Optimizing a VPN funnel through behavioral insights
CRO & UX Research @ CyberGhost VPN + PIA, 2023–2024
Company
CyberGhost VPN
Role
Product Designer
Team
2 Designers · CRO Team · Developers
01 — Background
CyberGhost VPN is a global VPN company focused on online privacy and security, serving millions of users worldwide. A large part of the business depended on how users moved through the pricing and checkout funnel.
CyberGhost VPN is a global VPN company focused on online privacy and security, serving millions of users worldwide. A large part of the business depended on how users moved through the pricing and checkout funnel.
02 — The Problem
Our test ideas were built on feedback nobody had refreshed in years.
The pricing page already had ongoing optimization work and constant experimentation. But over time, the team realized something important: we were making decisions based more on assumptions than actual user behavior.
The pricing page already had ongoing optimization work and constant experimentation. But over time, the team realized something important: we were making decisions based more on assumptions than actual user behavior.
03 — Research
We asked one question, two ways.
Phase 1 — Structured (April–May 2023). A two-question exit-intent survey on landing and pricing pages, across both brands: a closed list of reasons (built from our existing assumptions — devices, streaming, servers, pricing, refund) plus an optional open field.
We surveyed leaving visitors on both brands — 150K saw the survey, 5,600+ answered — and compared what they said with heatmaps and analytics. The open answers showed that what we had treated as one problem was actually several different problems.
Round 1 — preset answers, built from our assumptions, Round 2 — one open question, no options
04 — Insights
What users said, and what their behavior showed
36%
too expensive
Price wasn’t always the problem. Confusion was.
Price was the #1 stated reason — but the open answers revealed the texture: users weren't rejecting the price, they were confused by it. Renewal costs, upfront payment for 2 years, and a 30-day money-back guarantee that many misread as a free trial. Quote: "You said no credit card and then asked me for a credit card."
"You said no credit card and then asked me for a credit card."
33%
were existing users
A third of the people who “left” were not there to buy.
Up to 33% were existing customers. They came to download the app, log in, or get support — and the sales pages gave them nowhere to go. We had been counting them as lost buyers, even though buying was never their goal.
"When I go to the homepage, where is the login button?"
73%
never scrolled below pricing
The information was on the page, but it appeared too late.
73% never scrolled past the pricing cards. The information they were looking for — servers, devices, streaming — was further down the page.
The goal became clear: Show the right information before users decide — and give existing customers a way in.
05 — From research to experiments
What we decided to test
The research gave us a clearer direction, but not one final solution. We used experiments to answer three questions:
#1
Does keeping the next step visible help users continue?
#2
The source of truth was people, not files
#3
A system built alone stays alone
The goal became clear: one source of truth, built with the team, that everyone would actually use.
06 — The Experiments
Experiment 1 - A sticky CTA that boosted add-ons
Win
Fully implemented
Why we tested this
Users lost the buy button while scrolling, especially on mobile. And since most users decide fast, the button needs to be there the moment they do.
What we changed
We introduced a sticky checkout CTA that remained visible throughout the pricing flow.
Control
CTA disappears on scroll
Variation
CTA remains visible
What happened
+19.6%
add-on purchases
+6%
average order value
The sticky CTA didn’t increase VPN plan purchases directly, which surprised us. Instead, it increased users’ confidence to continue exploring add-ons.
What we learned - The button wasn't blocking the purchase — it was cutting exploration short. When the buy action stays within reach, users keep browsing. Fully implemented.
Experiment 2 - More content didn’t increase conversions
Inconclusive
Why we tested this
Research showed that many users felt unsure about what they were actually getting. We believed adding more product benefits, visuals, and reassurance below the pricing cards would help users feel more confident before purchasing.
What we changed
We expanded the pricing page with additional benefit sections, feature highlights, screenshots, and social proof placed below the pricing cards.
Control
Short benefit section near pricing
Variation
Detailed product sections added below pricing
What happened
-2.5%
Overall conversion rate
Low interaction
With the new sections
The additional content didn’t improve conversions. Heatmaps showed that most users never reached the new sections.
Users weren’t asking for more content — they were asking for the right information earlier in the flow.
What we learned - Most users never reached the new sections — the test confirmed it the hard way: placement beats volume.
Experiment 3 - Reordering the plans hurt conversions
Lost
Abandoned
Why we tested this
We wanted to understand how much the pricing layout itself influenced decision-making. We believed reordering the plans might help users compare options more easily and choose faster.
What we changed
We moved the highlighted 2-year “best value” plan out of its central position and reordered the pricing plans in a different sequence.
Control
Hero plan centered
Variation
Hero plan moved from center
What happened
-17.5%
conversion
-20%
two-year plan selection
Reordering the plans significantly hurt conversions. More users switched to shorter, cheaper plans instead of the higher-value 2-year option.
The original hierarchy wasn’t just visual styling — it was actively guiding users toward the recommended plan.
What we learned - The order of the plans was helping users compare the options and find the recommended plan. Moving it changed what people chose, so we reverted the change.
07 — Impact
The biggest impact wasn’t a single conversion uplift
it was changing how we understood user decision-making.
We understood who was actually leaving
Not every visitor who left was a lost buyer. Up to a third were existing customers with a different goal.
We protected what was already working
One losing test stopped us from rolling out a change that reduced conversion by 17.5%.
We changed what we tested
We focused on when information appeared, how easy it was to continue, and how the pricing layout affected people’s choices.
Next Case Study
