If you're hiring for 2026, the UI/UX design skills worth paying for are narrower than most portfolios suggest: designing to a performance budget, laying out for one thumb on a mid-range Android, making a 900-SKU catalogue navigable, writing the twelve words next to the COD toggle, working inside Shopify's theme architecture instead of against it, meeting accessibility rules that now carry legal weight in the EU, and reading enough data to know when an A/B test is noise. A designer who can do those seven things will outperform one with a prettier Dribbble page by a wide margin.
We've rebuilt enough storefronts to see the same pattern. The design is handed over as a Figma file that looks excellent at 1440px, and six weeks later the store is slower, the mobile PDP has a 3.1s LCP, and nobody can find the size chart. The design wasn't bad. It was made without the constraints that actually govern an ecommerce storefront.
Performance budgeting: the UI/UX design skill most teams outsource too late
Speed is settled at design time, not during a dev cleanup sprint. A hero video, a four-font stack, a product card with hover-swap images, an announcement bar that loads a third-party script: every one of those is a design decision with a measurable cost in milliseconds. By the time a developer sees the file, the budget is already spent.
Google's thresholds haven't moved much and they're the floor, not the target: LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1, measured at the 75th percentile of real users. INP is the one that catches designers out. It punishes heavy JavaScript on interaction, so the slick filter drawer that re-renders the whole collection grid on every checkbox tick will tank it even though the page loads fast.
What good looks like in practice: the designer specifies a weight ceiling per template before drawing anything. Something like 150KB of JavaScript and one web font family with two weights on a PDP. Then they design to it. When the founder asks for a testimonial carousel, the answer is which existing element it replaces, not "sure." That one habit does more for mobile conversion in India than any amount of visual polish, because a large share of sessions on the stores we work on arrive on mid-tier Android phones over patchy 4G.
If you've already got a theme and the damage is done, start with measurement rather than a redesign. Our Shopify speed optimisation work usually finds that two or three apps and one oversized hero image account for most of the gap.
Thumb-zone layout, because checkout happens on a bus
Desktop-first design is still the default in most agency files, and it still shows. On mobile, the usable zone for a one-handed grip on a 6.5-inch phone is roughly the bottom two-thirds of the screen, slightly favouring the right side. That's where the add-to-cart, the variant picker, the quantity stepper and the checkout button belong. Not pinned to the top because that's where the wireframe put them.
Specific things we check on every mobile PDP:
- Tap targets at 44px minimum, with real gaps between them. Variant swatches at 32px with 4px spacing produce mis-taps and returns.
- A sticky add-to-cart that appears after the fold, not one that covers the price from the first scroll.
- Address forms that respect the Indian flow: PIN code first, auto-fill city and state, phone number field with a numeric keypad, no "State" dropdown with 36 entries the user has to scroll.
- One-handed reach to the COD toggle, because for a lot of catalogues COD is still the majority payment method and burying it costs orders.
Shopify's checkout itself is mostly locked down, which is a good constraint. A designer who proposes a fully custom checkout on anything below Plus is telling you they haven't shipped on Shopify recently. On Plus, checkout extensibility gives you UI extensions and branding controls, and that is the honest limit: you can add a trust block, a delivery-date promise, an upsell. You cannot rebuild the form. Knowing which fights are available is itself a skill.
Making a large catalogue navigable
Past about 200 SKUs, the homepage stops mattering and the collection page becomes the store. Past 800, on-site search becomes the primary navigation and most of your merchandising effort should go there.
The design work is unglamorous. Facet order, because "Price" first is wrong for apparel and right for electronics. Whether filters live in a left rail or a drawer on mobile, and whether the result count updates before or after the user hits Apply. Zero-result handling, which almost every theme does badly: a blank page with "No products found" instead of the three nearest matches. Synonym mapping for how Indian shoppers actually type, so "kurti" returns kurtas and "saree" returns sarees regardless of spelling.
Most native theme filters fall over at scale, either on speed or on relevance. If that's where you are, a purpose-built layer is cheaper than a custom build; we publish FilterPro for exactly this, adding smart filters and AI search to the storefront without the collection page turning into a JavaScript bonfire.
Writing, which is a design skill whether anyone admits it
The highest-leverage text on an Indian D2C store is about eighty words long and sits in four places: the delivery promise on the PDP, the returns line near add-to-cart, the COD fee or restriction, and the error message when a payment fails on Razorpay. Get those wrong and you generate support tickets instead of orders.
