Why your Shopify speed score and your real shoppers disagree

You paste the tote page into PageSpeed and get a red number

You open the product page for your sand canvas tote, copy its address, and paste it into PageSpeed Insights, Google’s free checker for a web page. A moment later it shows one big number out of 100, labelled Performance, inside a circle that is orange or red.

That number is the same kind of grade Shopify’s old speed score was built from. It grades one pretend visit to the tote page, made by a Google computer, not by any of your shoppers.

Your shoppers have a number of their own: how long they waited for the tote photo to appear on their phone. It sits in a separate part of that same report, and in your Shopify admin. The two can disagree, and when they do, trust the shoppers.

So the red circle is a clue about the tote page. It is not the verdict on it.

What the big number grades, and what Shopify counts instead

OK, so let’s start with what Google did with your tote page. It loaded it once, on a pretend mid-range phone over a mobile connection, and timed how the page came together on screen.

Google’s page on PageSpeed Insights describes that as a simulated load on a single device with a fixed set of network conditions. It turns those timings into the grade, and 90 or above shows up green.

That pretend visit is what people call a lab test, and the tool that runs it is called Lighthouse. So when someone compares a Lighthouse score with real user speed on Shopify, they are making the same comparison you are making with the tote.

Shopify’s own speed score used to be built the same way. Shopify’s article on the speed score says it was a weighted average of Lighthouse scores for your homepage, your busiest product page, and your busiest collection page, and that Lighthouse only measures a snapshot in a simulated lab setting.

That same article says the Web Performance dashboard replaced the single score. The dashboard counts something different.

Every time someone opens the tote page in Chrome, Edge, Samsung Internet, or Firefox, Shopify notes when the biggest thing on their screen showed up. On a product page, that is usually the tote photo.

Shopify’s help page on web performance says each of these scores comes from real visitors, not simulated tests, and that good means the main content is on screen within 2.5 seconds.

That wait has a name, Largest Contentful Paint, or LCP, and numbers from real visits like these are what people call field data. Shopify reads it at the 75th percentile, which means three out of four visits to the tote saw the photo at least that fast.

PageSpeed has a real-shopper part too, and it is easy to miss next to the big circle. Google’s page on PageSpeed Insights says it covers Chrome users who agreed to share their data over the previous 28 days.

If the tote has too few visits, that part falls back to the whole store or shows nothing. Shopify counts more browsers than Chrome, which is why Shopify’s help page warns that its numbers can differ from PageSpeed’s. Same tote, a slightly different crowd of visitors.

Why a green score does not mean the tote arrived sooner

Here’s the catch. The grade and the photo time come from two different kinds of visit, so one can move while the other stays put.

Google’s page on PageSpeed Insights says it plainly: good lab data does not necessarily mean real shoppers’ experiences will also be good. It runs the other way too.

The pretend phone is one mid-range model on one mobile network, and your shoppers are on whatever they own, wherever they are. A red circle can sit on a tote page that real shoppers see quickly, and a green one can sit on a page where they wait.

The circle grades one visit Google made up. The photo time is every visit your shoppers made.

I know what you’re thinking. Google made PageSpeed, so surely Google ranks the tote on that circle. Shopify’s article on the speed score says Google ranks on the Chrome visitors’ data, the real-shopper part of that same report, not on the circle. Shopify’s number counts Firefox and more browsers as well, so when those two real-shopper numbers differ, trust the photo time in your admin.

That gives you permission to ignore a few things. The circle changing when you run the tote twice is normal, because Google’s page says the network and the test computer vary from run to run.

A red circle on a tote page where real shoppers already get the photo in time is not a fire. And chasing 90 for its own sake is the trap Shopify’s article warns about, where the score goes up and the page gets no better for shoppers.

What you still have to get right is the photo for real shoppers. Keep the lab report anyway, because Google says lab data is the useful kind for finding a problem, and the fixes listed under the circle are where you look if the real photo is late.

Read both numbers on the tote page, then decide

Here is the check, in the order you would click. You need the tote’s address, PageSpeed, and your Shopify admin.

1. Open the live tote page from a normal link, not the theme editor, and copy the address from the browser bar.

2. Paste it into PageSpeed Insights, leave it on Mobile, and note the big Performance number. Do not act on it yet.

3. In the same report, find the part about your real users and look for Largest Contentful Paint. If it says there is not enough data, that is normal for one product on a smaller store, so go straight to Shopify.

4. In your Shopify admin, go to Analytics, then Reports, search for contentful, and open Largest Contentful Paint: Page URL, as Shopify’s guide to the reports describes. Filter the device type to mobile and find the tote’s address in the list.

Now you have both numbers for the tote: the pretend visit, and what your shoppers waited for.

So does Whoosh turn the red circle green? A little, and that is the wrong reason to install it.

On the stores we measure, the product page on a phone had a median pretend-visit score of 56 with Whoosh on and 43 with it off. Those are two different middles, though, and when you line each store up against itself, the typical store’s own gap is 3.5 points.

The homepage on a computer is the same trap, where 73.5 against 61 looks like a 12-point jump while the typical store’s own gap is 3.

The photo barely moved at all. In our real-visit csv, the median store’s photo time at three out of four visits was about 1.87 seconds with Whoosh on and about 1.88 seconds off, and both were already inside the good band.

So the lab number moved a little, and the photo time for real shoppers barely did. Do not install Whoosh to turn a red PageSpeed number green.

Whoosh does have a job when the tote photo is late for real shoppers because a chat, review, or popup app starts first. Smart App Loading sequences those other apps so they do not race the tote photo, and they still load once the photo is up.

Smart App Loading is US$99 a month on the Shopify App Store, with a 7-day free trial.

You can stop once Shopify says real shoppers on phones see the tote within 2.5 seconds, the good line on Shopify’s help page, whatever colour the circle is.

If the photo is late for them, the next suspect is what holds it up. There are two places to look: an app nobody uses any more whose program still arrives first, which is why a Shopify store gets slow, or a photo file bigger than the phone needs, which is the copy of the photo your theme asks for.

Whatever you change, give it a week, then read both numbers on the tote page again. Shopify’s reports can run up to 36 hours behind, so set the date range to the days since your change and let the shoppers’ number catch up with the circle.