Mobile site speed testing on a cheap Android over a cellular connection will tell you more about your store than any dashboard, because that device is your median mobile customer. You experience your store on a flagship phone, on office WiFi, with every image already cached from the hundred times you have opened it this week. That experience is real for exactly one visitor: you.

Meanwhile the shopper who found you on Instagram is holding a $150 Android, three years old, on two bars of LTE in a parking lot. Budget hardware means slower processors, and processing is where modern storefronts hurt: every script your theme and apps ship gets parsed and executed by that modest chip. A page that feels instant on your phone can take three times as long to become tappable on theirs, and they do not know your store is “usually faster.” They only know this tab is not loading and the back button is right there.

The ritual, once a month

• Get the device. Buy a sub-$200 Android, used is fine, and keep it in a drawer as your store’s crash-test dummy. It may be the highest-ROI hardware your business owns.
• Turn off WiFi. Cellular only, ideally somewhere with mediocre signal. Clear the browser cache so you load the store the way a first-time visitor does, cold.
• Shop for real. Homepage, a collection, search for something, product page, add to cart, reach checkout. Feel where you wait, where you mis-tap, where the layout jumps as late images shove the button you were aiming at.
• No spare phone this week? PageSpeed Insights already simulates a mid-tier device for its mobile score, and Chrome DevTools can throttle the network and CPU to approximate the same pain. Useful, and still not a substitute for thumb-on-glass reality.

The lab numbers and the LCP work from tip 4 remain your instruments. But instruments are for measuring; the drawer phone is for believing. Nothing reprioritizes a roadmap like watching your own checkout take eleven seconds in your own hands.

Fifty-nine days out is the moment for this test, because whatever you find still has time to get fixed, and Cyber Monday traffic skews harder toward mobile, first-time visitors, and strained networks than any other day of your year. Your median customer is about to get even more median. Meet them where they shop.

A Shopify app bloat audit is the cheapest speed win available to most stores, because the average store is carrying apps the way an attic carries boxes: installed for a campaign in 2024, forgotten, and still shipping their JavaScript to every visitor on every page load. The countdown timer from a sale that ended. The quiz nobody finished building. The loyalty widget you evaluated for a week. Each one loads its scripts whether or not it renders a single pixel.

Here is the hot take: you do not have an app problem, you have a breakup problem. Merchants hoard apps “just in case” the way people keep exercise bikes as coat racks. Meanwhile every hoarded app is a tax paid by every visitor, and mobile visitors, your majority, pay it on hardware and networks with no slack to spare.

Run the audit this week

• List every installed app and write one sentence next to each: what it does for a customer or for you, today. If the sentence starts with “we might”, it fails.
• Uninstall, do not just disable. Toggling an app’s widget off in the theme editor often leaves its scripts loading. Only uninstalling removes the app’s ride-along, and even then, check your theme code: apps installed before Shopify’s app-embed era often left snippets in theme.liquid that keep loading forever. Search your theme for the app’s name before you close the ticket.
• Measure before and after. Run your homepage and a product page through PageSpeed Insights before the purge and after. The before/after is your receipt, and it is usually satisfying.
• Replace app-jobs with theme-jobs where you can. A countdown, a badge, an announcement bar: many single-widget apps do what twenty lines of theme code do, minus the third-party request chain.

The rule going forward: every app pays rent. Rent is measured in a job it demonstrably does, reviewed quarterly. This series will happily recommend apps that earn their slot, but the bar is exactly that: earning it. An app that is not making you money is making you slower, and slower costs money on every session. Evict accordingly.

Largest Contentful Paint (LCP) measures how long your biggest visible element takes to appear, and on most Shopify stores that element is the hero image. Which means most stores do not have a “site speed problem.” They have a one-image problem wearing a site speed costume.

This matters because slow paint is not a technical footnote, it is a conversion tax. A shopper who taps your ad and stares at a blank rectangle is deciding, right then, whether you are worth waiting for. Google’s own guidance calls for LCP under 2.5 seconds, and mobile shoppers on cellular are less patient than that number suggests.

Here is the hot take: do not buy a speed app yet. The speed app industry sells you minification, lazy loading, and preloading magic while a 4MB PNG exported straight from Photoshop sits at the top of your home page doing 80 percent of the damage. Fix the one asset first. It is free, it takes an afternoon, and it is usually the whole story.

The afternoon fix

• Find the culprit. Run your home page through PageSpeed Insights and look at the mobile score. It names the LCP element explicitly. Nine times out of ten, it is the hero.
• Weigh it. Aim for under 200KB for a full-width hero. If yours is over a megabyte, congratulations, you just found your speed strategy.
• Compress and convert. Drag the original into Squoosh, export as WebP, and eyeball the quality slider until just before you can tell the difference. Shopify’s CDN helps with resizing, but it cannot rescue a bloated source file with a huge intrinsic size.
• Right-size it. A 4000-pixel-wide image serving a 390-pixel-wide phone screen is pure waste. Upload something sane and let the theme’s srcset do its job.
• Do not lazy-load the hero. Lazy loading is for images below the fold. Lazy-loading your LCP element is asking the browser to please be slower.

Then re-run PageSpeed Insights and watch the LCP number drop. Repeat for your top collection pages and your bestseller product pages, since each template has its own biggest element. Total cost: zero dollars and one afternoon. We will get to image formats across the whole catalog later in the series; today is just the one image doing the most harm.