Why I Stopped Trusting "Best Practices" in Design
“Best practices” in design is mostly a social construct.
Not wrong exactly — but not nearly as reliable as the phrase implies. I spent years treating them like physics. Like something discovered, not invented. I’m more skeptical now.
Where the skepticism started
I was redesigning a client’s pricing page. Classic best practices: annual toggle prominent, most popular plan highlighted, three tiers, CTA button in the hero color. The whole checklist.
Conversion went down.
We ran tests. The version that won had four tiers, a barely visible toggle, and a CTA that looked almost like a ghost button. Everything the conventional wisdom said not to do.
That wasn’t a fluke. I’ve seen it enough times now that I can’t unsee it.
Why “best practices” travels so badly
Best practices usually describe what worked for someone else, in a different market, with a different audience, at a different moment in time. They get written up, shared, repeated — and somewhere in that process they lose their context. The context is the whole point.
“Button contrast should be high.” Sure. But high against what? For which user? On what device?
“Never make users scroll to see your value prop.” True on a landing page for cold traffic. Maybe not true for a product your existing users know deeply.
The pattern I keep seeing: best practices describe the average. And the average is where nobody wins.
What I do instead
I still read case studies. I still know the conventions. But I treat them as hypotheses, not answers.
The useful question isn’t “what’s the best practice here?” It’s “what’s the assumption behind this rule, and does that assumption hold for my situation?”
Most of the time the assumption doesn’t fully hold. Which means the rule needs adjusting — or breaking.
I’d rather have a clearly stated hypothesis I can test than a handed-down rule I can’t interrogate. At least then I learn something either way.
There’s a version of “best practices” that’s genuinely useful: accessibility standards, technical constraints, pattern libraries. Things that don’t depend much on your specific audience or context. Follow those religiously.
But the conversion optimization lore, the UX laws named after researchers, the A/B test folklore — treat all of it as prior art, not gospel.
The real best practice is learning to tell the difference.