Search That Works
Including misspellings, synonyms and how customers actually name things.
Ecommerce UX Design
Ecommerce UX Design Because Most Lost Sales Are Discovery Failures
Checkout gets most of the attention and accounts for a minority of lost sales. Far more people leave because they could not find the right product, could not narrow a large category, or searched and got nothing. Ecommerce UX design starts with discovery — search, filtering and navigation — because that is where the volume is.
We Start Where the Volume Is
Checkout optimisation improves a small share of sessions. Discovery affects nearly all of them.
Filters That Match Intent
The attributes people choose by, not the ones in the product database.
Zero Results Handled
A dead end is a lost sale and it is entirely designable.
Category Navigation
Large catalogues narrowed without twelve clicks.
Cross-Session
Most purchases involve several visits, and most sites forget everything between them.
Where an online store loses purchases
Retail UX is unusually measurable, because every step between arrival and payment can be observed. These are the checks that locate where intent is being lost, rather than describing the experience in general terms.
Category to product rateWhether listing pages give enough information to justify a click
Search result qualityWhat happens when a shopper searches for a product you actually sell
Zero-result handlingWhat a shopper sees when their search finds nothing
Filter usefulnessWhether filters match how shoppers describe the product, not how you catalogue it
Product page completenessWhether the page answers the questions that otherwise become support tickets
Image adequacyScale, detail and variant coverage sufficient to decide without handling the item
Cart abandonment pointsThe specific step where carts stop progressing
Unexpected cost exposureWhen shipping and tax become visible relative to when the decision is made
Guest checkoutWhether a purchase can be completed without creating an account
Form frictionField count, validation behaviour and mobile input types at checkout
Mobile completionWhether the full purchase path works one-handed on a phone
Returns clarityWhether the returns policy is findable before purchase rather than after
Unexpected cost exposure is consistently among the largest recoverable losses in online retail. A shopper who has invested effort and then encounters a shipping charge they did not anticipate abandons at the point of highest frustration. Showing the total earlier costs some entries to the funnel and recovers more at the end.
Ecommerce UX, Explained
Where Are Sales Actually Lost?
Four points, and only the last is checkout.
Discuss Your Store →- 1
Search Returns Nothing
The customer used a different word from your product data.
- 2
The Category Is Too Big
Four hundred items and no useful way to narrow them.
- 3
The Product Page Falls Short
A question went unanswered — see product page design.
- 4
Checkout
Real, and smaller than the three above combined.
What Ecommerce UX Work Covers
Discovery first, then the path from finding to buying.
Being found in the first place is covered by ecommerce SEO. Product page decisions are ecommerce product page design; the final stretch is checkout page design.
How We Improve Ecommerce UX
Start with the search log. It is the most direct record of what customers wanted and did not get.
Read the Search Log
What people typed and what came back.
Map Discovery Paths
How people get from arrival to a product.
Fix Search and Filters
Synonyms, spelling and the attributes that matter.
Design the Dead Ends
Zero results and empty filters given somewhere to go.
Test With Shoppers
Real purchase attempts, watched.
What the catalogue and the purchase decide
The design problem in retail is set by how many products there are and how much consideration each purchase requires. A single-product brand and a ten-thousand-SKU catalogue share almost none of the same challenges.
Large catalogues
Where the central problem is helping a shopper find the right item among thousands. Search quality, filtering and category structure do nearly all the work, and product page design matters much less than getting shoppers to the correct page at all.
Small ranges and single-product brands
Where discovery is trivial and persuasion is everything. The product page carries the entire sale, which means imagery, detail, objection handling and social proof matter far more than navigation does.
Considered and high-value purchases
Where shoppers research across multiple sessions and devices before buying. Saved carts, comparison, detailed specification and clear returns terms address the actual hesitation, and rushing the decision loses more sales than it closes.
Repeat and consumable purchases
Where the same customer buys the same thing repeatedly and the goal is reducing effort to near zero. Order history, quick reorder and saved preferences matter more than any acquisition-stage improvement.
Configurable and made-to-order products
Where the shopper builds the item and every option affects price and availability. This is closer to an application than a catalogue, and it usually needs the user flow mapped explicitly before any interface is drawn.
Discovery
Why the Search Log Is the Best Evidence You Have
It is a list of things customers wanted, in their own words, with a record of whether they got them.
What does the zero-results list tell you?
Three different things, and they need different responses. Products you do not stock — useful demand data. Products you do stock under a different name — a synonym problem. And misspellings, which are trivially fixable.
Most stores never look at it, so all three go unaddressed indefinitely while the same searches fail every week.
It is the highest-return data in ecommerce: specific, already collected, and directly actionable without any research being commissioned.
Why do filters so often not help?
Because they are generated from the product database rather than from how people choose. A clothing filter offering fourteen material types and no "suitable for" grouping reflects the data model, not the shopper.
Filter order matters too. The attribute most people narrow by should be first and open; the specialist ones can be collapsed below.
The test is whether someone can get from four hundred items to a shortlist in two interactions. Most large-category filter panels fail it while offering more options than anyone uses.
