Convert Lane

Chapter 10 of 11

Shopify UX, Design and Development

What design and front-end work really do to conversion on Shopify, when a redesign is justified, and the speed and app-bloat problems that cost orders.

Contents
  1. What design and front-end work actually do to conversion
  2. When a redesign is justified
  3. The cost nobody prices: you lose your accumulated learning
  4. Brand expression against usability
  5. Speed is a conversion variable, and Core Web Vitals is the scoreboard
  6. Where Shopify themes actually lose their speed
  7. App bloat is a tax you pay twice
  8. Accessibility is a conversion issue, not a compliance chore
  9. Build sections so future tests are cheap to ship
  10. What Shopify Plus actually unlocks, and what it does not
  11. What loses
  12. The order to do this in

Most Shopify design work is sold as a rebrand and bought as a conversion project. Those are different things, and confusing them is how stores spend six figures to end up flat.

This chapter covers what design and front-end development actually do to the number, when a rebuild is the right call, and the technical failures that cost Shopify stores orders without appearing on any dashboard.

What design and front-end work actually do to conversion

Design does not persuade anyone to buy something they did not want. It removes the reasons they had to leave.

There are three mechanisms, and everything else is decoration:

It reduces effort. Fewer decisions, fewer taps, fewer moments where the customer has to work out what to do next. A variant selector that takes one tap instead of three is not prettier, it is cheaper to use.

It controls speed. Front-end code decides whether the answer to a customer’s question arrives before their patience runs out. This is a development problem wearing a design costume.

It puts the right information in the right place. The objection a customer has at the point of the add-to-cart button is not the objection they had in the hero. Most wins that get filed as “design” are actually information architecture: the same content, ordered by when the customer needs it.

Notice what is missing. Colour palettes, typefaces, illustration style, and animation are not on that list. They matter for brand, and brand matters commercially over years. They are just not the lever you pull when the number needs to move this quarter.

If you cannot say which of those three mechanisms a change operates through, it is a taste change. Ship it if you want, but do not budget it as conversion work.

When a redesign is justified

There are real triggers. They are all structural.

The theme cannot be extended. On an old theme architecture, every change is a developer ticket and nothing can be configured. That is a compounding tax on every future test.

The catalogue has outgrown the navigation. You launched with 12 SKUs, you now have 400 across six categories, and the menu is still the one you built in year one.

The mobile experience is structurally broken, not cosmetically dated. Layouts that break at common widths, tap targets that miss, a buy box that takes two scrolls to reach.

You are replatforming anyway, or moving off a heavily customised legacy build.

The brand genuinely misrepresents the product, usually after a price repositioning. A store that moved from $40 to $180 average order value and still looks like a discount site has a real mismatch.

What is not a trigger: the team is bored, a competitor launched something shiny, or a new hire wants to make their mark. Those produce the most expensive projects in ecommerce, because they carry the full cost of a rebuild and none of the diagnosis.

Before anyone briefs a designer, run the diagnostic work in the conversion research chapter. A redesign brief written from research is a list of problems to solve. A redesign brief written from taste is a mood board with a budget attached.

The cost nobody prices: you lose your accumulated learning

This is the part that gets left out of every redesign proposal, and it is the largest hidden cost.

Everything you have learned about your store is attached to the store you have. Every test result was a result on a specific page, with specific copy, in a specific layout. Rebuild that page and those results become history rather than knowledge. You cannot say “we tested that” about a page that no longer exists.

It goes further than tests:

Your analytics baselines break. Event tracking, funnel steps, and page groupings all change. The comparison you want to make in month three, new site versus old, is usually not clean enough to make.

Your speed profile resets. A new theme with new apps has a new performance story, and it is frequently worse than the one you spent two years fixing.

Your search performance is at risk. URL changes, template changes and content cut in the redesign show up in organic traffic six to twelve weeks later, by which point nobody connects it to the launch.

Three things reduce the damage:

Keep a decision log. One document, one line per test or change: what you changed, what happened, what you concluded. This is the cheapest insurance in CRO and almost nobody keeps it.

Port your learning as requirements, not as inspiration. If a test showed that delivery dates near the button mattered, that is a line in the new build spec, not a nice-to-have for phase two.

