Shopify Speed Optimization

Shopify speed optimization services for a store where everything slow is something you installed

You cannot touch the server, the CDN or the JavaScript Shopify injects on every page. That part is already fast. Everything left is your theme, your apps and your tags, which means Shopify performance is less an engineering problem than an inventory one.

App auditing, theme weight, image delivery and INP, measured on real mobile field data rather than a lab score.

The basics

What are Shopify speed optimization services?

Shopify speed optimization services are the work of getting a Shopify store through Core Web Vitals within the constraints of the platform: auditing installed apps for the scripts they load on every page, reducing theme weight, fixing hero image delivery and layout shift, cutting third party tags, and improving Interaction to Next Paint on the cart, filters and variant selectors.

This is deliberately narrower than the general speed service, because Shopify removes most of the usual levers.

If you are on WordPress or a custom build, website speed optimization covers hosting, caching and server work that simply does not apply here. Shopify speed optimization services start where that list runs out.

Straight talk

What you can and cannot fix on Shopify

Four things decide how this work goes, and the first one saves you a lot of wasted effort.

The infrastructure is not your problem, and not your lever. Shopify runs a global CDN, converts images to WebP automatically, and serves over HTTP/2 with Brotli compression, keeping time to first byte under about 300 milliseconds in most regions. You cannot improve it and you do not need to. You also cannot remove the core JavaScript Shopify injects on every page.

Apps are the single biggest cause of failure. Reviews widgets, chat, upsells, popups, loyalty, tracking pixels. Most load JavaScript on every page of the store rather than only where they are used, including pages where the feature never appears. This is where the weight is, and it is weight somebody chose to add.

INP replaced First Input Delay in March 2024, and it is stricter. Interaction to Next Paint measures how quickly the store responds to every interaction rather than just the first, with a target under 200 milliseconds. That means slow cart drawers, sluggish variant selectors and laggy collection filters now count against you for the whole session.

The median store has almost no headroom. In a benchmark of a thousand real Shopify stores, only 48% passed all three Core Web Vitals on mobile, and the median mobile largest contentful paint sat at 2.26 seconds against a 2.5 second threshold. That is a quarter of a second of slack. One more app, or one unoptimised hero image, is the difference between passing and failing.

Reported gains from getting into the good thresholds have run around 15 to 30% on conversion and 12 to 20% on organic traffic. Treat those as what has been possible rather than what is promised, because the starting point decides most of it.

The dividing line

Yours, Shopify’s, and the bit in between

Worth knowing before you pay anyone to optimise something they cannot reach.

 Who controls itWhat that means for you
Server and hostingShopifyAlready fast. Nothing to do, and nobody can sell you an improvement
CDN and compressionShopifyGlobal CDN, HTTP/2 and Brotli by default
Core platform JavaScriptShopifyLoads on every page and cannot be removed
Image formatShopify, mostlyWebP conversion is automatic. Dimensions and priority are still yours
Theme codeYouWeight, render blocking assets, how much loads before anything paints
Installed appsYouThe largest controllable factor by a distance
Third party tagsYouTag managers, pixels, chat and review widgets

Three rows are yours. That is the entire surface area of Shopify speed work, and it is why an audit starts with a list of apps rather than a Lighthouse report.

Scope of work

What our Shopify speed optimization services cover

Ordered roughly by how much they typically return, which is not the order most agencies work in.

App audit and removal

Every app mapped to what it loads, on which templates, and whether anyone still uses the feature. Removals usually deliver more than every other item on this list combined.

Conditional app loading

For the apps you are keeping, restricting their scripts to the templates that actually need them instead of every page in the store.

Hero image and LCP work

The largest element on the page loaded with priority, correctly sized, and not sitting behind a carousel or a script. Usually the fastest single LCP gain.

INP and interaction work

Cart drawer, variant selectors and collection filters profiled and fixed, because INP now measures every interaction rather than the first one.

Theme weight reduction

Unused sections, abandoned experiments, duplicate libraries and render blocking assets left behind by previous developers.

Layout shift fixes

Reserved space for images, banners and dynamically injected app content, so the page stops moving under the customer’s thumb.

Tag and pixel consolidation

Tracking rationalised, deferred where it can be, and duplicates removed. Most stores are firing at least one pixel twice.

Font loading

Weights trimmed to what the design actually uses, loaded so text is visible immediately rather than after a blank pause.

Field data monitoring

Tracked on real user data rather than a lab score, per template, so a regression from a newly installed app is caught in days.

You can get a first read on your own templates with our bulk PageSpeed checker, which is worth running across a home page, a collection and a product page rather than the home page alone.

How we work

How the speed work runs

Inventory before measuring

Every app, tag and third party script listed with what it loads and where. This tells you where the weight is, which a performance score on its own never does.

Remove before optimising

Deleting an unused app beats optimising it every time. This step is unglamorous, occasionally awkward internally, and it is where most of the gain lives.