"Free shipping on prepaid orders above ₹999" beats "Free shipping available" because it tells the shopper what to do next. "Delivered by Thursday, 14 Nov to 560068" beats "Fast delivery." A returns line that says "7-day return, pickup from your address, refund to source in 3-5 working days" removes the exact objection that stalls a first-time buyer on a ₹3,000 item. Designers who think copy is someone else's job will hand you a lorem-ipsum PDP and a 2% drop in add-to-cart rate they'll never attribute to themselves.
Designing inside Shopify instead of around it
Online Store 2.0 changed what a handoff should contain. Sections, blocks, section groups, metafields, metaobjects: a designer who understands these produces a file the merchandising team can actually operate. One who doesn't produces a beautiful set of flat screens that have to be rebuilt as rigid templates, and then every seasonal change becomes a developer ticket.
Concretely, the handoff we want includes which parts of a page are merchant-editable, what the empty state of each block looks like when the merchant leaves a field blank, and what happens when the product title runs to three lines. That last one trips up roughly a third of files we review on the first pass. Indian catalogues have long product names with specs baked in, and a card designed around "Linen Shirt" breaks on "Pure Linen Half Sleeve Shirt for Men - Olive - XL."
This is also where cost control lives. Every bespoke section that could have been a theme section is billable now and billable again at every change. We've written about where the money actually goes in Shopify development pricing in India, and design decisions drive more of that number than most founders expect.
Accessibility is now a compliance line, not a nice-to-have
The European Accessibility Act has applied since June 2025, which means any brand selling into the EU has a legal exposure it didn't have in 2023. The UK and US have their own routes to the same place. If you ship to those markets, WCAG 2.2 AA is the working standard and your designer needs to hit it without being asked.
The practical checklist is short: 4.5:1 contrast on body text, visible focus states that survive the theme's CSS, form labels that aren't placeholder text, alt text that describes the product rather than repeating the SKU, and no information carried by colour alone. Low-contrast grey on white is the single most common failure we find, and it's usually in the price. Designers love #999. Customers over 45 cannot read it on a phone in daylight.
Enough statistics to know when a test is noise
This is where most design teams oversell. Say your baseline conversion rate is 1.8% and you want to detect a lift to 2.1%, a 0.3 percentage point absolute change. Using the rough rule that you need about 16 × p(1−p) ÷ δ² visitors per variant at 95% confidence and 80% power: p(1−p) is 0.018 × 0.982 = 0.01768, δ² is 0.003² = 0.000009. That gives 16 × 0.01768 ÷ 0.000009 ≈ 31,400 sessions per variant. Roughly 63,000 sessions total for one test.
If your store does 25,000 sessions a month, that test runs for two and a half months, which is longer than the seasonality it's sitting inside. So don't run it. Make the change on judgement, watch the funnel, and spend your testing budget on things with bigger expected effects: the PDP layout, the payment method order, whether COD is a toggle or a radio. A designer who tells you they'll "A/B test everything" on a small store is either going to waste months or quietly call a 40-conversion difference a win.
Session recordings and funnel drop-off are the better tool at that scale, and reading them well is a skill. Watch twenty checkout abandonments and you'll usually find one specific thing, often a form field, that nobody designed for.
The UI/UX design skills that matter less than people think
Homepage carousels. Nobody scrolls to slide three, and the layout shift hurts CLS. One hero, one message.
Dark mode on a storefront. It's work, it breaks product photography, and we have never seen it move a number.
Elaborate scroll-triggered animation. On a flagship phone it's lovely. On a ₹14,000 Android during a festive sale it drops frames and raises INP.
Full redesigns. The instinct to rebuild everything after a flat quarter is expensive and usually misdiagnosed. Three targeted fixes to the PDP, collection filters and the mobile cart will typically beat a six-week redesign, and you'll know which one worked.
How to test for these skills when you're hiring
Give the candidate your actual PDP URL and ask three questions. What would you remove, and what does it cost us in milliseconds. How would this page look if the product title were twice as long and the brand had no lifestyle photography. Which element here do you think the merchandising team changes monthly, and can they do it without a developer.
Good answers are specific and slightly impatient. Bad answers are about brand storytelling. If you'd rather not run that process yourself, we do design and front-end as one team, and you can also hire a Shopify developer on a dedicated basis if the design side is already covered.
Pick one template, probably your top-selling PDP, and run it through PageSpeed on mobile today. Note the LCP element. Nine times out of ten it's an image a designer chose, at a size nobody questioned. That's your first fix, and it costs nothing to find. If you want a second pair of eyes on the whole storefront, our free audit covers speed, UX and the checkout path in one pass.