handful of JavaScript for interactions. There’s no database, no server of my own, and deployment is copying files. That’s easy to maintain for years. For another project, the answer might be WordPress. For an e-commerce site, Shopify, or Tienda Nube if you’re in Latin America. No option is universal.

A boring stack is a good stack

A colleague once said something I liked and still remember (among many things he used to say): “boring technology is the kind that lets you sleep.” It’s not about avoiding innovation, but about putting innovation where it really counts. The stack doesn’t have to be exciting. The work does.

In the end, nobody remembers whether you used the trendy framework. They remember whether the site loads fast, makes sense and is still working two years later.

Next time you choose a stack, don’t ask which one is the best. Ask which one will still work when you’re not around. That’s the real standard.

Next post

Home office: the balance the office couldn't give me