Most of the usability problems we find on Shopify stores sit in four places: the search box, the variant picker, the mobile filter drawer, and the first screen of checkout. Fix those and you've done more for conversion than a theme redesign will. The rest of e-commerce UX design is housekeeping, useful housekeeping, but it won't move the number on its own. What follows is the list we work through on an audit, in roughly the order we work through it.
Search is the highest-intent surface on your store
A shopper who types into the search box has told you exactly what they want. Shopify's native search is keyword matching with some leniency, and on a catalogue above a few hundred SKUs it starts failing in boring ways. "Kurta set" doesn't find "Kurta Set with Dupatta" if the handle and tags drifted. Someone types a SKU from a WhatsApp catalogue and gets nothing.
1. Make search actually find things. Add synonyms for the words your customers use rather than the words your merchandiser used. Transliterated terms matter in India: "saree" and "sari", "chappal" and "sandal", "anarkali" spelled four ways. Make SKUs searchable. Allow one or two character typos.
2. Kill the dead end. A zero-results page with nothing on it is the clearest abandon signal we see in recordings. It should show the closest matches, your top collections, and the search term in an editable field so the shopper can correct a typo without retyping.
Here's arithmetic you can run on your own Shopify analytics this afternoon. Say you get 60,000 sessions a month and 9% of them use search, so 5,400 search sessions. If your search log shows an 18% no-results rate, that's about 970 searches a month that end in a wall. Those sessions had intent. Even recovering a quarter of them at your site conversion rate is real money, and the fix is a config change, not a rebuild.
3. The mobile filter drawer. Roughly four in five sessions on the Indian D2C stores we work on are mobile, so the filter experience is the drawer, not the sidebar. Three things break it: no result counts next to each option, filters that apply on every tap and re-render the grid under the shopper's thumb, and a drawer that closes and dumps you back at the top of the collection instead of where you were. Show counts, batch the apply, preserve scroll position. If your catalogue is large enough that merchandising it by hand has stopped being realistic, a dedicated filter and search layer like FilterPro does this out of the box, including the metafield-driven filters Shopify's native faceting won't give you.
The product page decisions people get wrong
4. Variant pickers. Dropdowns for colour are a waste of a good product photo. Swatches, with the image swapping on selection. The bigger issue is out-of-stock handling: most themes let you select a sold-out size and only tell you after you've tapped Add to Cart. Grey it out with a strikethrough and put a notify-me there instead. On a two-option product (size and colour) the picker should also disable combinations that don't exist, which sounds obvious and is wrong on most themes we open.
5. Put the answer to the buying question above the fold. For apparel that's fit and size, in centimetres and in garment measurements, not just a chart linked in a modal. For skincare it's the ingredient list and the FSSAI or CDSCO detail if you're making claims. For electronics it's the spec that drives the decision: battery, warranty, what's in the box. Shoppers scroll past the carousel looking for one fact. Make them scroll less.
6. Show the delivery date, not a promise. "Fast shipping" means nothing. "Delivered by Tue, 14 Oct to 560068" means something, and pincode-based delivery estimates on the product page cut the number of people who get to checkout and bail when the real date appears. Same logic on price: show the GST-inclusive figure, show the COD fee before checkout if you charge one, and do not surprise anyone with a shipping charge at step three. Surprise cost is the most common abandon reason we can actually see in funnel data.
Collection pages carry more traffic than you think
7. Give the product card enough to decide on. Image, name, price, and nothing else is a card from 2014. Add the rating if you have reviews, the colour swatches if the product has them, and a size-availability line if you sell apparel. Resist the quick-add button on mobile unless the product has one variant; it opens a modal that half of shoppers close by tapping outside, which looks like engagement and converts like nothing.
Two columns on mobile, not one. One-column grids feel premium in a design review and make people scroll through nine products to see nine products. Default sort is worth an argument with your merchandiser: best-selling is usually right, newest is right for a fashion drop, and "featured" is right only if someone is actually maintaining the manual order.
