Why Do Most Ecommerce Websites Fail at SEO?
Two different questions get asked here. Why a shop never ranked, then why a shop that used to rank stopped. They have different causes and different fixes. Treating them as one problem is how months get spent on the wrong thing.
Two Different Problems
A shop that never ranked has something wrong with it. A shop that used to rank and stopped had something happen to it. Those are not the same investigation and they do not share an answer.
Never ranked. The shop has been live for a year or more and has never received meaningful search traffic.
The cause is present in the site now and can be found by examining it. Blocks three to six cover where to look, in the order the causes usually sit.
Used to rank and stopped. Traffic was arriving and then it declined.
The cause is an event rather than a condition, so examining the site tells you less than establishing what changed and when. Blocks seven and eight cover that. The first question is whether the change was yours.
Why mixing them wastes so much time. Because the wrong investigation finds real problems.
Audit a shop that used to rank and you will find plenty wrong with it, since every shop has faults. None of them explains the decline, because they were all there while it was working. Months then go into fixing things that were never the cause.
How to tell which you are. One question.
Was there ever a period when this shop received search traffic it was pleased with? If yes, block seven. If no, block three.
The third possibility. Both, which is common.
A shop that ranked modestly, then declined after a change, now has an event to unpick and underlying conditions that were limiting it anyway. Deal with the event first, since that part is recoverable.
Why These Keep Happening
The same faults appear on shop after shop, built by different people on different platforms. That pattern is structural rather than a comment on anybody's competence.
Platforms default to what suits the platform. Reasonably.
A platform's defaults are set to make a shop work for the largest number of merchants with the least support burden. Making filters generate findable pages is easier to support than not doing so. Nothing about that decision was aimed at your search visibility.
Design agencies are judged on appearance. Because that is what a client can assess.
A shop is approved on how it looks and how it feels to use. Nobody signs off a build on how its category tree maps to demand, so nothing in the process rewards getting that right.
Nobody is asked about search until after launch. The one that causes the most damage.
Structure, addresses and how filters behave are all decided during a build. By the time anybody with a search brief is involved, those decisions have been made and are expensive to revisit, which is why our ecommerce web design guides exist separately.
What follows from all three. The faults are inherited rather than committed.
That matters practically as well as diplomatically. A store owner who believes they made a mistake looks for a decision they took. A store owner who understands these are defaults looks at the configuration, which is where the answer usually is.
Technical Failures
Five faults that stop a shop being found regardless of how good everything else is. They come first because nothing further down this page matters while one of these is present.
Indexing problems. Pages excluded from being found at all.
Directives left over from a staging site, sections excluded by a previous supplier, a setting nobody knew was on. The most common single cause of a shop that never ranked, also among the fastest to fix.
Crawl waste. Attention going somewhere other than your products.
Per our structure guide, this is almost always faceted navigation on a large catalogue.
Faceted pages consuming the catalogue. The severe version of the above.
Where generated combinations substantially outnumber real pages, the shop is effectively hiding itself behind itself.
Broken redirects. What happened to everything that used to exist.
Discontinued products, renamed categories and anything changed in a previous redesign. Each one is a page that earned something, now leading nowhere.
Pages nobody can reach. Products no internal link points to.
Common at scale and invisible from inside the business, since the product works perfectly if you have the link.
How to check the lot quickly. Compare pages that exist against pages that are indexed.
A large gap in either direction is the headline finding. Our technical guides cover the detail.
Structural Failures
Four failures that limit what a shop can achieve rather than preventing it outright. These are harder to see than technical faults, because nothing is broken.
A category tree inherited from a supplier. The largest of the four.
A tree organised by manufacturer and range, on a shop whose customers search by use, fit and material. Per our structure guide, no amount of content fixes this, because the pages target terms nobody searches.
Products too deep. Items several levels and several listing pages from anywhere.
They exist, they are reachable in principle. They are visited rarely enough that they are effectively not part of the shop.
No internal linking beyond the theme. Whatever the template produces, then nothing else.
That distributes evenly rather than deliberately, so nothing accumulates anywhere and the categories that matter are treated exactly like the ones that do not.
Categories that do not match how anybody searches. The version of the first failure that survives a rebuild.
A shop can rebuild its tree from scratch and still organise it the way the business thinks rather than the way customers look, which produces a tidier version of the same problem.
Why these are worth finding on a shop that never ranked. They explain a specific pattern.
Where the technical position is sound and the shop still receives almost nothing, structure is usually the answer, because the shop is competing for terms that do not exist.
Content Failures
Four content failures. Each leaves a shop with pages that work but give nobody a reason to prefer them.
Supplier copy everywhere. Identical text across every shop selling the item.
Per our guide to product descriptions, this is not a penalty. It means the choice of which page to show is made on everything except the words. Those factors favour whoever is larger.
Category pages that are bare grids. A heading and a list of products.
The most valuable pages on the shop, carrying the most commercial search, saying nothing. Our category page guide covers what they should do instead.
Nothing beyond the catalogue. No guides, no comparisons, no answers.
A shop publishing only products can only be found by people already looking for products, which is a ceiling rather than a fault.
No answer to any question a buyer asks before choosing. The one that connects the other three.
Somebody deciding between options has questions. A shop with supplier copy, bare categories and no guides answers none of them, so it competes on price and delivery alone.
Why content is listed fourth rather than first. Because it is where shops start.
Content is the visible work, so it is what gets commissioned. Done before blocks three and four it is money spent on pages that cannot be reached or cannot be found for anything.