Roll out by template, not all at once. New product page template first, then collection, then homepage. Each phase is readable on its own. A big-bang launch answers one question, “is the new site better or worse”, and answers it badly, because a whole-site test compares two bundles of hundreds of decisions and tells you nothing about which decision did what. The mechanics of why that is hard to read are in the A/B testing chapter.

Brand expression against usability

These are not enemies. They compete for the same pixels and the same seconds, which is different.

The workable split: brand owns the surfaces that do not carry a decision, usability owns the transactional path.

Brand gets the hero, editorial content, photography direction, category headers, the about page, packaging, the newsletter. Spend real money there. It is what makes a $180 product feel like a $180 product.

Usability owns the buy box, the variant selector, the cart, the checkout entry, and every form. On those surfaces, convention wins. A customer should not have to learn how your store works at the moment they are deciding to spend money.

The specific places brand expression usually costs orders:

  • Body copy below 16px on mobile, because the type specimen looked elegant at desktop sizes
  • Light grey on white, which fails contrast and is unreadable outdoors, where a large share of phone shopping happens
  • Scroll-triggered reveals that delay the content a customer came for
  • Clever navigation labels that do not name the thing being sold
  • Hover-only affordances, which do not exist on touch devices

An illustrative example: a fictional homeware brand, Bramble & Oat, sets product titles in a display serif at 14px and 40% opacity to match its print catalogue. It reads beautifully on the designer’s monitor, and on a phone in daylight customers cannot read the product names. That is not a brand decision, it is an unlogged conversion decision.

Where the two genuinely conflict on a transactional surface, test it rather than argue about it. That is what testing is for.

Speed is a conversion variable, and Core Web Vitals is the scoreboard

Google publishes thresholds for Core Web Vitals, and you can treat these as authoritative because they are the published definitions rather than someone’s benchmark study. Measured at the 75th percentile of real page loads:

  • Largest Contentful Paint at or under 2.5 seconds
  • Interaction to Next Paint at or under 200 milliseconds
  • Cumulative Layout Shift at or under 0.1

INP replaced First Input Delay as the responsiveness metric in 2024, which caught out a lot of stores that had quietly passed on the old measure.

We will not quote you a revenue-per-100-milliseconds figure. Those numbers circulate endlessly, they come from a handful of very large sites with traffic nothing like yours, and repeating them here would be the borrowed statistic this course argues against.

The mechanism is enough. A slow page loses people before they see anything, and a page that shifts under a thumb causes mis-taps at the moment of purchase.

Two measurement rules:

Use field data, not lab scores. The speed report in your Shopify admin runs a lab test against a few pages. It catches regressions; it does not describe what your customers experience. Real-visitor data through PageSpeed Insights or Search Console is the version that counts.

Measure the templates that carry money. Site-wide averages get dragged around by whichever page has the most traffic. Pull the product and collection templates out and look at them on mobile, on a mid-range Android rather than the newest iPhone in the office.

Where Shopify themes actually lose their speed

Four culprits account for most of it. In rough order of how much they cost:

Apps. Most front-end apps inject JavaScript that loads on every page whether or not the feature appears there. Fifteen apps each adding a modest script is a slow store, and no single one of them looks like the problem. This is the biggest item on nearly every store we look at.

Sliders and carousels. A hero carousel is a heavy JavaScript component that loads several large images so that most customers can see the first one. Worse, if your LCP element sits inside a carousel or behind a lazy-load, your headline metric is being set by a library. Carousels also perform poorly on their own terms: people interact with the first slide and largely ignore the rest.

Unoptimised imagery. Shopify’s CDN will serve correctly sized, modern-format images with responsive srcset if the theme asks it to. Plenty of themes do not. The recurring faults are the same everywhere: images uploaded at 4000px and displayed at 600, no width and height attributes so the layout shifts as they arrive, below-fold images loading eagerly, and the LCP image lazy-loaded so the browser deliberately delays the one thing that matters.

Render-blocking script and fonts. Synchronous third-party tags in the head, a jQuery dependency inherited from an old theme, and custom fonts loaded without a sensible fallback so text either flashes or does not appear at all.

Two more worth checking: templates that loop over large collections without pagination limits, and pages that render twenty sections and hide most of them with CSS. Hidden is not free. The browser still does the work.

App bloat is a tax you pay twice

Every app costs you a monthly fee and a slice of page performance, and the second cost is invisible until someone measures it.

