I’ve had this argument with clients more times than I can count: “The copy is what sells; the design is just packaging.” So I decided to settle it ...
For further actions, you may consider blocking this person and/or reporting abuse
The "proof before persuasion" point really stands out to me, it makes sense that people need a reason to trust the page before they're willing to believe the claims on it. I also like that you controlled for traffic source, that's the kind of detail a lot of A/B test writeups skip and it makes your 2.9x result much more credible. The scroll depth data is the clincher too, since it shows Layout B didn't just convert better, it actually got more people to read the same words. Definitely bookmarking this as a reminder to test structure before touching copy again.
The "proof before persuasion" idea took me a while to fully believe myself, even after seeing the numbers, since it means the order of information matters as much as the information itself.
Controlling for traffic source felt non-negotiable once I thought about it. If paid and organic visitors weren't split evenly across both variants, someone could reasonably argue the result was about audience quality rather than layout, and I didn't want that gap in the data.
The scroll depth stat is still the one I keep coming back to as well. It reframed the whole test for me. I wasn't just measuring which layout converted better, I was measuring which layout earned enough trust for people to actually read what was already there.
Glad it's useful enough to bookmark. If you try testing structure on your own pages, I'd genuinely like to hear what you find, especially if your audience or offer is different from mine.
This is a great breakdown, thanks for actually running the numbers instead of just arguing about it. The "proof before persuasion" point really stands out to me. It makes sense that people want a reason to trust the page before they're asked to believe the copy's claims. The scroll depth data is the most convincing part too, since it shows Layout B didn't just convert better, it actually got people to read more of the same words. Curious whether you've tested if this proof-first pattern holds for B2C pages too, or if it's mostly a B2B trust thing since the buying process is usually longer there. Solid experiment overall.
Thanks so much for reading it that closely! The traffic source detail almost got cut from the writeup, but I felt like skipping it would've left a hole in the argument. Too many A/B tests get dismissed because someone can point to a variable that wasn't controlled for, so I wanted to close that door early.
The scroll depth number was honestly the part that surprised me most too. I went in expecting a conversion lift, but seeing that people also read more of the identical copy made me rethink what "good copy" even means in isolation. It seems like copy can't really be judged on its own, only in the context of whether the layout gives people a reason to keep reading it.
Appreciate you pulling out "test structure before touching copy again," that's basically the whole point I was trying to make, just said better than I said it. Let me know if you end up running something similar on your own pages, I'd be curious what you find.
This is a great breakdown. The "proof before persuasion" point especially resonates, so many pages ask visitors to trust a claim before giving them any reason to. It also lines up with something I've noticed anecdotally: dense text blocks feel salesy even when the writing itself is solid, so the whitespace change alone probably did a lot of the trust-building work. Curious whether you tested any middle-ground layouts (like just moving testimonials up without doubling the CTA) to see which single change carried the most weight, or if it was really the combination that mattered. Thanks for sharing the actual numbers instead of just the takeaway, that's rare to see.
Thanks, really appreciate that! You're onto something with the whitespace point, I think dense blocks do carry a subconscious "trying too hard to convince you" signal even when the copy itself is solid.
I didn't isolate every single change with its own test, so I can't say for certain which one carried the most weight on its own. My gut says it was the combination rather than one lever doing all the work, but if I had to guess, moving the testimonials up probably mattered more than the CTA change by itself. Proof earlier in the page seems to lower the guard before you even ask for anything. Something I'd like to test properly next time, isolating each change instead of shipping them together. Thanks for asking, it's a good nudge to actually run it that way.
This is a great case study because it isolates a variable most teams never actually test in isolation. The proof-before-persuasion point tracks with how trust works generally, people don't evaluate claims on merit until they have a reason to believe the source first. What stands out most is the scroll depth data. A conversion lift alone could be explained by fewer people bouncing before the CTA, but higher scroll depth means the same words genuinely got read more, so the layout change didn't just remove friction, it earned attention. Would be curious to see this replicated on a B2C page with a shorter consideration cycle, since the "proof first" pattern might matter less when the purchase decision doesn't require as much trust-building. Appreciate that you published the actual numbers instead of just the conclusion.
The scroll depth data is the part that makes this convincing. It's easy to test layout and only look at conversion rate, but showing that people actually read more of the identical copy in Layout B proves the point in a way conversion numbers alone can't. It's not that the words got better, it's that the structure earned the right to be read.
The "decision friction" framing is also useful outside of CTAs. I've seen the same pattern in onboarding flows: two screens with identical instructions, but one buries the action behind extra scrolling or a wall of text, and completion rates drop even though nothing about the actual task changed.
One thing I'd be curious about: did you track whether the second CTA (the one at the end of the page) pulled meaningfully more clicks than the first, or did the early one absorb most of the conversions once people had the trust strip up front? Might tell you whether repeating the CTA is doing real work or just insurance.
This is such a clean way to isolate the variable. The "proof before persuasion" framing is the part I'll keep thinking about, most pages assume the reader already trusts them enough to read a feature list, when really that trust has to be earned first. The scroll depth number is the real proof here too, since it shows the layout change didn't just move the CTA closer, it actually made people willing to read the same words more carefully. Controlling for traffic source was smart as well, that's usually the first thing people poke holes in when a test shows a big swing like this. Would love to see a follow up where each change (testimonial position, CTA repetition, whitespace) is isolated on its own to see which one is really doing the heavy lifting.