Case 01 · CRO · Experimentation · Data analysis
What makes someone buy a VPS? I tested three answers
Locaweb’s VPS page serves everyone from small businesses to experienced developers. I planned a test to find out what moves more people toward a purchase: the business benefit or the technical spec.
The challenge
Product page copy was treated as a communication piece, with no measurement of its own. There was no evidence of how much the copy alone affected conversion. Without it, every new page came down to opinion.
Reasoning
Two readings of the same audience
To isolate the effect of the content, the design stayed the same in every version. Version A, the control, was the live page and received 50% of the traffic. Versions B and C, with the new copy, received 25% each, and each one had a numeric hypothesis.
Version B · Persuasion Shipped
For people still comparing options on the market and looking for a solid brand. Focus on business pain points, authority, security and support.
Hypothesis: +20% click-through to the order form.
Version C · Features
For developers and sysadmins who need exact specs before checkout. Focus on hardware, virtualization and networking, in direct, technical language.
Hypothesis: +12% click-through to the order form.
Top of the page in variants B and C. Image, price and CTA are identical; only the copy changes. B: “Turn your code into a business without the infrastructure chaos.” C: “VPS server with maximum performance and native AI integration.”
Process
- Hypotheses written as “if [variable], then [expected impact] for [segment].”
- Both variants written on the same layout.
- Weighted KPIs: revenue and conversion (weight 1), checkout rate (weight 2), CTR and heatmap (weight 3).
- A 30-day A/B/C test, with cross-analysis of clicks, checkout and revenue.
- Planned next phase: a refinement A/B test focused on statistical validity.
Results
The Persuasion version confirmed the hypothesis
Total page conversion for the Persuasion version versus the control: a 57% increase, with the same layout.
| Funnel step | Persuasion vs. control |
|---|---|
| Product page → order form | +19.8% |
| Order form → payment | +21.1% |
| Total page conversion | +57% |
The partial read, at day 8, had a small sample and served only as a direction. At the end of the 30 days, the Persuasion version kept its lead and was shipped to production.
The Features version had the highest revenue per user, but that number came from a few customers with large purchases. We chose Persuasion: its higher conversion was a consistent gain, while the other version’s revenue depended on isolated cases.
Learnings
- Copy is a business variable that can be isolated and measured.
- Reading data in context changes the decision: higher revenue driven by a few orders doesn’t beat higher, consistent conversion.
- The test became a method: the next plan included copy tests on other product pages and a persuasive copy guide for the Content System.