Last quarter we reworked the primary landing page for a B2B SaaS client — I'll leave them nameless. Original baseline was 2.1% conversion to demo request. Ending CVR after a 3-week sprint was 3.0%. Net lift of 43%.
I want to write this one up honestly because a lot of CRO case studies read like every change was heroic. Most of them weren't. Two of the changes did most of the work; the other five were small wins that added up. Here they are, roughly ordered by impact.
1. Rewrote the headline
This was the biggest single lift, probably around +18% on its own, though it's hard to isolate cleanly.
The original headline was something like "Engineering momentum for modern teams." I don't know what that means and neither did half the people who landed on the page. New version: "Ship code 40% faster — without hiring more engineers." Specific outcome, specific number, specific constraint that their ICP feels in their bones.
There's a temptation, especially with brand-conscious clients, to be clever on the headline. Clever is fine for awareness campaigns. On a landing page getting paid traffic, clear beats clever every time.
2. Shortened the demo form
From 9 fields to 4. We cut phone, company size, industry, use case, and how-did-you-hear. Every field you require costs you somewhere between 1–3% of conversions. Fields the sales team asks again on the call anyway are pure friction.
This lifted CVR around +7%. The sales team pushed back because they thought lead quality would drop. It didn't. Turns out making someone type their phone number before a demo doesn't make them a better lead.
3. Moved social proof above the fold
Customer logos and one specific testimonial with a number in it. Previously these lived two screens down. Now they're visible without scrolling. About +5%. Standard advice, still works.
4. Added an interactive product tour
A 45-second click-through of the product, gated nowhere, above the demo form. This is slightly counterintuitive because you're giving people a way to satisfy their curiosity without converting. But the people who come back after watching convert at a much higher rate, and the ones who don't come back probably wouldn't have converted anyway.
Maybe +5% on CVR, and a meaningful improvement in sales-call show rates because prospects came in better informed. Harder to measure cleanly.
5. Rewrote feature copy as outcomes
"Real-time collaboration with unlimited users" became "Your whole team in the same doc, no version conflicts." Small lift, maybe +3%. This one is mostly hygiene — doesn't hurt to do, doesn't hurt not to.
6. Sticky mobile CTA
+1% blended, but more like +4% on mobile alone, which is where all the gains on landing pages live these days. Easy implementation, no design cost.
7. Sped up LCP
From 3.4s to 1.1s. This one is interesting: measured CVR barely moved on the existing traffic. But bounce rate dropped 22%, which means the funnel was getting more qualified visitors further down. Whether that shows up as CVR depends on what stage you're measuring at. I count it as a win, but I wouldn't have built a whole project around it.
What this doesn't tell you
We had clean data. Enough traffic to run these sequentially with confidence (about 40K visits/month). A buyer's committee that was willing to approve changes in under 48 hours. Most of that is not typical.
Also — I'll admit it — we got lucky on the headline. I've had headline rewrites move CVR by 1%. This one moved it by 18%. That's a tail event, not the expected value. If you're budgeting for a CRO engagement, don't assume that one.
The actual lesson: there isn't one big thing. Seven specific decisions, most of them free to make, stacked into a meaningful number. That's how CRO works when it works.