Here's a conversation we have almost every week. A merchant comes to us frustrated. Their store is slow, they know it, their customers know it, and Google knows it.
They've already tried a caching extension. They've upgraded their hosting plan. They've run GTmetrix and PageSpeed Insights until the screenshots are burned into their retinas.
They're sitting at a 38 on mobile. Sometimes lower.
They want to know what else to try.
The answer, in most cases, is that they've been treating the symptoms rather than the cause. Caching helps. Better hosting helps.
But if your store is built on Magento's Luma theme, you're fundamentally fighting the architecture itself. And architecture always wins.
This isn't a minor tuning issue. Google's research shows that 53% of mobile users abandon a site if it takes longer than 3 seconds to load.
The average Luma-based Magento store loads in 5–9 seconds on mobile.
The gap between those two numbers is your lost revenue.
Let me walk through the real causes of Magento slowness in priority order, because not all of them are equally important — and fixing the wrong ones first wastes money and time you don't have.
The Business Case for Caring About This Right Now
Before the technical breakdown, let's be clear about what's actually at stake, because "your PageSpeed score is low" undersells the problem.
A 1-second delay in page load time reduces conversions by 7%, page views by 11%, and customer satisfaction by 16%.
For a Magento store doing £50,000 per month, that single second costs roughly £3,500 every month. Not annually, monthly.
Beyond conversions, Google has been using Core Web Vitals as a ranking signal since the Page Experience update rolled out. Stores that consistently fail LCP, FID/INP, and CLS benchmarks are at a structural disadvantage in organic search — which, for most merchants, is their highest-margin acquisition channel.
A slow store isn't just a user experience problem. It's an SEO problem and a revenue problem wrapped together.
The Causes of a Slow Magento Store— Ranked by Real-World Impact
Not all performance problems are equal. The biggest mistake merchants make — and the most expensive — is optimising for the wrong thing first. Here's what actually moves the needle, in order.
1. The Luma Theme's JavaScript Stack
This is the one most performance guides bury in a footnote or skip entirely, which is bizarre because it's responsible for the majority of mobile performance failures on Magento stores.
Luma — Magento's default theme, and the base of most custom Magento themes — was built on RequireJS, KnockoutJS, and jQuery. These were reasonable technology choices when Magento 2 launched in 2015.
In 2026, this stack ships between 400kb and 900kb of JavaScript to every single page visitor, regardless of what they're actually doing on that page. Visiting the homepage? Here's 600kb of JS. Looking at a product page? Same 600kb. The bundle doesn't meaningfully change based on what the page actually needs.
On a fast desktop connection, this is slow. On a mid-range mobile device — which is how approximately 63% of global ecommerce traffic is now generated — it's devastating. The browser has to download the JavaScript, parse it into machine-readable instructions, and execute all of it before the page becomes interactive.
This process can take 3–8 seconds on a typical mobile device. No amount of caching makes this fast. You cannot cache your way out of a JavaScript execution problem. It's the architecture.
The fix — the only real fix — is switching to the Hyvä theme. Hyvä replaced RequireJS and KnockoutJS with Alpine.js and Tailwind CSS, reducing frontend JavaScript to 60–130kb per page. That's a reduction of 80–90% in JavaScript weight.
The real-world results are consistent: stores migrated to Hyvä routinely hit 80–96 on mobile PageSpeed Insights, compared to 25–55 on an equivalent Luma build. Google's own Lighthouse scoring reflects this directly in Core Web Vitals: LCP (Largest Contentful Paint) and INP (Interaction to Next Paint) improve dramatically because the browser simply has less work to do.
If you're sitting at a 35–50 mobile score right now and wondering why nothing you try seems to help, this is almost certainly why.
Want to know if Hyvä is the right move for your store?
We've helped dozens of Magento merchants go from sub-50 mobile scores to 90+ through Hyvä migrations.
2. Full-Page Cache Not Properly Configured
Magento has a built-in full-page cache (FPC). When it's working correctly, it stores rendered HTML pages and serves them from memory rather than rebuilding every page from scratch on each request.
When it's disabled, misconfigured, or being bypassed by an extension, every page request hits PHP and the database from scratch — a process that can take 800ms to 3 seconds on its own, before a single byte of HTML reaches the browser.
With FPC enabled and Varnish deployed in front of it, Time to First Byte (TTFB) can drop from 1,500ms to under 100ms.
That's a 15x improvement in server response time. Varnish is a reverse proxy cache that sits in front of your web server and serves cached responses at memory speed — it's one of the most impactful server-side changes you can make to a Magento store.
To check whether FPC is working, go to System → Cache Management in your Magento admin. If Full Page Cache shows as disabled, that's your first action.
