Audit run by UXFix on 8 September 2026 against 115 rules. Every finding below is shown with the screen we captured, in the order they cost conversions. Nothing here was sponsored or reviewed by the store.
Grade C, needs work. Most points are lost on mobile and confidence.
Each category is scored 0–100 from the rules that apply to it.
Largest Contentful Paint under 2.5s on mobile. Tap targets at least 44x44 px on all interactive elements. And 3 more.
Returns/exchange policy visible on PDP (not buried in footer). Image zoom available on desktop and mobile. And 3 more.
Delivery expectation shown before final purchase. Contact/help option accessible during checkout (phone, chat).
Total form fields 8 or fewer (industry average is 11.3). Address autocomplete (Google/Loqate) present. And 6 more.
Add-to-cart CTA above the fold on mobile. Delivery/arrival date estimate shown. And 2 more.
Across 28 audited stores, as of 9 September 2026. Method and full ranking: the UXFix ecommerce benchmark.
Scores above 74% of audited stores
The same rule pass UXFix runs on every store. No store data and no analytics access — everything below was observable from the public storefront.
Product page to payment step, one real product, both viewports.
When shipping, tax and totals first become visible to the shopper.
Field count, labels, autofill, error placement, recovery after a failed submit.
Payment marks, security copy, returns and contact reachability at the pay step.
Tap target sizes, thumb reach, keyboard type, viewport zoom on focus.
Whether an AI shopping agent can read price, stock and shipping from the page.
What we could not check: 32 rules could not be judged from the public storefront on the date above. They were dropped from the score, not guessed at.
115 rules were run. 32 could not be assessed and were dropped rather than guessed, leaving 83 scored. A further 3 findings are held back from this page because our wording for them was not fit to publish.
Rules already passing. No action needed on any of these.
The issues, in the order they appear below.
Right idea, wrong placement or timing: full total including shipping and tax visible before payment entry, in-scale/lifestyle image included (product in real context) and no layout shift (CLS) when images load. And 6 more.
Each finding shows the screen as we captured it on 8 September 2026, what the problem is, and what it costs in conversions.
The full order total (including shipping and tax) isn't visible before the buyer starts entering payment details.
Surprise costs at the payment step are the #1 abandonment trigger in every study. Buyers want the final number before their card leaves the wallet.
The Add-to-Cart button is not visible above the fold on mobile. Shoppers have to scroll to find how to buy.
On mobile especially, buyers scan the top of the page and decide within seconds. If the primary purchase button isn't in that first view, many will leave before scrolling.
Since this exact tee is unavailable, replace the 'Sorry, this item is no longer available' text block with a prominent CTA like 'Shop The Box-Cut Tee in White' or 'Shop The Box-Cut Tee in Black' linking directly to the in-stock variants shown below, so shoppers can purchase without scrolling to the 'More to Love' section.
The product page doesn't tell shoppers when their order will arrive. There is no delivery date or lead-time estimate.
Buyers often decide between two stores based on delivery timing, especially for gifts and urgent purchases. Uncertain timing pushes them to competitors who are clear.
There's no returns or exchange policy communicated on the product page itself. Shoppers have to hunt the footer or FAQ to find it.
"What if I don't like it?" is the last question in every purchase. If the answer isn't obvious, hesitation kills conversion, especially for higher-price items and apparel.
Add a short 'Free Returns & Exchanges within 30 days' line directly under the price/rating block (around y=0.17) near the 'Sorry, this item is no longer available' notice, so shoppers see the policy even when the item is unavailable.
The main content of the product page takes longer than 2.5 seconds to appear ("Largest Contentful Paint" is slow).
Every extra second of load time drops conversion by ~7% (Google, 2024). 2.5s is Google's threshold for a good experience. Above 4s is considered poor and hurts search rankings too.
We judged this on the live page but could not confirm this picture is of the same step, so we are not showing it.
The cart has no shipping estimator; shoppers can't preview shipping costs without starting checkout.
Unknown shipping costs are the single biggest cause of cart abandonment. An estimator answers the question before it becomes a reason to leave.
The checkout has more than 8 form fields. Industry-best checkouts collect 6-8 fields; industry-average is 11.3.
Every extra field drops conversion by roughly 2%. Fields cost you money in a direct, measurable way. A checkout with 13 fields loses ~10% of buyers compared to one with 8.
