The 6 Elements of Your Website Worth Testing
Button color isn't worth testing. Headlines are. Here are the six elements that actually move conversion rates.
Most A/B testing advice is about method. Sample size, statistical significance, test duration. Almost none of it is about what to actually test.
Here's what I think after running hundreds of tests: you should almost never test button color, font size, or image crops. Those are rounding errors. Test the elements that answer "should I do this?"
Six of them matter.
1. Headlines
Your biggest lever by far. The hook decides if a visitor stays or bounces within three seconds.
Test framing (loss vs gain), specificity (number vs adjective), and emotional angle. "Stop losing 30% of your leads" beats "Get 30% more leads" most of the time in B2B. Loss aversion is a real cognitive bias and it shows up in test data consistently.
2. Subheadlines
The bridge between hook and action. One benefit, zero fluff.
The mistake most sites make: the subhead just restates the headline in different words. Waste of space. Use it to make the headline's promise concrete.
3. CTAs
First-person, benefit-led CTAs beat generic ones by 20-30% in most funnels. "Start my free trial" beats "Sign up" almost every time I've tested it.
Test pronouns ("my" vs "your"), verb specificity ("Book a 15-min demo" vs "Contact sales"), and whether the action is framed around what you give ("Get pricing") vs what the user does ("See pricing"). Present tense tends to win.
4. Button labels
Distinct from CTAs. These are the actual words on the button.
"Start free" beats "Sign up" by 20%+ almost universally. "Get my pricing" beats "Contact sales" for SaaS. "See plans" beats "View pricing" more than you'd expect.
This is micro-copy with macro impact. Four words on a button can swing a conversion rate by a third.
5. Body copy
The long-form stuff on landing pages, product pages, and blog posts.
Test length (concise vs detailed), structure (benefits-first vs story-driven), and tone (professional vs conversational). You'll usually find one length wins decisively — which tells you whether your audience wants to be convinced or informed.
6. Product descriptions
Outcome-focused. Concrete. Specific.
"Ship 3× faster" beats "AI-powered workflow automation." Always. Tell people what they get, not what you built.
What not to test
Button color. Font size. Image crops. Spacing. Alignment.
These can matter for accessibility and brand consistency, but they rarely change conversion rates enough to care. Test once, move on.
The wins are in the copy that changes the decision. Everything else is polish.
Stop guessing which copy converts
Helix is an AI agent that runs A/B tests on your website copy 24/7 — writing variants, measuring results, and learning what works for your audience.
Start Free