The Duplication Family
Four faults that look separate and are one problem wearing four costumes. Treating them individually is why shops fix one and see nothing.
Variants. Size, colour and configuration as separate near identical pages.
Filters. Combinations generating pages that duplicate the category they came from.
Near identical products. Items differing in one specification, each with the same supplier description.
The same content across categories. A description or block of text repeated wherever a product appears.
The shared cause. A catalogue generating pages automatically, with nobody deciding which should exist.
Every one of these comes from the same place: a system producing pages from data, doing exactly what it was built to do, with no judgement applied about whether anybody was looking for the result.
What they do collectively. Divide the shop against itself.
Several pages competing for the same search means none of them holds it, while all of them consume attention that should be reaching pages that matter.
Why the grouping is practical rather than tidy. One decision resolves all four.
The question is the same in every case: does anybody search for this specifically? If yes, it deserves one proper page. If no, it should not exist as a findable page at all. Applying that once fixes the family. Fixing them one at a time does not.
Losing Rankings After A Change
Where a shop declined, the first question is whether anything changed. Self inflicted declines are the most common kind, the easiest to confirm and frequently the most recoverable.
Four changes that cause it. In order of how often we see them.
A migration or replatform without redirects. Every address changes and nothing points the old ones anywhere.
Everything the shop earned is discarded at once. Our migration guides cover doing this properly. This is the single most damaging avoidable event in ecommerce.
A redesign. Where the addresses survive and the content does not.
A cleaner design frequently means less text, so category pages lose the content that was earning their position while everybody agrees the site looks better.
Address changes. Renaming categories, restructuring the tree, tidying up.
Each one breaks an address that had accumulated something. These are frequently done in small batches nobody records.
A configuration change nobody connected. An application installed, a setting altered, a directive updated.
How to confirm this is what happened. Three checks.
Find the date the decline started and ask what happened that week, including anything a developer did without telling anybody. Check whether old addresses still resolve. And look at whether the drop was sudden, since a self inflicted change usually produces a cliff rather than a slide.
Losing Rankings Without A Change
Where nothing was done and traffic fell anyway, four explanations cover almost every case. Telling them apart matters, because three of them need no repair at all.
An algorithm update. The assessment changed rather than the shop.
Typically affects a whole site or a whole category of sites at once, over days rather than instantly, with other shops in your market moving at the same time.
A competitor improving. The most overlooked explanation.
Somebody else did the work in this guide. Your position fell because theirs rose, which shows as a gradual slide on specific terms rather than a site-wide drop.
Seasonality. Not a decline at all.
Per our seasonal guide, a shop comparing to last month in a seasonal business will read a collapse the calendar already decided.
Stock going out. Specific to shops and frequently missed entirely.
Products unavailable for a period take their pages with them. So do those discontinued entirely. A shop that cleared a range last quarter has lost whatever those pages were earning. Nobody connected the two because it felt like a stock decision rather than a website one.
How to tell them apart. Breadth, shape and comparison.
Site-wide and sudden suggests an update. Specific terms sliding suggests a competitor. Matching last year suggests seasonality. And a decline confined to one part of the catalogue points at what happened to that part.
Diagnosing Which One You Have
Six questions, in this order. Each one either produces an answer or sends you to the next. The sequence ends somewhere definite rather than in a list of possibilities.
One. Did this shop ever rank well? If no, go to question four. If yes, continue.
Two. Did the decline match a change you made? Find the date it started and establish what happened that week.
If something did, that is your answer and block seven is the repair.
Three. If nothing changed, what shape was the decline? Sudden and site-wide. Gradual on specific terms. Or the same as last year.
Per block eight, those three shapes point at three different explanations. Only the first two need any action.
Four. Is the shop actually eligible to be found? Pages that exist against pages that are indexed.
A large gap in either direction is your answer and block three is the repair.
Five. Do the pages you want to rank target terms anybody searches? Per block four.
If the category tree came from a supplier, this is your answer regardless of what else is wrong.
Six. Do those pages give anybody a reason to choose them? Per block five.
What you should have at the end. One cause rather than a list.
An underperforming shop is rarely broken in twenty ways. It is usually broken in two, with eighteen symptoms. The sequence above is designed to find the two.
What To Fix First
In this order, always, whatever the diagnosis. Each stage makes the next one work, so reversing the order wastes whatever comes first.
One. Indexing and crawl. Nothing else counts while pages cannot be found or attention is going elsewhere.
Also the fastest to show results, since it releases what the shop already earned rather than building anything.
Two. Structure. The ceiling everything else works under.
Frequently more of this can be fixed without changing addresses than shops expect, which our structure guide covers.
Three. Categories. The pages carrying the commercial search.
Four. Products. The ones that can pay, rather than all of them.
Our product page guide covers why most product pages are not a target.
The one exception to the order. A recent self inflicted decline.
Where block seven applies, repair that first regardless. Restoring what a shop had is faster and cheaper than building something new. Until it is done, nobody can tell what the underlying position actually is.
What we ask before any of it. Record what the shop currently earns.
Per our measurement guide, a baseline taken before work starts is what makes the next report a comparison rather than a fresh start.
Broken in two ways.
Eighteen symptoms.
Audit a shop that used to rank and you will find plenty wrong, because every shop has faults. None of them explains the decline, since they were all there while it was working. Finding the actual cause is the whole job.
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 diagnosis. The rest of the series covers the sequence, structure, category and product pages, technical health, measurement and everything a store owner has to decide.
