Why Page Speed Affects Your Ecommerce Rankings
Speed is one factor among several. On a large catalogue it is also something else that almost nobody mentions: a slow shop is worked through more slowly, so parts of it are visited rarely and changes take longer to register.
What Speed Actually Does To Rankings
Speed is a ranking factor. It is not the ranking factor. A fast page with nothing useful on it will not rank. Overstating this is how shops end up fixing the wrong problem for months.
What it genuinely influences. Two things, in different ways.
It contributes directly, as one signal among many. And it affects whether somebody stays long enough to do anything, which feeds back into how the page is judged over time.
Where it sits against everything else. Below relevance, comfortably.
A slow page that answers the question beats a fast page that does not. Speed separates pages that are otherwise closely matched rather than lifting a page over a better one.
Why the overclaiming is so common. Because it is measurable.
Speed produces a number. Whether a category page helps somebody choose does not. Anything with a score attached attracts attention out of proportion to its weight. There is an industry selling improvements to that score.
Why it still deserves a page. Because of block two.
On a shop specifically, speed is not only a ranking question. It decides how much of a large catalogue gets looked at, which nothing else on this list does.
The boundary worth stating. This page covers speed as a ranking and crawling problem.
Speed as a build and conversion problem sits in our ecommerce web design guides. Same subject, different consequence.
The Crawl Problem At Catalogue Scale
A search engine spends a finite amount of effort working through any site. On a shop with thousands of pages, how fast those pages respond decides how many of them get seen.
The mechanism. Time, rather than page count.
The effort spent on a site is closer to a budget of time than a number of pages. A shop where every page takes twice as long gets roughly half as much of itself looked at in the same window.
What that means on a small site. Nothing much.
Twenty pages get worked through regardless. This is why speed advice written for ordinary websites never mentions it.
What it means on a large catalogue. Two compounding problems.
Deep pages are visited rarely. Products far from the home page may go long periods without being looked at, so they are judged on old information.
Changes take longer to register. Work done today is not seen until the page comes round again, which on a slow large shop can be weeks.
Why this changes the commercial case. It affects everything else you are paying for.
A shop that fixes its speed does not only rank slightly better. It gets more of itself considered. Every subsequent change lands faster. That makes speed work earlier in the sequence than its weight as a ranking factor alone would suggest, which our complete guide reflects.
What Actually Makes A Shop Slow
Four causes, listed in the order they usually contribute. That order is stable enough across the shops we look at to be worth treating as a starting assumption.
Images, by a distance. Product photography at the resolution it was shot, across a whole catalogue.
On most shops this outweighs everything else combined, which our image guide covers in full.
Third party scripts. Chat widgets, analytics, advertising tags, review displays, personalisation.
Each is small. Together they are frequently the second largest cost. Each one is somebody else's code running on your shop at whatever speed their servers manage.
Applications. Block four, because they behave differently from ordinary scripts.
The theme itself. Carrying features the shop does not use.
A theme built to suit many shops includes sliders, galleries and layout options you never enabled. That code frequently loads anyway.
What is usually not the problem. Hosting.
It is the first thing shops suspect and rarely the main cause on a hosted platform, where the server is not yours to blame. Worth checking, worth not assuming.
The App Problem
Applications are how a shop gains functions its platform does not provide. They are also the most common reason a shop that used to be fast is no longer fast.
How they accumulate. One at a time, each justified.
A reviews app, an upsell app, a stock notification app, a currency switcher. Every one of them was a reasonable decision on the day. Nobody ever decides to install fourteen.
Why they cost more than their function suggests. Each adds code to every page.
An application affecting one part of the shop frequently loads on all of it. A currency switcher used by a small proportion of visitors may be loading for every visitor on every page.
Why several compound rather than add. The part that surprises people.
Applications frequently load their own copies of shared components, so two doing similar things can each bring the same underlying code separately. They also compete for the same moment in the loading process, which is worse than the same weight spread out.
The one that catches everybody. Removing an application does not always remove its code.
Uninstalling frequently leaves fragments in the theme, added when it was installed and not removed when it went. A shop can be carrying weight from applications it stopped paying for years ago.
What to do about it. Audit rather than assume.
