Online store speed: a page-by-page plan

A store has five kinds of page, and each needs different speed work. What the data says about store platforms, and a plan from category pages to checkout.

IntermediateHow-to7 min readUpdated October 30, 2026

In this guide
  1. Why stores are different
  2. What the data says about platforms
    1. Should you change platform for speed?
  3. The five kinds of page
    1. 1. Category and collection pages
    2. 2. Product pages
    3. 3. The cart
    4. 4. The checkout
    5. 5. Search
  4. Where the time goes on a typical store
  5. How to measure a store
  6. A priority order
  7. Questions for your developer or agency
  8. Mistakes to avoid
  9. Check your own page
  10. Sources

Why stores are different

A brochure site is a set of pages that look the same to every visitor. A store is a set of pages that change with the visitor: the basket, the prices, the stock, the address and the payment. That has two consequences for speed.

First, some pages cannot be cached the way others can. A shared copy of a cart page would show one shopper another's basket. Platforms handle this in different ways. For example, WooCommerce marks its cart, checkout and account pages as not cacheable, and Shopify renders its pages on its own servers.

Second, a store is heavy by nature. It has many pictures, many scripts for payments, reviews and tracking, and many plugins or apps. The 2025 Web Almanac's ecommerce chapter says that stores are unusually sensitive to performance because every extra second compounds: slower category pages reduce product views, slower product pages reduce add-to-carts, and slower checkouts reduce conversion. That is the chapter's reasoning about how a store works, not a measured result, but it matches how shopping flows run.

What the data says about platforms

The Almanac's chapter lists the share of stores that pass all three Core Web Vitals, using real-visitor data. On mobile, in 2025:

The weak point on the lower platforms was mostly Largest Contentful Paint, the loading measure. Interaction was close across the group, and Magento also did poorly on layout stability.

The Almanac groups platforms by who runs the servers. Shopify is software as a service, so the platform owns the code and the hosting. WooCommerce is self-hosted, so a merchant or an agency runs it. The chapter's conclusion is that control tends to give steadier results, but nothing is fixed in advance: disciplined self-hosted and customized stores can do well.

So the data supports two statements at once. Some platforms give a better starting point. And a store's own choices decide a great deal. See Shopify speed, WooCommerce speed and hosted site builders for each.

Should you change platform for speed?

Rarely, and not on this evidence alone. A platform change is a large project with risks to your orders and your search visibility. These are medians across many stores, and they say little about one store with careful work. First work through the plan below. Consider a move when the numbers stay poor after you have fixed what you control, and when other needs, such as features or costs, point the same way.

The five kinds of page

1. Category and collection pages

These pages show many products at once, so they are made of images. What decides their speed is the number of products shown, the size and format of the thumbnails, how many scripts the filters and sorting add, and whether the page is cached.

What to do:

2. Product pages

The main product image is often the Largest Contentful Paint element. The gallery tools, reviews, recommendations and variant pickers add weight around it.

What to do:

3. The cart

The cart is personal, so it depends on the server, plus the scripts for the mini-cart or cart drawer.

What to do: keep small any script for a mini-cart or cart drawer, since it can load on every page. Test adding to the cart on a phone, because that tap is where interaction speed shows.

4. The checkout

The checkout depends on the server, the number of fields, the payment tools and the recalculation of totals.

What to do: see WooCommerce checkout speed for how the classic checkout works and how to test it. The 2025 Almanac calls payments a major dependency surface, because of third-party scripts, redirects and fraud tools. Offer the payment methods that your customers use, and test every change with a real purchase.

Predictive search and filters run scripts as the shopper types.

What to do: make sure the search script loads without blocking the page, and that results come from a fast query. Shopify's reference theme, for example, loads its predictive search script with defer.

Where the time goes on a typical store

Across all five, the same four causes come up:

  1. Hosting and caching. The pages that can be cached must be, and the ones that cannot must be built quickly. See WordPress hosting and WordPress caching for WordPress stores.
  2. Images. Product photos are most of the weight. Right-sized, well-compressed, correctly prioritized images are the largest single gain on most stores.
  3. Scripts from other companies. Reviews, chat, pop-ups, loyalty tools, tracking pixels and payment tools. See third-party scripts.
  4. Plugins, apps and theme features. Each adds something to every page. Remove what you do not use.

How to measure a store

A priority order

If you do not know where to start, this order tends to pay off, and the reasoning is in the guides linked above:

  1. Make sure the cache works for the pages that can be cached, and that the cart and checkout are excluded.
  2. Fix the main image on product pages and the thumbnails on category pages.
  3. Review the scripts from other companies, and remove or delay the ones you can.
  4. Review plugins or apps.
  5. Trim the checkout and test it with a real purchase.
  6. Re-measure, and repeat for the page type that is still weakest.

Questions for your developer or agency

Mistakes to avoid

Check your own page

Run the free audit on a category page and on a product page. Open LCP request discovery, Improve image delivery, 3rd parties and Reduce unused JavaScript. Fix the largest problem on the weaker of the two, then repeat.

Sources

Start here

Free website audit

Live Google data for any public page: mobile and desktop.

Mobile + desktop · Lighthouse lab test · Real-user Chrome data when Google has it

Rather have this done for you? Speed optimization · Book a call

Related guides

All guides →