UX Audit E-Commerce: Why Customers Buy Less Online
A brick-and-mortar drugstore chain approached us with a specific problem. Their physical stores averaged 38.99 PLN per basket with 6 products. Their online store, selling the same products at the same prices, averaged 34.99 PLN per basket with only 4 products.
That is a 33% drop in items per cart. Not because of pricing, assortment, or demand. The products were identical. The prices were identical. The only variable was the buying experience itself.
This UX audit e-commerce case study documents exactly what we found, what we recommended, and what changed after implementation.
The Baseline: Physical Store vs Online Store Performance
Before diving into methodology, it is worth understanding why this gap matters commercially.
A customer walking into a physical drugstore picks up 6 items and spends 38.99 PLN. The same customer, browsing the same retailer online, adds 4 items and spends 34.99 PLN. Over thousands of transactions per month, that per-cart difference compounds into significant revenue loss. If the online store processes 5,000 orders per month, those 2 missing items per cart represent tens of thousands of PLN in unrealized sales every month.
The interesting part: this was not a marketing problem. Traffic was healthy. Product pages existed. Prices matched the physical store. The gap was purely experiential.
The client had Google Analytics data and point-of-sale records from physical locations. They could see session counts, bounce rates, and average order values. They knew the gap existed. They did not know why. Their hypothesis was that online product discovery was broken, that customers simply could not find what they would normally buy in-store.
Our job was to confirm or reject that hypothesis through a structured UX audit e-commerce analysis, identify the specific friction points, and deliver actionable fixes with measurable outcomes.
UX Audit Methodology
The client came to us without any user personas. They had raw data (GA sessions, bounce rates, cart metrics, physical store transaction logs) but no structured model of who their customers actually were and how they behaved.
We started with workshops.
Persona Development
Using the physical store data combined with GA demographic segments, we built two personas through structured workshops with the client's team. The primary persona was Anna: a 35-year-old mother of two, dog owner, who does most of her online shopping in the evening after putting the kids to bed. The secondary persona covered a different demographic segment, but Anna drove the majority of the audit because she represented the highest-volume customer type.
This aligns with a core UX psychology principle: if you do not understand the mental model your user brings to the interface, you cannot diagnose why the interface fails them.
Shopping List Simulation
With Anna defined, we built a realistic shopping list based on physical store purchase patterns for her demographic. Baby food, hair accessories, children's toothpaste, hybrid nail polish, pet supplies. Then we walked through the online store trying to find and purchase every item on that list.
This cognitive walkthrough method is detailed in our full persona analysis. The goal was not to test whether items existed in the catalog. They did. The goal was to measure how much effort it took to locate and purchase them.
Google Analytics Data Review
Before running any qualitative tests, we analyzed the existing GA data to identify where users were dropping off. Bounce rates on category pages were elevated compared to industry benchmarks. The internal site search had a high exit rate, meaning users who searched often left immediately after seeing results (or lack thereof). Cart abandonment was above average, but not dramatically so. The problem was upstream: users were not getting enough items into the cart in the first place.
This data-driven approach matters. Too many UX audits start with subjective opinions about what "feels wrong." We started with numbers. The GA data told us where to look. The persona walkthrough told us why.
Testing Scope
We audited both desktop and mobile versions of the store. Given that Anna's profile indicated evening shopping (likely from a phone on the couch), mobile UX was a critical priority. On mobile, every friction point from desktop was amplified: smaller tap targets, more scrolling to browse products, and less screen space for navigation cues.
Our focus areas were: search functionality, navigation and categorization, cart abandonment signals, and upselling/cross-selling mechanisms. We also evaluated the store against established UX design principles to identify where standard best practices were being violated.
Finding 1: Broken Product Discovery
This was the single biggest problem. The store's navigation actively prevented customers from finding products they intended to buy.
Category confusion. Anna needs baby food. She opens the menu. Should she look in "Food" or "Mom and Baby"? Both seem logical. The correct answer depends on the specific product, but a first-time user has no way to know that. Hick's Law tells us that decision time increases logarithmically with the number of choices. When two categories seem equally valid, users either guess (50% chance of wrong choice) or give up.
Children's toothpaste had the same problem. "Hygiene" or "Mom and Baby"? Hair brushes were expected in "Hair" but actually placed under "Beauty Accessories." These are not edge cases. These are everyday drugstore products that a mother of two buys regularly.