Checkout is where the money is lost, and rarely where it is designed
Checkout receives less design attention than any other part of a store, because it is the least interesting page to work on and it usually arrives as a platform default. It is also the point at which every shopper who was going to buy has already declared their intent, which makes each failure here more expensive than a failure anywhere else.
The recurring causes are consistent across stores. A forced account creation that presents a registration form to someone who wanted to buy something. Shipping costs revealed at the final step, after the shopper has committed effort. Forms with more fields than the transaction requires, validating aggressively, rejecting formats that should be accepted. Mobile inputs that summon the wrong keyboard for a card number.
None of these is difficult to fix and all of them are easy to leave, because they are invisible to anyone who has not tried to complete a purchase on a real phone with a real card. The single most useful exercise available to a retail team is to buy their own product on a mobile connection and note every point of irritation.
Product pages have to answer the questions that would otherwise be asked
A shopper cannot handle the item, so the page has to substitute for that entirely. Every question left unanswered becomes either a support enquiry or an abandonment, and abandonment is by far the more common outcome because asking takes effort the shopper is not obliged to spend.
The questions are usually predictable and specific to the category: actual dimensions, how it fits relative to a known reference, what is included, what it looks like in ordinary lighting, how long delivery takes, and what happens if it is wrong. The support inbox contains a complete list of them, sorted by frequency, and is the fastest route to an accurate product page specification.
Returns policy belongs on the product page rather than in a footer link, because it is directly part of the purchase decision for anything where fit or suitability is uncertain. Making it hard to find does not reduce returns; it reduces purchases by shoppers who were not willing to accept an unknown risk.
Search and filtering carry more weight than navigation
Shoppers who use on-site search are substantially further along in their intent than those browsing categories, because they have arrived with a specific item in mind. This makes search quality one of the highest-leverage areas in a store — and it is frequently left as an untuned default that fails on synonyms, plurals, misspellings and the terms customers actually use.
The diagnostic is straightforward: read the site search logs. They show what shoppers wanted in their own words, which queries returned nothing, and where the vocabulary of the catalogue diverges from the vocabulary of the customer. Queries returning zero results for products the store genuinely stocks are direct, recoverable lost revenue.
Filtering has the same character. Filters built from how the business catalogues products — supplier, internal category, product code — are useless to shoppers who think in terms of size, use, price and compatibility. Rebuilding filters around customer vocabulary rather than internal taxonomy is often the single largest improvement available to a large-catalogue store.
Questions, answered.
Ecommerce UX design covers how shoppers find and choose products — search, filtering, category navigation and the path to purchase — as well as the basket and checkout.
Usually not. Checkout affects the minority of sessions that reached it. Discovery — search, filters, navigation — affects nearly all of them, and that is where the larger loss sits.
Start with the zero-results log. It separates into products you do not stock, products named differently in your data, and misspellings — three problems with three straightforward responses.
Enough to reach a shortlist in about two interactions, ordered by how people actually narrow. Generating one filter per database attribute produces a long panel that helps nobody.
Alternatives, a corrected spelling suggestion, and popular items from the closest category. A dead end is a lost sale and it is entirely designable.
Unexpected costs at checkout, followed closely by forced account creation. Both take a shopper who has decided to buy and introduce a new obstacle at the moment of highest commitment. Both are usually platform defaults rather than deliberate choices, which is why they persist — nobody decided to do it, so nobody has revisited it.
Usually not as a first step. Store performance problems tend to concentrate in a small number of places — search, the product page, or one checkout step — and those can be identified from analytics before anything is redesigned. A full rebuild changes everything simultaneously, which makes it impossible to tell what helped and what hurt, and risks losing whatever was working.
For most retail categories it is where the majority of sessions occur, and frequently where completion rates are worst. The gap between mobile traffic share and mobile conversion share is the clearest signal available that the mobile experience is the constraint. Testing the full purchase path on an actual phone, on a mobile connection, is the check that matters — your own analytics will show the specific split for your store.
They affect confidence, particularly for unfamiliar brands and for products where suitability is uncertain, and how they are presented matters as much as whether they exist. Reviews that are obviously filtered read as untrustworthy and can perform worse than none. Displaying them honestly, including the critical ones, is more persuasive — a product with mixed but genuine feedback is more credible than one with uniformly perfect scores.
The platform sets what is straightforward and what requires development, so it constrains the work rather than determining it. Most of the improvements that matter — checkout steps, product page content, search configuration, filter taxonomy — are achievable within any established platform. Where a limitation is genuinely structural it is worth naming early, because that becomes a development decision rather than a design one.
Still deciding if ecommerce ux design is right for you?
Talk to UsThey Told You What They Wanted
Ecommerce research usually means commissioning something: user testing, surveys, a heatmap tool, a consultant with a heuristic checklist.
Meanwhile the site search log contains thousands of statements of intent, typed by real customers in their own words, each one paired with a record of whether the site could answer it.
The zero-results list in particular is a queue of specific, fixable failures — and on most stores nobody has opened it since the search was installed.
Free Store Review
Find Out What Shoppers Cannot Find
Send us your store and access to your search log. We will tell you which searches fail and where discovery breaks down.
Claim Your Free Marketing Audit
Enhance Your Brand Potential At No Cost!
- Expect a response within 24 hours
- NDA available upon request
- Dedicated product specialists