List what is installed, establish what each is genuinely used for, then check what remains from anything removed. On a shop nobody has examined this is one of the larger improvements available.
Measuring It Properly
Most speed measurement produces a number that is technically accurate and practically misleading. Three distinctions matter.
Laboratory against field data. The most important of the three.
A laboratory test loads your page once, on a simulated device and connection, under controlled conditions. Field data is what actual visitors experienced on their own devices and connections. They frequently disagree. Where they do, the field data is the one that describes reality.
Why a single score is a poor target. Because it compresses several things into one number.
Two shops with the same score can have entirely different problems. Chasing the number encourages fixing whatever moves it most cheaply rather than whatever affects visitors most. Those are not the same list.
Why testing the home page misleads. The specific mistake on a shop.
The home page is usually the most carefully built page on the site and the least commercially important. Your money is on category and product pages, which carry the grids, the filters and the photography that cause the problems.
What to test instead. A representative set.
A busy category page, a typical product page, one page deep in the catalogue. The basket. Measured on a mobile connection rather than on office broadband, per our mobile guide.
What To Fix First
Ranked by what each returns against what it costs to do. The order is unglamorous, which is part of why shops skip to the interesting work instead.
First: images. Largest effect, lowest difficulty, applies across the whole shop at once.
Sizes, formats and delivery are settings rather than a project. This is almost always the biggest available win and it is rarely the thing anybody is excited about.
Second: scripts and applications. Per blocks three and four.
Removing something unused is faster than optimising something needed. The audit that identifies it costs an afternoon.
Third: what loads when. Deferring what is not needed immediately.
Real gains, more care required. The place where a careless change can hide content from search entirely.
Fourth: the theme. Removing unused features or replacing it.
Substantial effect, substantial cost. The point at which this becomes a project rather than maintenance.
What sits below all four. Anything sophisticated.
Advanced techniques are real and they are the last few percent. A shop still serving full resolution photography has no business considering them. The fact that they are more interesting to discuss is exactly why they get discussed first.
Platform And Theme Limits
Some of this cannot be fixed without a rebuild. Knowing which situation you are in before spending is worth more than any individual improvement.
What a hosted platform decides for you. The layer beneath your shop.
Server response, caching and delivery are the platform's responsibility. You cannot improve them. On a mainstream platform they are rarely the constraint anyway. Our platform guide covers what each allows.
What a theme decides. Considerably more. It is the usual culprit.
How the page is constructed, what loads in what order, whether features you never use are still being carried. Some of that is adjustable. Some is built in.
How to tell which situation you are in. One question.
After images and applications are dealt with, is the shop acceptable? If yes, the theme was never the problem. If it is still slow with the obvious causes removed, the construction of the pages is the constraint.
What a rebuild is worth. Less than it appears, on its own.
Replacing a theme for speed alone is expensive and risks the layout and conversion work already done. It is usually right only when combined with something else you were doing anyway. Our migration guides cover protecting what the shop already earns.
Common Mistakes
Chasing a perfect score. The most expensive of the four.
The last stretch of any score costs disproportionately more than the first. It buys almost nothing a visitor notices. A shop that has gone from poor to good has captured nearly all the value available.
Installing an application to fix applications. The one with a certain irony.
A speed application adds its own code to every page while attempting to reduce other code. Sometimes it helps. Sometimes it produces a heavier shop with an extra subscription. Per block four, it is one more thing that may leave fragments behind.
Optimising the home page only. Per block five.
The page most likely to be tested and least likely to matter commercially.
Blaming speed for a ranking problem with another cause. The most consequential.
Speed is measurable, so it becomes the explanation for a decline nobody has diagnosed. Months go into improving a score while the actual cause, usually structural or a change nobody connected, continues untouched. Our diagnostics guide exists for exactly that situation.
What connects them. Each substitutes a measurable thing for a difficult one.
Half as fast means
half as much seen.
The effort spent working through a site is closer to a budget of time than a number of pages. On a large catalogue that means speed decides how much of your shop gets considered at all, not only how it ranks.
What is included every month:
£350 per month, one target area. No setup fee, nothing billed separately.
Twenty-two guides.
One subject.
This guide covers speed. The rest of the series covers the sequence, structure, category and product pages, technical health, measurement and everything a store owner has to decide.