Search engine failure. When navigation confused Anna, she tried search. Typing "mascara" returned zero results. The search engine had no synonym matching, no fuzzy logic, no natural language processing. If the user's search term did not exactly match the product database entry, the search returned nothing. For an e-commerce store with thousands of SKUs, this is a conversion wall.
According to Baymard Institute research, 61% of e-commerce sites fail to return useful results for product-type searches that do not match the site's exact terminology. Our client was in that majority.
Quantified impact. We tested 35 problematic subcategories with 100 users through tree testing. Results were below average across the board. Users showed high backtracking rates, meaning they entered a category, realized the product was not there, went back, and tried another path. Every backtrack is a moment where the user can abandon the task entirely.
To put this in perspective: if Anna's shopping list has 6 items and she encounters a navigation dead-end on 3 of them, she might persevere and find 2 of those 3 through trial and error. But the effort spent navigating drains her patience. By item 5 or 6, the likelihood of abandoning the remaining items increases. This is how a 6-item physical cart becomes a 4-item online cart. It is not that the products are unavailable. It is that finding them costs more effort than the customer is willing to spend.
Finding 2: Missing Feedback and Cart Visibility
In a physical store, you pick up a product and place it in your basket. You can see the basket. You can feel its weight increase. You have constant, ambient confirmation that items have been added.
Online, that feedback must be designed deliberately. This store had almost none.
No visual feedback on Add to Cart. When a user clicked "Add to Cart," nothing visible happened. No animation, no color change, no counter increment, no toast notification. The user had no confirmation that their action registered. This is a documented cognitive bias trigger: when users are uncertain whether an action succeeded, they either repeat it (creating duplicate cart entries) or assume it failed (and move on without the product).
Cart icon ignored. Our heatmap analysis showed that the cart icon in the header received almost no interaction. Users did not scroll up to check it. They did not hover over it. The cart, as implemented, was invisible during the shopping flow. In a physical store, the basket is in your hand. Online, the basket was hidden in a corner of the header, with no dynamic indicators to draw attention.
This finding alone is significant. If users cannot confirm items are in the cart, they do not add as many items. That maps directly to the 4-items-vs-6-items gap.
The heatmap data revealed something else worth noting: users spent considerable time on product pages but rarely proceeded to the cart review page. They were browsing, evaluating, and intending to buy. But the interface gave them no confidence that their actions were registering. In usability terms, the store violated the visibility of system status principle, the most fundamental of Jakob Nielsen's ten heuristics.
Finding 3: Filter and Categorization Gaps
Even when users found the right category, the filtering system failed to help them narrow down options.
Baby food: no taste/flavor filter. Parents buying baby food online need to select specific flavors their child will eat. The store offered no filter for taste or flavor in the baby food subcategory. Users had to scroll through dozens of products, reading each label individually. In a physical store, you scan the shelf and grab the apple flavor. Online, you scroll, read, scroll, read. This is a direct contributor to cognitive load, and it makes the online experience slower and more frustrating than the physical one.
Baby food: no ingredient composition. Product pages for baby food did not display ingredient lists. For a parent choosing food for an infant, ingredients are a primary purchase criterion. Without them, the parent cannot make a buying decision online without already knowing the product from the physical store. This eliminates one of the core advantages of e-commerce: the ability to discover and evaluate new products.
Hybrid nail polish: no color filter. The nail polish category in "Hair Dyes" had no filter by color. Users browsing for a specific shade had to open individual product pages. Compare this to a physical store where colors are immediately visible on the shelf. For a visual product category, the absence of a visual filter is a fundamental UX failure.
Inconsistent category logic. As documented in our tree testing and card sorting study, the store's information architecture reflected internal warehouse logic, not customer mental models. This is common in retailers that build their online store around their inventory management system rather than around how people actually shop.
The pattern was consistent: wherever the physical store let customers use their eyes and hands to quickly assess products (reading labels, comparing colors, scanning shelves), the online store forced them to click through individual product pages one by one. Every category that lacked proper filtering was essentially asking users to open 20 to 50 product pages to find one item. That is not browsing. That is a search operation disguised as a product listing.
Finding 4: Friction in Purchase Flow
The final category of findings covered micro-interactions that added friction to every purchase.
Quantity selection required repeated clicking. There was no quantity input field. To add 10 units of baby food (a normal purchase for a parent stocking up), the user had to click "Buy Now" 10 separate times. Each click required the same effort as the first. No increment buttons, no input field, no "add to cart with quantity" option.