Run the audit annually. It takes an afternoon:

List every app, its monthly cost, and what it does. Most stores find at least two they are paying for and not using, and one nobody can identify.

Ask what would break if you removed it. If the answer is “nothing visible”, you have your answer.

Check for overlap. Reviews, user-generated content, upsells, popups and loyalty tools have converged. Four apps frequently do the work of two.

Check the theme after uninstalling. Removing an app does not reliably remove the code it added to your theme files. Orphaned snippets and script tags from apps you dropped two years ago are common, and they still load.

Hold new apps to the standard you hold any other change to. An app that adds a front-end script should show a measured result. “It might help” is how stores end up with thirty of them.

Consolidation is unglamorous and it is usually the highest-return technical work available on a mature store, because it improves speed and reduces cost at the same time.

Accessibility is a conversion issue, not a compliance chore

Almost everything on an accessibility checklist is also a mobile usability fix. That is not a coincidence. Both are about whether a person in imperfect conditions can complete a task.

  • Contrast. WCAG 2.2 AA requires 4.5:1 for body text. The customer squinting at your grey-on-white spec table in bright sunlight is not an edge case.
  • Tap targets. WCAG 2.2 sets a minimum target size of 24 by 24 CSS pixels at AA. Undersized swatches and quantity steppers cause mis-taps, and a mis-tap in a variant selector means someone buys the wrong size or gives up.
  • Real labels on form fields. Placeholder-only forms fail for screen readers and fail for everyone else the moment they start typing and the label disappears.
  • Visible focus states. Keyboard users need them, and so does anyone whose touch input is imprecise.
  • Errors described in text, next to the field. Colour alone is not a message.
  • Keyboard operation of the cart drawer, the menu, and any modal. A drawer you cannot escape from is a trap, not a feature.

Two honest notes. Overlay widgets that promise instant compliance do not fix the underlying markup, and the accessibility community is broadly hostile to them. And the regulatory picture is real: the European Accessibility Act’s requirements began applying to consumer ecommerce in the EU in June 2025, and litigation over inaccessible storefronts is routine in the US. That is context, not legal advice.

The commercial framing is simpler than the legal one. The accessible version of your variant selector is the usable one.

Build sections so future tests are cheap to ship

Here is why most stores test slowly: every test is a development ticket. Someone briefs it, someone builds it, someone QAs it, and by the time it ships the quarter is over.

The fix is architectural, and it is the most valuable thing a good Shopify build gives you.

Expose the things you will want to change as theme settings. Headline text, badge on or off, the order of trust elements, whether the size guide is inline or in a drawer. If changing it needs a code deploy, you will not change it.

Build reusable blocks rather than one-off sections. A trust bar, a comparison table, an FAQ accordion, a review block. Once these exist as blocks, a new landing page is an afternoon rather than a project. That compounds directly into the work in the landing pages chapter.

Drive content from metafields where it varies by product. Sizing, materials, care, delivery class. Merchandisers then update products instead of queueing theme changes.

Prefer theme-level variants over client-side script injection. Tests that rewrite the page after it loads cause the flicker problem covered in the A/B testing chapter, and flicker biases the result against your variant.

One caution: do not parameterise everything. A theme with 200 settings is its own maintenance problem. Expose what you actually intend to test.

This is most of what separates a store that ships four tests a month from one that ships four a year, and it is the reason we treat conversion-focused Shopify builds as testing infrastructure rather than as a design deliverable.

What Shopify Plus actually unlocks, and what it does not

Plan features move, so verify against Shopify’s current plan comparison before you budget anything against this. The categories have been stable, though.

What Plus genuinely buys you: deeper checkout customisation, including checkout branding and more extension points now that checkout.liquid has been retired in favour of checkout extensibility; B2B and wholesale functionality on the same store; expansion stores for other regions; higher platform and API limits; and more direct support and launch resource.

What it does not buy you: speed, a better product page, full HTML control of checkout, or a team that knows what to change. Plus is a bigger set of tools. It is not a strategy, and it does not fix anything that is wrong with the storefront you already fully control.

The honest test for upgrading: write down three specific things you would ship in the first 90 days on Plus that you cannot ship now. Price them. Compare that to the annual plan difference. If the list is thin or the items are speculative, the money buys more on the front end you already own.

