Lessons from 20 years building online: what changed, what never did

I’ve been building things on the internet for about twenty years. In that time I’ve watched algorithm updates erase traffic overnight, platforms I relied on simply die, “guaranteed” tactics stop working, and now AI rewrite the whole production cost of content. You learn a particular kind of lesson living through that many cycles: not the clever tactic, but what’s left standing after the tactics burn off. Here’s the honest version.

What died (and kept dying)

A graveyard of things that “worked” until they didn’t:

• Every loophole and hack. Whatever the algorithm shortcut of the year was, it closed — and usually punished the people still leaning on it.
• Renting your whole business on someone else’s platform. Channels I depended on changed the rules or shut down, and the income built purely on them went with them.
• “Set and forget” passive-income fantasies. The ones that needed no maintenance and no real value always decayed.

The pattern: anything that tried to skip doing something genuinely valuable had a short half-life. That’s not moralising — it’s just what I watched happen, repeatedly, to me and everyone around me.

What never changed

Against all that churn, a small set of things just kept working:

• Useful beats clever. Content and products that genuinely help people have outlasted every tactic designed to game the system.
• Own your audience. An email list and traffic you control survived platform deaths that wiped out people who only rented attention.
• Trust compounds; shortcuts to it collapse. Reputation built slowly and honestly kept paying; every fast route to fake credibility eventually broke.
• Distribution is half the job. Making something good was never enough — getting it found mattered just as much.

These are unglamorous, and that’s exactly why they’re reliable: they were never the trick, so there was nothing to stop working.

What AI actually changed

The newest cycle is the loudest, so let me be precise about it. AI has crushed the cost of producing content and lowered the barrier to building — which means more competition and an ocean of generic output. But it has raised the value of the things it can’t fake: real first-hand experience, genuine expertise, original data, a trusted name. The whole site you’re reading is built on that bet — E-E-A-T from actually having done the thing.

What twenty years actually taught me to do

If I compress it into how I now operate:

• Build on the constants, treat tactics as disposable. Learn the current tricks, but never depend on them — depend on useful work, owned audiences and trust.
• Build clean and to last. The legal, durable, sellable kind of business is the one still standing after each shake-out.
• Be patient past the point most people quit. Compounding is real but invisible early; almost everyone leaves before it shows.
• Keep my own experience at the centre. It’s the one asset that got more valuable as everything else got automated.

The takeaway

• The tactics, platforms and algorithms churn constantly — build on what survives them.
• What died: loopholes, rented businesses, value-free “passive” schemes.
• What never changed: useful beats clever, own your audience, trust compounds, distribution matters.
• AI raised the premium on authenticity — real experience and trust are worth more now, not less.
• After twenty years I trust the boring fundamentals precisely because they outlived everything flashier.

If there’s one thing two decades hand you, it’s a long enough memory to stop being seduced by the tactic of the moment — and to keep quietly building the durable things that are, predictably, still here.