Operational notes
- Never redesign a site without a control group.
- Test big swings: 'Ugly' vs. 'Polished' is a valid test.
- Look at secondary metrics: Polished sites might increase time-on-page but decrease conversion.
- Data doesn't care about your design system.
Designers hate A/B testing. Why? Because data has a habit of hurting their feelings. Time and time again, we’ve seen a beautiful, award-worthy redesign go live, only to see conversion rates tank by 15%.
Why does this happen? And how can you test “Aesthetics” scientifically?
The “Pretty Perception” Bias
Users say they prefer beautiful sites. If you run a survey, they will vote for the polished version every time. But users vote with their wallets for clarity and ease of use. Attitudinal data (what they say) vs. Behavioral data (what they do).
How to test the “Ugly” Hypothesis
If you want to see if a stripped-down approach works for you, don’t burn your site down. Run a split test on a specific high-traffic landing page.
The Variable: Keep the copy identical. Keep the offer identical. Keep the price identical. Only change the container.
- Control: Your current branded, styled template.
- Variant: A stripped-down, whitespace-heavy, “ugly” version. High contrast, system fonts, few images.
What to look for
You might find that the “Ugly” version has a higher Bounce Rate (people leaving immediately because they think it works poorly) but a significantly higher Conversion Rate for those who stay (because the path to purchase is clearer).
Iterative “Uglification”
You don’t have to go full brutalist. You can test “uglifying” specific components:
- Test a native HTML select dropdown vs. a custom styled React dropdown. (Native usually wins on mobile).
- Test a plain text link vs. a button.
- Test a static image vs. a carousel.
Let the data dictate the design. If the “ugly” version wins, it’s not ugly—it’s profitable.