If it's enabled but your TTFB is still high (you can measure this in Chrome DevTools under the Network tab, or via WebPageTest), the cache may be getting invalidated too frequently, or certain extensions may be generating unique cache keys that prevent page caching from functioning properly.
One important caveat: some extensions — particularly personalisation tools, cart price rule calculators, and certain layered navigation extensions — disable Magento's full-page cache by design, because they need to render dynamic content.
If you recently installed an extension and your TTFB got worse, this is likely why. A good Magento developer can identify which extensions are bypassing the cache and find ways to restore cacheability, often using ESI (Edge Side Includes) for dynamic blocks.
This fix is one of the easier wins on this list. But to be clear about what it does and doesn't do: it dramatically improves server response time. It has no effect on the JavaScript weight problem above. You can have a sub-50ms TTFB and still score 38 on mobile PageSpeed because of Luma's JS bundle.
3. Under-Resourced or Poorly Configured Hosting
Magento is a genuinely complex PHP application. It has more than 40,000 PHP files. Its bootstrapping process — the work it does before rendering a single product — involves autoloading hundreds of classes, reading configuration, and initialising the dependency injection container. This is not a WordPress site. It requires real resources.
Shared hosting is essentially incompatible with good Magento performance. On shared hosting, your store competes with dozens or hundreds of other sites for CPU and memory.
PHP memory limits are often set at 256MB or lower — Magento recommends a minimum of 2GB. Database servers are shared, slow, and not tuned for Magento's query patterns.
The minimum viable Magento hosting setup in 2026 looks like this: a VPS or cloud server with at least 4GB of dedicated RAM, PHP-FPM with OPcache enabled (OPcache alone can reduce PHP execution time by 50–70% by caching compiled PHP bytecode), a dedicated MySQL or MariaDB server, and Redis for session and full-page cache storage. Managed Magento hosting platforms — Cloudways (with a properly sized plan), and Adobe Commerce Cloud — handle most of this correctly by default.
It's worth saying again: better hosting improves TTFB and page generation time. It doesn't touch JavaScript weight. A store on enterprise-grade hosting, with perfect caching, will still score poorly on mobile PageSpeed if it's on Luma. The ceiling imposed by the theme is real.
Not sure if your hosting is the bottleneck?
Our free performance audit includes a server response analysis that tells you exactly whether your host is contributing to the problem — and by how much.
4. Too Many or Poorly Coded Extensions
The Magento extension ecosystem is enormous — there are thousands of third-party extensions available through the Magento Marketplace, individual vendors, and custom development. Most are fine. Some are not. And the impact of extensions on performance varies enormously based on how they're written.
A well-coded extension that adds two targeted database queries is invisible in terms of performance impact. An extension that runs 40–80 database queries on every page load, loads a full JavaScript library globally when it only needs it on one page, or hooks into events that fire on every request — that extension can add 500ms to 2 seconds to your page load on its own.
The problem is that extension performance issues are invisible without profiling. You can't diagnose them by looking at the extension manager. Blackfire.io and New Relic are the two tools most commonly used for this. Blackfire, in particular, produces a call graph that shows exactly which PHP functions are consuming the most time — it's one of the best investments you can make for ongoing Magento performance management.
Common extension-related performance issues to look for include extensions that disable block caching (check for `cacheable="false"` in layout XML files), extensions that add JavaScript to every page when it's only needed on specific pages, and extensions that haven't been updated to support Magento's flat catalogue or have inefficient EAV queries.
A good rule of thumb: every extension you're not actively using should be disabled. Even extensions that "do nothing visible" on a given page add to Magento's PHP bootstrapping process. A store with 80 enabled extensions will bootstrap slower than the same store with 40 — even if 40 of those extensions aren't actively rendering anything.
5. Unoptimised Images
Large, uncompressed product images are a genuine contributor to slow page load times — particularly on mobile, where bandwidth is constrained and images can represent 40–60% of total page weight for a product listing page. This is usually the first thing an agency recommends because it's visible, easy to demonstrate progress on, and safe. It's also the fix with the smallest ROI if your score is already below 60, because you're adding polish to a broken foundation.
That said, it's absolutely worth doing, especially for stores with large catalogues. Magento includes built-in image resizing, but it doesn't compress aggressively by default. Switching to WebP format (which Google reports produces images 25–34% smaller than comparable JPEG files at equivalent visual quality) and serving images through a CDN (Content Delivery Network) with edge caching can meaningfully reduce page weight. Fastly, Cloudflare, and Amazon CloudFront are all solid options for this.
For stores on Hyvä, image optimisation can push an already-good 80+ score to 90+. For stores on Luma, it will move the needle modestly — typically 5–10 points on PageSpeed — but won't address the fundamental problem.
