A tenth of a second: what site speed changes for an online store

The study most often quoted on site speed is Milliseconds Make Millions, which Deloitte and the agency 55 ran for Google at the end of 2019. They tracked mobile load times on 37 brand sites in Europe and the US, over more than 30 million sessions, for 30 days. None of the sites changed their design during the study, so changes in speed could be compared with what people did.

For retail sites, a 0.1 second improvement in mobile speed went with an 8.4 percent rise in conversions, and shoppers spent 9.2 percent more.

There is a condition that usually gets left out when people quote those numbers. The gains only showed up when every page on the way to checkout got faster. A fast homepage in front of a slow product page did not count.

That matches what we find when we look at slow stores. People rarely give up on the homepage. They give up on a filter that takes two seconds to respond, or a cart drawer that freezes while some app loads.

How Google measures site speed now

Google’s Core Web Vitals are three numbers taken from real visits to your site in Chrome.

• Largest Contentful Paint (LCP) is how long the main content takes to appear. Good is 2.5 seconds or less.
• Interaction to Next Paint (INP) is how long the page takes to react when someone taps, clicks or types. Good is 200 milliseconds or less, and anything over 500 is poor.
• Cumulative Layout Shift (CLS) is how much the page jumps around while it loads. Good is 0.1 or less.

INP replaced First Input Delay as a Core Web Vital in March 2024, and the change matters for stores. The old metric looked only at the first interaction. INP looks at the slowest one in the visit, ignoring one outlier for every 50 interactions. Size pickers, filters, quantity buttons and the add to cart button all count.

Google judges each number at the 75th percentile, so three out of four visits have to be within the limit. A fast office laptop tells you very little about that.

Where stores usually lose time

Apps and scripts. On Shopify, each app can add its own scripts to every page: reviews, upsells, chat, pop-ups, tracking. Each one is small. Together they decide how quickly the page reacts. Removing an app in the admin does not always remove its code from the theme, either.

Big first images. A 3 MB hero photo or an autoplaying video at the top of the page pushes LCP past the limit on a mobile connection. Images should be sized for the screen that shows them and served in a modern format such as WebP or AVIF.

Heavy filters and variant pickers. These run code on every tap. If that code redraws the whole product grid for each change, INP climbs.

Fonts and late banners. Fonts that load late and banners that slide in after the page appears move content around just as someone tries to tap it.

How to check your own store

1. Open PageSpeed Insights and test a product page, not only the homepage. Look at the section with data from real users. If the store has enough traffic, this is the data Google uses.
2. Try the store on a mid-range Android phone on mobile data. It is much closer to what many of your customers use than your own laptop is.
3. Go through the steps a buyer takes: a collection page, a product, the cart, the start of checkout. Write down every tap that feels slow.
4. List every app and third-party script on the store, and ask what each one earns. Keep the ones you can give a reason for.

Our Shopify team and website care service work on site speed page by page, starting with the slowest step to checkout.