The pattern we see most often is a store upgrading for checkout control when its actual problem sits well before checkout, in the cost transparency and delivery expectations covered in the cart and checkout chapter.

What loses

The house rule applies here as everywhere: the losses are more useful than the wins. These are the Shopify design and development decisions we see cost the most money.

The big-bang redesign with no baseline. Everything changes on one night, nobody recorded the old numbers properly, and six weeks later the argument about whether it worked cannot be settled with data.

Homepage obsession. The homepage gets the design attention and the review meetings, while a large share of sessions land straight on a product or collection page from ads, email and search. Check your own landing page report before allocating the budget, then read the product page chapter.

Buying a premium theme and stacking thirty apps on it. The theme was fast when you bought it. The apps are why it is not now.

Hiring a design studio with no measurement plan. You will get a good-looking store and no way to know what it did. Ask, in the pitch, how the work will be measured and who is accountable for the number.

Testing a whole redesign as a single test on low traffic. A store with modest traffic cannot read a site-wide test in a sensible window. Run your own numbers in the significance calculator before you plan a launch around a result you will never get.

Popups that fire immediately on load. They damage the first impression, they hurt layout stability, and they are usually configured by whoever installed the app rather than by anyone thinking about the visit.

Animating things that carry decisions. A fade-in on the add-to-cart region is a delay dressed as polish.

One illustrative case: Northwind Kettlebells rebuilds on a new premium theme, keeps every app, adds a full-screen video hero and launches it all on a Monday. Sessions look normal, orders are down, and the team spends a quarter arguing about the hero. The cause could be the video, the slower product template, a lost delivery-date line, or an unlucky month of traffic mix. With that launch structure, they will never know. That is the real cost of a big-bang redesign: not the drop, the unanswerable question.

The order to do this in

If your store is slow or structurally broken on mobile, fix that first and do not test it. Those are defects, not hypotheses, and you do not need an experiment to justify making the site work.

Then get the information architecture and the buy box right, because that is where the decisions happen. Then build the section toolkit that makes tests cheap. Then test, with a real backlog and honest measurement.

Doing it in the other order, testing on a theme nobody can change, is how stores spend a year busy and compound nothing. If you want that sequence run with the build and the tests under one roof, that is what our growth engagement does.

What founders ask before they book.

Will a Shopify redesign increase my conversion rate?

Sometimes, and nobody can tell you in advance. A redesign changes hundreds of things at once, so if the number moves you will not know which change did it, and if it drops you will not know what to roll back. Redesigns are justified by structural problems, such as a theme that cannot be extended, a catalogue that has outgrown its navigation, or a mobile experience that is genuinely broken. Being bored of the current design is not one of them.

How fast does a Shopify store need to be?

Google publishes the thresholds it considers good for Core Web Vitals, measured at the 75th percentile of real visits. Largest Contentful Paint at or under 2.5 seconds, Interaction to Next Paint at or under 200 milliseconds, and Cumulative Layout Shift at or under 0.1. Use field data rather than the lab score in your Shopify admin, and look at the product and collection templates specifically rather than a site-wide average.

Do Shopify apps slow down my store?

Front-end apps usually do, because most inject JavaScript that runs on every page load whether or not the feature is in use. The cost is rarely one app; it is fifteen apps each adding a little. Audit them by asking what each costs per month, what it adds to page weight, and what would break if it were removed. Uninstalling does not always remove an app's code from your theme, so check the theme files afterwards.

Is Shopify Plus worth it for conversion?

Plus buys deeper checkout customisation, B2B functionality, expansion stores and higher platform limits. It does not make your store faster, fix your product page, or hand you a team that knows what to change. Before upgrading, write down three things you would ship in the first 90 days that you cannot ship now, price them, and compare that to the plan difference. If the list is thin, the money buys more on the front end you already control.

Book A Call

Thinking about rebuilding your Shopify store?

Before you brief a designer, it is worth knowing which parts of the current site are earning and which are costing you. We will walk the templates, check the speed and the mobile path, and tell you whether a rebuild is the right call. Free 20-minute call.

  • 437+ stores built
  • 2,000+ tests shipped
  • No call needed to start

Claim your free audit of one page

Send us one page of your store and we’ll review it on video: what’s leaking, what to fix first, and what to leave alone. Changes you can ship this week to lift CVR, AOV and revenue per visitor.

Get My Free Audit