The checkout has no address autocomplete (Google Places, Loqate, SmartyStreets, or similar). Shoppers type every line of their address manually.
Address autocomplete cuts address-entry time roughly in half on mobile and drops delivery-failure rates from typos. It also feels premium, buyers notice.
The checkout page does not show when the order will arrive. Buyers commit to buying without knowing the delivery date.
Delivery timing is often the deciding factor between two stores. If your competitor tells the buyer "arrives Tuesday" and you say nothing, you lose. Especially for gifts.
The structured product data that AI shopping agents read (ChatGPT, Google, Copilot) is incomplete on this page. Some or all of these are missing: product ID (GTIN), brand, review data, availability, shipping details.
AI-referred traffic converts 42% higher than regular traffic (Adobe, 2026). But agents only recommend stores whose product data they can read. Incomplete data means invisible to AI shoppers.
The checkout has no signals declaring it can be driven by AI agents. No ACP or UCP meta tags, no /.well-known/agentic-commerce endpoint, no Stripe ACP integration.
OpenAI's Agentic Commerce Protocol (ACP) and Google's Universal Commerce Protocol (UCP) are the emerging standards for agent-driven checkout. Stores that declare readiness get preferential routing from ChatGPT and Google AI. This is 2026's biggest new distribution channel.
On this long product page, the Add-to-Cart button disappears once the shopper scrolls down. There is no sticky bar that keeps it reachable.
Shoppers scroll deep into reviews and details, decide to buy, and then have to scroll all the way back up. Every extra scroll is a chance to get distracted and leave.
Clicking the product image doesn't open a zoom view. Shoppers can't inspect details, fabric, or quality up close.
Zoom is how online shoppers replace touching the product. Baymard finds inadequate image inspection among the top reasons for product-page abandonment.
The product page does not have a Q&A section or a real specifications table. Shoppers cannot answer their own questions before buying.
Considered purchases (electronics, appliances, tools, tech gear) live and die on the details page. When shoppers cannot find dimensions, materials, compatibility, or answers to obvious questions, they leave to research on Amazon or a competitor and often buy there instead.
The product description is one big block of text with no headings, bullet points, or sections. Shoppers can't scan it.
Nobody reads walls of text online. If the description isn't scannable, the buyer misses the key features and either bounces or buys the wrong thing and returns it.
Add a structured description block below the 'Sorry, this item is no longer available' message (or above it) with sub-sections like 'Fabric & Care', 'Fit', and 'Details', using bullet points instead of leaving the area blank so users still learn about the tee even when out of stock.
Some interactive elements on the mobile page are smaller than 44x44 pixels. Fingers struggle to hit them accurately.
44x44 pixels is the minimum touch target Apple and the WCAG 2.2 guidelines recommend. Below that, mis-taps happen: buyers hit the wrong swatch, wrong quantity, wrong link, and abandon in frustration.
We judged this on the live page but could not confirm this picture is of the same step, so we are not showing it.
The cart does not show a delivery estimate for each item. Buyers cannot see when their order will arrive.
Delivery timing is the second-biggest factor after price, especially for gifts and urgent purchases. Uncertain arrival dates push buyers to competitors who are clear.
We judged this on the live page but could not confirm this picture is of the same step, so we are not showing it.
The cart only offers a "Remove" action. There is no "Save for later" or "Move to wishlist" option to hold the item without buying today.
A shopper who is not ready to buy today but likes an item will either remove it (and forget) or leave the tab open (and forget). Either way you lose the sale. A save-for-later button captures that intent.
The checkout doesn't show which step the buyer is on or how many steps remain. It feels like an infinite form.
Not knowing how long a task will take is one of the strongest predictors of abandonment. A visible "Step 2 of 3" gives buyers the confidence to push through the next 30 seconds of typing.
Add a horizontal step tracker above 'Contact' (e.g. 'Information > Shipping > Payment') with the current step highlighted in the purple used for the SHOP button, so shoppers know they're on step 1 of 3 before reaching Shipping method and Payment sections.
There's no way to edit the cart from inside the checkout; fixing a quantity means abandoning the checkout flow.
Buyers often spot a mistake at the order summary. If fixing it costs their progress, some just quit.
The store logo in the checkout header links straight back to the homepage.
A stray click on the logo dumps the buyer out of checkout, and on some stores it also loses the form data. Small leak, easy fix.