Speed is part of UX design, not a separate project
8. Set a weight budget and defend it. A 2.4-second LCP on 4G is a usability problem dressed up as a technical one. The usual culprits on an Indian D2C store are a 1.5 MB hero image that was never resized, four review and upsell apps injecting scripts into the head, and a slider on the homepage nobody swipes past slide two of.
Compress the hero to under 150 KB in WebP, lazy-load everything below the fold, defer third-party scripts, and audit your app list once a quarter. Uninstalling an app does not always remove its theme code; we find orphaned script tags in theme.liquid on most audits. If you want the full method and what it typically costs, our Shopify speed optimization page goes through it. During festive peaks the gap matters more, because that's when everyone is on a congested network and your ad traffic is at its most expensive.
Checkout and the address form
You can't redesign Shopify checkout below Plus, and on Plus you can do less than you expect now that checkout.liquid is gone. What you can control is everything leading into it and the fields inside it.
9. Fix the address form. Ask for pincode early and auto-fill city and state from it. Indian shoppers type their address in an order nothing like the Western form, and they abandon when the form fights them. Keep the phone field numeric-only with a +91 prefix already set. Don't ask for a company name on a D2C store.
10. Frame payment choice honestly. If COD is 55% of your orders, it's not a problem to design around, it's your business. Put prepaid discount messaging next to the COD option where the decision happens, not in a banner at the top of the page. A ₹50 prepaid discount on a ₹1,200 order is 4%, and if your COD RTO rate is 22% against 4% on prepaid, the arithmetic for giving that discount is easy: every COD order that converts to prepaid saves you a forward and reverse leg of shipping plus the handling. Run your own numbers before copying anyone's discount value. Razorpay's UPI intent flow on mobile should be one tap, and if your checkout is showing a QR code on a phone, something is misconfigured.
11. Write error messages like a person. "Invalid input" tells nobody anything. Validate inline as the field loses focus, say what's wrong and how to fix it, and never clear a form on error. The OTP login step deserves a look too: auto-read the code where the browser supports it, give a resend timer that doesn't start at 60 seconds, and let people check out as a guest. Forced account creation costs more orders than the email list is worth.
Accessibility fixes that double as conversion fixes
12. Tap targets, contrast, focus states. A 44 by 44 pixel minimum on anything tappable. Grey-on-grey price text at 3:1 contrast fails WCAG and also fails the forty-year-old reading it on a train. Visible focus outlines so keyboard users can see where they are. None of this is charity. Thumb-sized buttons reduce misclicks, and a misclick on mobile is an exit.
While you're in there: alt text on product images helps screen readers and helps Google Image search, which still sends meaningful traffic to apparel and jewellery stores.
How to find your own problems in an afternoon
Install a session recording tool and watch twenty recordings of sessions that reached cart and didn't convert. Not a hundred. Twenty is enough to see a pattern, and you will see one in the first five. Then open your search log and read the top fifty queries and every query that returned nothing.
Then do the thing most founders skip: hand your phone to three people who aren't in e-commerce and ask them to buy a specific product. Watch without helping. It is uncomfortable and it is the cheapest research available. We've had sessions where the tester couldn't find the size chart for ninety seconds on a page the team had looked at every day for a year.
What we'd tell you to skip
Mega menus with eight columns. Mobile shoppers don't use them and they push your actual categories three taps deep. Homepage personalisation before you have the traffic to make it statistically meaningful, which for most stores means before roughly 100,000 sessions a month. Chat widgets that auto-open after four seconds and cover the add-to-cart button. Pre-loader animations. Sticky bars stacked two deep that eat a third of a small screen.
And exit-intent popups on mobile, where exit intent doesn't exist, so the trigger is a timer and it fires while someone is reading your ingredient list.
Pick the two items from this list that match what your recordings show, ship them, and measure for a fortnight before touching anything else. If you'd rather have someone else do the looking, our free store audit covers the search log, the mobile funnel and the speed numbers, and if the answer is "your filters are fine, your product photography isn't", we'll say that. Teams who want the work done rather than diagnosed can see how we scope and price it on the Shopify development cost page.