One client came to us having spent $800 on "Magento performance optimisation" from a freelancer — image compression and some caching tweaks. His mobile score went from 29 to 34. He'd paid to fix the small problems while the big one sat untouched. Six weeks after a Hyva migration, he was at 91.
The Diagnostic: Where to Start Before Spending Anything
Before committing to any of the fixes above, run the diagnostic first. It takes 10 minutes and tells you exactly what you're dealing with.
Step 1: Run Google PageSpeed Insights
pagespeed.web.dev on three pages: your homepage, a product page, and your checkout.
Run both the mobile and desktop versions. Screenshot everything. Mobile scores below 60 are broadly failing by Google's standards; below 50 is severely failing.
Step 2: Note your Core Web Vitals scores.
LCP (Largest Contentful Paint) should be under 2.5 seconds — this measures how quickly the main content appears. INP (Interaction to Next Paint) replaced FID and should be under 200ms — this measures how quickly the page responds to user input. CLS (Cumulative Layout Shift) should be under 0.1 — this measures visual stability. These are the three signals Google uses directly as ranking factors.
Step 3: Check your TTFB.
In Chrome, open DevTools (F12), go to Network, reload the page, and click the first HTML document in the list. Under Timing, look for "Waiting (TTFB)". Under 200ms is good. 200–800ms is acceptable. Over 800ms is a problem that requires server or cache work.
Step 4: Check your total JavaScript weight.
In PageSpeed Insights, look at the "Reduce unused JavaScript" recommendation. If it's flagging 400kb or more of JavaScript, you're on Luma and the theme is the primary cause of your mobile score.
What these numbers tell you: a low mobile score with high JS weight and an acceptable TTFB is a frontend (theme) problem. A high TTFB with acceptable JS weight is a server/cache problem. Both failing means you have both problems, in which case — fix the server side first because it's faster and cheaper, then address the theme.
Not confident interpreting these results yourself?
Send us your PageSpeed scores and we'll tell you exactly what they mean and what to do first.
The Fix That Actually Moves the Needle
Let's be direct about something, because too many performance guides dance around it: if your Magento store is on Luma and scoring below 60 on mobile PageSpeed, you can optimise images, tune your server, configure Varnish perfectly, and disable every unnecessary extension — and you will move from a 35 to maybe a 48. That's still a failing score.
The move from 48 to 90+ requires fixing the frontend. And that means Hyvä.
This isn't positioning dressed up as technical advice. Hyvä's own published case studies consistently show before/after scores of 20–45 mobile improving to 85–96 after migration. Independent audits and merchant testimonials corroborate this. The delta between Luma and Hyvä on mobile PageSpeed is typically 40–60 points — not a marginal improvement, but a categorical jump from failing to excellent.
The reason is simple: Hyvä removes the architectural cause of the problem. It doesn't optimise Luma's JavaScript — it replaces it entirely. Alpine.js (Hyvä's interaction framework) weighs around 15kb minified and gzipped. Tailwind CSS, purged to only the classes your theme actually uses, typically compiles to 10–25kb. Compared to Luma's 400–900kb JS bundle, the reduction in parse time and execution time on mobile devices is enormous.
According to Hyvä's performance benchmarks, a typical Hyvä store reduces Time to Interactive by 3–5 seconds on mobile compared to an equivalent Luma build. For a merchant with significant mobile traffic — which is most merchants — this translates directly into higher conversion rates and better organic search rankings.
If Hyvä Migration Isn't in Your Budget Right Now
Fair point — Hyvä migration isn't completely free, and not every store can prioritise it immediately. Here's what to do in the meantime, in priority order.
First, make sure full-page cache is enabled and working. Check System → Cache Management. If FPC is disabled, enable it. If you have Varnish available on your hosting plan, configure it. The official Magento documentation on Varnish configuration is solid and worth following carefully.
Second, run an image audit. Use a tool like ImageMagick or Squoosh to batch-convert your product images to WebP format. For a store with a few hundred products, this is an afternoon of work. For stores with tens of thousands of SKUs, it may be worth an extension like TinyPNG for Magento or setting up an image optimisation pipeline.
Third, audit your extensions. Go through every enabled extension and ask: is this actively being used in the last 30 days? If the answer is no, disable it. For the remainder, check whether any are flagged in PageSpeed Insights under "Reduce unused JavaScript" — that's a sign of an extension loading JS globally when it shouldn't be.
Fourth, check your PHP version.
PHP 8.2 is significantly faster than PHP 7.4, and PHP 8.3 improves further. If you're on an older PHP version, upgrading (after compatibility testing) can meaningfully reduce PHP execution time without any code changes.