The checkout uses separate First Name and Last Name fields instead of a single Full Name field.
Two fields cost two taps and two keyboard focuses on mobile. Baymard research shows a single Full Name field measurably lifts checkout completion, especially on phones.
Personal-info fields are missing correct `autocomplete` attributes, or one is set to `autocomplete="off"`. Password managers and browser autofill do not work smoothly.
WCAG 2.2 requires autocomplete on personal-info fields for accessibility. Beyond compliance, correct autocomplete makes checkout a 2-tap experience for returning buyers, huge for mobile conversion.
The checkout does not offer a "Same as shipping" checkbox for the billing address, with the box checked by default. Buyers have to type their address twice. (This rule only applies to physical-product stores.)
For most orders, billing and shipping addresses are the same. Making the buyer re-enter their address is duplicate work and a common source of typos.
The phone number field is required with no explanation why. Buyers who value their privacy hesitate or abandon.
"Why does a t-shirt store need my phone number?" is a real objection. Without a rationale, some shoppers will not give it and will leave.
The checkout form uses placeholder-only labels: the field's purpose is shown as grey text inside the input, and disappears the moment the buyer starts typing.
Placeholder-only labels violate WCAG accessibility, but they also just cost sales. A buyer who types a wrong value, then forgets what the field was for, has to erase everything to see the label again. Frustration compounds.
Add persistent labels above each input (e.g. 'Email', 'First name', 'Address', 'Postal code') similar to the 'Discount code or gift card' field style, so users can still see what each field is for after they start typing, especially in the Delivery and Payment sections.
The card-payment form is missing the `autocomplete` attributes mobile browsers need to autofill saved cards.
iOS and Android can autofill a saved card in one tap when the fields are labeled correctly. Without it, the buyer types 16 digits, expiry, and CVV manually, and many abandon.
The checkout does not declare support for delegated-payment (ACP tokenization). Agents cannot pass a tokenized payment method on the buyer's behalf.
Delegated-payment is the piece that actually lets agents complete a purchase without asking the buyer to type their card. Stripe and PayPal both support this via their ACP modes. Without it, agents can prep the order but hand it back to the buyer to finalize.
When an error message appears, the form visibly jumps, moving fields under the buyer's cursor or thumb.
Layout jumps cause mis-clicks and lose the buyer's place in the form. On mobile they routinely cause wrong-field input.
UXFix scores every store twice: once for human shoppers, once for the agents now doing the browsing on their behalf. Everlane scores 62 for humans and 54 for agents. The gap is the part most stores have not looked at yet. 5 of the 12 agent rules are not fully met.
Derived from this store's own captured product-page content. Coverage judges whether that content answers each query — it is not a measurement of search volume or of what any AI currently recommends.
Everlane (everlane.com) scored 62/100 on 8 September 2026 in UXFix's independent audit of its product page, cart and checkout, judged on 83 of the 115 rules run.
The audit found 29 failing checks. The costliest: full total including shipping and tax visible before payment entry, add-to-cart CTA above the fold on mobile and delivery/arrival date estimate shown.
Everlane scores 54/100 on UXFix's AI Shoppers rules, which test whether an AI agent can read price, stock and shipping from the store and complete a purchase there.
UXFix ran the same 115 deterministic rules it runs on every store: a full purchase walk on desktop and mobile, from product page to payment step, using only the public storefront. No store data or analytics access was used, and every finding is shown with the captured screen.
Against 28 scored stores in the UXFix benchmark, Everlane ranks 8 with 62/100. The benchmark median is 58. As of 9 September 2026.
Everlane is not a badly built store. It passes 45 of 83 assessable rules. What it has is a band of friction between the product page and the payment step, and every part of it is named above.
The 29 findings above come to roughly 84 developer hours of estimated effort in total. Most are settings or template changes rather than rebuilds. A further 9 rules are only partly met and worth a look after these.
The pattern repeats across the stores we audit: the score is rarely lost on visual design, and it is rarely lost in one place. It is lost on when the cost appears, how many decisions stand between a shopper and paying, and whether a machine can read any of it.
Modelled, not measured. We re-audit for free to check.
The same rules, the same evidence, the same estimates, on your product page, cart and checkout. Free, no card, about two minutes.
Quote the score with its date, or embed the live badge. The badge links back to this page and stays current if we re-audit.
Recent audits from the same benchmark corpus.
