How Komfortkissen Increased Revenue by 7.5% Through a Structured CRO and Experimentation Program

1. +7.5% Overall Revenue Uplift From a Single Experiment

‍The image-first layout combined with chip-style size selector and updated CTA color delivered statistically significant revenue growth across the primary product pages.

2. +13.9% Conversion Rate Lift on the Premium Topper PDP

‍Benefit-first image ordering: showing product specs and outcomes before lifestyle imagery outperformed the default with 89% odds to win and 8.3% more revenue per visitor.

3. Returning User Behavior Unlocked Through Segmentation

‍Segmenting results by user type revealed that returning users convert 29% better with a dropdown size selector, while new users benefit from chips. This behavioral split points to a personalization opportunity estimated to drive +7.3% incremental revenue from returning users alone.

4. 6 Sequential Experiments Completed in 3 Months

‍A structured roadmap of experiments across layout, UX, visual hierarchy, and CTA design, each building on the learnings of the last delivered compounding improvements rather than one-off tests.

5. Scalable Experimentation Foundation Delivered

‍Full hypothesis backlog, Figma designs, experiment results, and tooling guidance handed over, giving the Komfortkissen team everything they need to continue experimentation independently.

‍

Komfortkissen is an e-commerce brand specializing in premium sleep products including mattress toppers and pillows. Despite strong, consistent traffic and a growing product range, the team knew there was significant untapped conversion potential, but had no structured way to find or unlock it.

The specific challenges coming into the engagement were clear. There was no existing experimentation framework: no process for designing, running, or analyzing A/B tests. There was no heatmap or session replay analysis to diagnose where and why users were dropping off. The conversion rate was below target, with a goal to increase it and grow Q1 revenue by 20–40%. And key product page elements like the size selector, image order, and CTA design had never been validated against real user behavior.

The team was making UX decisions based on assumptions. The engagement was built to replace assumptions with data.

‍

Technical UX/UI Audit

The program started with a full audit of Komfortkissen's Shopify setup and existing tooling: Hotjar and AB Convert, alongside a deep review of their product pages. Using session recordings and heatmap data, key friction points were identified. Most notably, users were barely interacting with the size selector, and the default PDP layout was creating unnecessary cognitive load on mobile, two issues that were invisible in standard conversion data but immediately visible in behavioral data.

Experiment Roadmap

Working closely with the Komfortkissen team, baseline KPIs were defined :conversion rate, add-to-cart rate, average order value, and revenue per visitor. A prioritized experiment roadmap was built with each experiment scoped around a clear hypothesis, success criteria, and targeting logic. Every decision would be data-driven, and each experiment would build on the learnings of the one before it.

Sequential A/B Experimentation

Six experiments ran over approximately three months, each informed by the results of the last.

Experiment 1 tested a dropdown versus the default size selector on mobile. It delivered a +1.9% CVR lift and a statistically significant +11.4% CTR improvement, and pointed toward testing a chips-based layout next.

Experiment 2 replaced the dropdown with a chip-style size selector and restructured the PDP layout. No meaningful CVR difference was found versus control, but chips were confirmed as the superior UX. The lower add-to-cart rate was explained by users seeing price changes inline, a behavioral insight that shaped the next test.

Experiment 3 combined images above the fold, chips, and a new CTA color. AOV increased by +4% with statistical significance, CVR trended positively at +2%, and overall revenue uplift reached +7.5%. This variant was shipped.

Experiment 4 tested two image sequences, lifestyle-first versus benefit-first. The benefit-first order delivered +13.9% conversion rate lift with 89% odds to win and +8.3% more revenue per visitor. Shipped as the new default.

Experiment 5 re-validated chips versus dropdown on the improved layout. The overall result favored chips,. but segmenting by user type revealed a more nuanced story. Returning users converted 29% better with the dropdown, while new users benefited from the visual clarity of chips as they explored sizes and prices for the first time. Rather than a one-size-fits-all selector, the data opened the door to a personalization opportunity, estimated to unlock +7.3% incremental revenue from the returning user segment alone.

Experiment 6 tested an on-brand orange CTA against the current green. Green won with statistical significance, with orange showing -7.93% CVR. Control was shipped.

Handover and Enablement

As the engagement concluded, a comprehensive handover document was prepared covering all hypotheses, Figma designs, experiment results, uplift data, and tooling guidance across AB Convert, Shoplift AI, Hotjar, and session replays. Komfortkissen now has a fully documented, scalable foundation to continue experimentation independently.

‍

The three-month program delivered measurable revenue growth, a validated set of PDP improvements, and a repeatable experimentation process the team can run without external support.

+7.5% overall revenue uplift from the image-first, chips, and CTA color experiment. +13.9% conversion rate lift from benefit-first image ordering on the premium topper PDP. A personalization opportunity identified through user segmentation, estimated to drive +7.3% incremental revenue from returning users by serving each segment the size selector that matches their decision-making style. Six experiments completed across layout, UX, visual hierarchy, and CTA design. A full hypothesis backlog, Figma designs, and results documentation handed over for independent use. Session replays and heatmap analysis established as an ongoing source for hypothesis generation.

‍