These steps will improve your scores. Based on what we've seen across dozens of Magento audits, they'll typically move a store from catastrophic (25–35) to functional (45–60). They won't get you to 90+ on mobile. But they'll get you to respectable while you plan the proper fix — and they'll ensure that when you do migrate to Hyvä, you're starting from a clean base.
Ready to plan your Hyvä migration? Our Hyvä migration guide walks through the entire process — from audit to launch. If you want us to handle it end to end, our Hyvä development service is where to start.
What About Magento Performance Extensions?
There are dozens of extensions that claim to speed up Magento — lazy loading, JavaScript deferral, CSS minification, critical CSS extraction, image optimisation.
None of them fix the underlying JavaScript weight problem that causes poor mobile scores on Luma.
Think of it this way: you can tune a car all you want, but if the engine is the wrong size for what you need, tuning has hard limits. The theme is the engine. Performance extensions are the tuning. JavaScript deferral extensions can delay the execution of Luma's bundle, but they can't reduce its size.
Critical CSS extraction can improve First Contentful Paint, but it doesn't change how much JS the browser ultimately has to execute. The core metric: INP, Interaction to Next Paint, which measures how quickly your page responds to user input, is dominated by JS execution time. You can't defer your way out of that.
If you're on Hyvä, performance extensions can genuinely help push an already-good score higher. Lazy loading improvements, CDN configuration, database query caching, these are meaningful marginal gains on top of a solid foundation.
On Luma, they mask the problem without solving it. The PageSpeed score might move from 38 to 47. Google's field data and your actual user experience will still reflect a slow site.
Magento Speed — Common Questions
Does Magento 2.4.x have better performance than older versions?
Yes, but not as much as you'd hope if you're on Luma. Magento 2.4.x introduced improved asynchronous operations, better database indexing, and GraphQL support — but the Luma frontend JavaScript stack remained largely unchanged. The performance improvements in recent Magento versions are primarily on the backend (server side) rather than the frontend. A 2.4.7 store on Luma will still have the same JavaScript weight problem as a 2.3 store on Luma.
How much does a Hyvä migration actually cost?
It depends primarily on how many custom design elements and third-party extensions your current theme has that need to be rebuilt or integrated. For a store with a relatively standard Luma-based theme and a moderate extension set, a Hyvä migration typically costs between £8,000 and £25,000. Heavily customised stores with complex checkout flows, custom layered navigation, and many extension integrations can be higher.
Will Hyvä work with my existing extensions?
Most popular Magento extensions have Hyvä-compatible versions available, either from the original vendor or through the Hyvä Compatibility Repository. Before committing to a migration, it's worth auditing your extension stack against the compatibility list. Some extensions — particularly those with heavily customised frontend JavaScript — may need partial rebuilding for Hyvä. This is typically a known cost that can be scoped before the project starts.
My Google Search Console is showing Core Web Vitals failures — is this the same issue?
Almost certainly yes, if you're on Luma. Core Web Vitals failures in Search Console fall into two categories: field data (real user experience, measured by Chrome User Experience Report) and lab data (synthetic testing). LCP and INP failures on mobile in particular are strongly correlated with high JavaScript weight. If your Search Console is showing a "Poor" URL status for Core Web Vitals, this is an active SEO issue — not a future risk, but a current ranking penalty.
We have Redis configured. Should that help?
Redis improves Magento's session handling and full-page cache storage significantly, and is a strong recommendation for production Magento hosting. It reduces database load, improves cache hit rates, and makes cache invalidation faster. But it's a server-side optimisation — it reduces TTFB and server load. It has no impact on the JavaScript weight your users download and execute. Redis is necessary for a well-configured Magento server; it's not sufficient for good mobile performance on Luma.
What's the difference between PageSpeed Insights and Google Search Console's Core Web Vitals?
PageSpeed Insights shows both lab data (simulated performance, run in a controlled environment) and field data (real user experience from Chrome's user database, if enough data is available). Google Search Console shows field data only — your actual users' experience over the past 28 days. For SEO purposes, field data is what matters. Lab data is useful for diagnosing issues and testing fixes before they go live. A store can have decent lab scores (because it was tested on a fast connection) and still show Core Web Vitals failures in Search Console (because real mobile users on real 4G connections are having a slow experience).
Let's Find Out Exactly What's Slowing Your Store
We run a free Magento performance audit; PageSpeed scores across key pages, Core Web Vitals field data, server response analysis, JavaScript weight breakdown, and extension impact assessment. You'll know exactly what your problems are and what fixing them involves, before you commit to anything.
No junior developer running a Lighthouse report and calling it an audit. A proper diagnostic, explained in plain terms, with a clear recommendation on where to spend your effort first.