EP Beauty's online sales were significantly lower than the sales generated in its clinics. The demand and the product were clearly there, proven by the clinics' in-person success, yet that wasn't translating online. As the brand's e-commerce platform, the website was where that gap needed to be closed.
The business knew the website was underperforming, but not why. Without behavioural data, it was impossible to identify where users were dropping off or which parts of the journey were causing friction.
S
My task was to uncover where customers were disengaging across the online journey, using Mouseflow behavioural data to identify friction points and provide the evidence needed to guide the redesign.
Understanding the Users
The digital audience was already there. Over one month (18 May – 18 June 2025), Mouseflow tracked 498 users across roughly 500 sessions on the live site. With most traffic coming direct rather than through search or ads, the data confirmed the initial thinking: EP Beauty had no digital awareness problem, and the issue lay in the website experience itself rather than in how users were reaching it, as existing traffic was failing to translate into online sales.
The findings were synthesised into key insights that shaped the direction of the redesign.
Traffic Sources
1. Users were searching, not finding.
With a click-to-view ratio of 4.22 but only 27 seconds of engagement per session, users were clicking frequently while spending very little time on any single page. The pattern pointed to people navigating in search of something specific and failing to land on it.
Site Averages
2. The majority were on mobile.
73.7% of all traffic came from mobile devices, against 26.1% on desktop. Yet the experience hadn't been built around that reality, leaving the largest group of users on the site it suited least. Mobile needed to be the primary context the redesign was built around, not an adaptation of the desktop layout.
Device Breakdown
3. Users were leaving at the first screen.
A 58.1% homepage exit rate meant more than half of all sessions ended where they began. Although the hero drew a 96.8% entry rate, attention thinned sharply beyond it, with average scroll depth reaching only 32.8% against a fold line at 782px, leaving roughly 68% of the content unseen. Users landed, engaged briefly with the top of the page, and left before the site had a chance to show them anything. The homepage needed to earn attention at the top and make the rest worth discovering.
Homepage Engagement Zones
Homepage Scroll Depth
4. Friction peaked where users were closest to buying.
Subscriptions exited at 83.3%, product pages at 75% and 67.9%, and account at 69.2%, alongside the blog page at 66.7%. Product pages listed ingredients and formulation detail but answered none of the questions a buyer actually asks: does this suit my skin, how do I use it, why should I buy it. The pages were static, offering nothing to interact with or explore at the moment users were weighing the product up. The blog pointed to something else. Scroll depth ran high while engagement time stayed short, indicating that users were drawn in by the story-led headlines but skimming the content once inside. The concept of skincare stories and ingredient education was capturing users, but it was doing nothing with that interest, offering no meaningful connection to EP's products. Account and subscriptions failed differently, leaving users unable to see what they were signed up to or how to change it, a gap the support team absorbed daily. These pages carried the highest intent and the weakest support, and needed rebuilding around what users actually needed to decide and manage.
Exit Rates
Ideation
Insights derived from Mouseflow behavioural data informed the ideation process.
Ideation 1 — Insight 1: Users were searching, not finding
ID1.1 Parallel information architecture: Flatten the nested category tree into equivalent top-level routes, so each way of thinking about the range is a starting point rather than a submenu.
ID1.2 Recognition over reading: Make navigation visually identifiable at the point of the click to reduce wrong turns.
ID1.3 Faceted filtering: Let users narrow the range to their own requirements.
ID1.4 Content-led discovery: Let content act as a path into the range rather than an endpoint.
ID1.5 Decision shortcuts: Surface a default choice for users who arrive without a specific target.
Ideation 2 — Insight 2: The majority were on mobile
Mobile-first design: Treat mobile as the primary context rather than a secondary one.
ID2.1 Touch-friendly targets: Position primary actions within natural thumb reach.
Ideation 3 — Insight 3: Users were leaving at the first screen
ID3.1 Above-the-fold: Place the highest-value entry point within the first screen, so the shortest possible session still has somewhere to go.
ID3.2 Visual hierarchy: Structure the page so attention carries downward rather than settling at the top.
ID3.3 Visual emphasis: Give commercial offers enough weight that a passing scroll notices them.
ID3.4 Section-level actions: Give each section of the homepage its own entry point, so users can move inward from wherever they stop.
ID3.5 Sticky navigation: Keep a way forward within reach wherever a user stops.
Ideation 4 — Insight 4: Friction peaked where users were closest to buying
ID4.1 Information hierarchy: Lead with what a user needs to know, not with everything there is to tell.
ID4.2 Progressive disclosure: Keep detail collapsed by default so users expand only what they want to see.
ID4.3 Trust signals: Carry the expertise behind the brand onto the page itself.
ID4.4 Interaction at the point of decision: Replace static presentation with something users can act on while deciding.
ID4.5 Clear CTAs: Make the available action obvious at every point of commitment.
ID4.6 Contextual commerce: Make content part of the buying journey rather than separate from it.
ID4.7 Account transparency: Make what a user has committed to visible and changeable at any time.
Design & Prototyping
Each ideation principle was translated into a specific design decision across the site.
Image 1–10 veya Video 1–10 bağla
A
Image 1–10 veya Video 1–10 bağla
A
A month of behavioural data became a structured set of design decisions, each one traceable to something users had actually done. Product pages were rebuilt around what a buyer asks rather than what a formulation contains, with clinical answers on the page and an AR view on serums. Content that people were already reading now leads to the products behind it, and subscriptions can be managed without contacting support.
The redesign hasn't gone live yet, so its impact is untested. The next step is a preview release to EP Beauty's mailing list to collect preference and open feedback at scale, then measuring the live site against the same Mouseflow metrics that shaped the work.
Gallery
Image 1–10 veya Video 1–10 bağla
EPBeauty
Image 1–10 veya Video 1–10 bağla
EPBeauty AR Experience