This is a pattern that simply does not exist in physical retail. You grab 10 jars from the shelf in one motion. Online, you click a button 10 times, each time wondering whether the click registered (see Finding 2). The compounding effect of poor quantity UX plus missing add-to-cart feedback is significant.
For e-commerce UX audit practitioners, this is a textbook example of how individual small frictions multiply. One minor annoyance is forgettable. Four of them in a single purchase flow make the user close the tab. Nielsen Norman Group research classifies these as "slips," the most preventable category of user error.
Recommendations and Results
Based on the UX audit e-commerce findings, we delivered a prioritized list of recommendations. The client implemented changes in three phases.
Navigation restructuring. We ran card sorting sessions with 100 users to rebuild the category tree based on how customers think, not how the warehouse is organized. Products appeared in the categories where users expected them. Ambiguous items were cross-listed.
Result: users navigated to products 2x faster with 29% less backtracking.
Persistent cart with visual feedback. We recommended a sticky mini-cart with animation on add, item count badge, and a brief toast confirmation. The cart became visible and responsive throughout the shopping flow.
Result: after implementing cart visibility improvements, the client saw +33% completed sales. Full details in our heatmap-driven redesign case study.
Search engine improvements. We recommended implementing synonym matching, typo tolerance, and basic NLP. "Mascara," "tusz do rzes," and "eye makeup" should all return relevant results. Algolia and Elasticsearch both offer these features as standard configurations. The gap was not technology availability but awareness of the problem.
Filter additions. Flavor filters for food categories, color filters for cosmetics, ingredient information on product pages. These are standard e-commerce features documented in Baymard's product page UX benchmark, yet approximately 20% of large e-commerce sites still lack them.
Quantity input fields. Replace repeated button clicks with a direct quantity input, including increment/decrement buttons and a text field for bulk entry.
Product page enrichment. Add ingredient lists, nutritional information, and usage guidelines to product pages in categories where purchase decisions depend on that data. Baby food, skincare, and supplements all require this information. Without it, the online store is less informative than the product packaging on a physical shelf.
Mobile-specific improvements. The mobile version needed larger tap targets for add-to-cart buttons, a sticky bottom cart bar (always visible without scrolling to the header), and simplified filter panels that do not require precise gestures on small screens. Given that Anna's primary shopping context was evening mobile browsing, mobile was not a secondary consideration. It was the primary one.
Combined Impact
The cumulative effect of these changes was substantial. Navigation restructuring alone cut product discovery time in half. Adding cart feedback increased completed purchases by 33%. Reducing backtracking by 29% meant users spent less time lost in the wrong categories and more time adding products to their cart.
None of these improvements required a full platform rebuild. They were targeted fixes to specific, measured problems. That is the value of a structured UX audit e-commerce methodology: it replaces guessing with evidence, and prioritizes changes by measured impact rather than subjective opinion.
Key Takeaways for E-Commerce UX Audits
1. Measure the right gap. This client knew their online cart was smaller than their physical cart. That single metric (38.99/6 vs 34.99/4) framed the entire audit. Without a concrete baseline, a UX audit e-commerce project becomes a list of opinions.
2. Build personas from data, not assumptions. The client had no personas. We did not guess. We built Anna from transaction data, GA segments, and workshop synthesis. Every finding in this audit traces back to Anna trying to complete a realistic shopping task.
3. Test navigation with real users. Internal teams are blind to their own category logic. Tree testing with 100 external users revealed problems that the client's team had never noticed because they knew where everything was.
4. Micro-frictions multiply. No single finding in this audit was catastrophic in isolation. A confusing category here, a missing filter there, a broken search query. But stacked together across a 6-item shopping list, they reduced the cart by 33%.
5. Physical retail is your UX benchmark. When your online and physical stores sell identical products, the physical store experience is the target. Every place where the online experience is worse than picking items off a shelf is a conversion opportunity.
This UX audit e-commerce approach, starting with data, building personas, simulating real tasks, and measuring specific outcomes, is the methodology we apply across all our UI/UX design projects.
If your e-commerce metrics show a gap between potential and performance, and you want to know exactly where the friction is, get a free validation call. We will tell you whether a full audit engagement makes sense for your case, or whether smaller fixes can close the gap. See our full case study portfolio for more examples of this work.
Want to discuss your store's UX data with our team? Schedule a call.