Fix the templates that matter

Home, collection and product, in that order of traffic. Hero images, layout shift and interaction handlers, measured per template rather than as a site average.

Watch the field data

Real user metrics tracked afterwards, because the number that counts updates over weeks and a lab score does not tell you whether customers felt anything.

The newer metric

INP is where Shopify stores quietly fail

Largest contentful paint gets the attention because it is easy to measure. Interaction to Next Paint is where a store that loads acceptably still feels bad to use.

The cart drawer

Add to cart triggers an app, which triggers a fetch, which redraws the drawer. Every one of those steps is measured, and the customer feels all of them.

Variant selectors

Changing colour or size re-renders price, availability and images. On a heavy product template this is routinely the slowest interaction in the store.

Collection filters

Filtering a large collection can block the main thread while it rebuilds the grid, which is exactly the interaction shoppers repeat most.

The target is under 200 milliseconds, and because INP looks at the whole session rather than the first click, a single sluggish interaction repeated across a visit is enough to fail it. Fixing this rarely means removing the feature. It usually means the feature was doing more work than it needed to.

Honest limits

What speed work will not do

Three things worth saying before anyone buys this, because the category attracts inflated claims.

A perfect score is not the goal and is often not reachable. Shopify injects its own JavaScript on every page and you cannot remove it. Any agency promising a hundred on a Shopify store is either testing a page nobody visits or is about to break something you rely on.

Some slow apps are load bearing. The reviews widget that costs you 300 milliseconds may also be worth more in conversion than it costs in speed. We measure both sides rather than deleting things because they show up in a waterfall.

Speed will not fix a store that does not convert. It removes friction from a journey that already works. If the product page does not answer the question people are asking, a faster version of it will not sell more, and we would rather tell you that than take the speed budget.

And one more: results decay. A store optimised today drifts back as apps get added over the following year, which is why the monitoring matters more than the initial project.

Investment

How much do Shopify speed optimization services cost?

A fixed project fee, with monitoring available afterwards because that is where the gains are held rather than won.

Speed audit

App inventory, per template field data, and a ranked list of what to remove, defer and fix, with the expected effect of each. Yours to act on with or without us.

Optimisation project

App removal and conditional loading, hero and layout shift fixes, INP work and theme cleanup. Fixed scope and price, quoted after the audit.

Ongoing monitoring

Field data watched per template with alerts when a newly installed app costs you headroom. Modest monthly fee, and the thing that stops this recurring annually.

Judge the work on field data and on conversion, not on a Lighthouse number. Lab scores are useful for diagnosis and close to meaningless as a measure of whether customers had a better time.

Common questions about Shopify speed

Because the slow part is almost never Shopify. The platform runs a global CDN, converts images to WebP automatically and serves compressed over HTTP/2, keeping time to first byte under roughly 300 milliseconds in most regions. What slows a store down is the theme, the installed apps and third party tags, all of which are things that were added to it.

They are the biggest controllable factor by a distance. Most apps load JavaScript on every page rather than only where the feature appears, so a reviews widget used on product pages is often also loading on your home page, your collections and your policies. Removing unused apps typically returns more than every other optimisation combined.

Interaction to Next Paint replaced First Input Delay as a Core Web Vital in March 2024, with a target under 200 milliseconds. The difference is that it measures every interaction across the session rather than just the first one, so slow cart drawers, variant selectors and collection filters all count. It is where plenty of Shopify stores that load acceptably still fail.

The score is the wrong target. What matters is passing Core Web Vitals on real user data, meaning largest contentful paint under 2.5 seconds, INP under 200 milliseconds and cumulative layout shift under 0.1. For context, a benchmark of a thousand Shopify stores found only 48% passing all three on mobile, with a median mobile LCP of 2.26 seconds, so most stores are close to the line rather than comfortably inside it.

Not honestly, and neither can anyone else on a real store with apps. Shopify injects its own JavaScript on every page and you cannot remove it. Anyone promising a perfect score is testing a page nobody visits or planning to strip functionality you actually need. The realistic goal is passing the thresholds on field data, which is what affects both ranking and customers.

Reported improvements from reaching the good thresholds have run around 15 to 30% on conversion and 12 to 20% on organic traffic, but the honest answer is that it depends entirely on where you start. A store failing badly has a great deal to gain. A store already inside the thresholds has very little, and should spend the money on the product page instead.

Not on their own. Stores drift back as apps get installed over the following year, and given the median store has around a quarter of a second of headroom, it does not take many additions to slip back under. That is why we would rather sell modest ongoing monitoring than the same optimisation project again in eighteen months.

Find out what your apps are costing you

We will inventory every app and tag on your store, show you what each one loads and where, and give you a ranked list of what to remove, defer and fix with the expected effect of each. Most stores have a surprising amount to delete. No obligation, no sales pitch.

Get my free speed audit
  • App and tag inventory
  • Field data per template
  • Ranked fix list