How to Fix Indexing Errors in Google Search Console
Two messages account for most of the confusion here. One means Google never came to look at the page. The other means it came, read it and decided not to keep it. Those need opposite responses. Telling them apart is most of the skill.
Not Every Reported State Is A Problem
A reader who sets out to fix every address in the not stored list has already lost. Many of those exclusions are correct, several are your own instructions being followed and clearing the list was never the objective. The job is finding the handful that matter.
What the list actually contains. Everything.
Every address Google knows about that it is not storing, including things you retired years ago, addresses generated by filters and pages you deliberately kept out.
Which ones matter. Two tests.
Is it a page you want stored, then is the category it sits in growing. Anything failing both tests can be left alone indefinitely without consequence.
Why this framing matters commercially. The list gets sold.
A four thousand line export makes an impressive attachment to a proposal. Most of those lines describe a site working as intended, which is worth establishing before anybody quotes for the work.
What this page covers. The two that matter.
The two most searched messages in the whole tool, what each one means and why they need opposite treatment. Then the categories to leave alone.
What a genuine problem looks like. Sudden or commercial.
A section disappearing from the stored list at once. Or a page you sell from sitting in the not stored list. Both are specific, both are traceable and both are worth an afternoon. A slowly accumulating pile of tag pages is neither.
Where the report itself is explained. Its own guide.
The page indexing report covers what the reporting shows and its naming history. This page is what to do about it.
Discovered But Not Indexed
Google knows the address exists and has not been to look at it. Nothing about the page has been assessed, because nothing about the page has been read. This is a queue rather than a verdict.
How Google found it without visiting. Several ways.
Your sitemap declared it, another page linked to it or it appeared somewhere else on the web. Discovery and fetching are separate events and this is the gap between them.
What it usually points at. Prioritisation.
Google gives every site a share of attention. A site publishing more addresses than that attention justifies will have a queue. On a large site this is normal and expected.
The second common cause. Nothing points at it.
A page reachable only through a sitemap, with no internal links from anywhere that matters, has been declared unimportant by your own site. Google reads that.
What to do about it. Make it easier to justify.
Link to it from pages that already perform, put it somewhere sensible in the navigation and reduce the number of low value addresses competing for the same attention.
The version that is genuinely urgent. A new section.
Where you have just published a set of pages and they are all sitting in this state, the site is telling you it cannot justify them yet. That is worth acting on quickly, because publishing more will make it worse.
What not to do. Request it repeatedly.
Asking moves one page up a queue once. It does nothing about why the queue exists, which the requesting guide covers.
Crawled But Not Indexed
Google fetched the page, read it and chose not to store it. That is a judgement about the content rather than a technical failure, which makes it the more uncomfortable message and by some distance the more useful one.
What it is not. A fault to repair.
Nothing is broken. The page loads, the instructions are fine and Google reached it without difficulty. It simply was not considered worth keeping.
What usually causes it. Three things.
Content too thin to say anything. Content near identical to another page on your site. Content that exists in a hundred better versions elsewhere on the web.
Where we see it most. Generated sets.
Town pages differing by a place name, service pages differing by a word and product pages differing by a size. Each one exists for a search term rather than for a reader.
What it is telling you. The page has no reason to exist.
Stated less politely than the report puts it. That is the content of the message. Something on the page has to be genuinely worth having, otherwise the page does not need to be there.
Why it feels unfair. The page cost money.
Somebody wrote it, somebody paid for it and now it is being told it was not worth keeping. That reaction is understandable and it is not an argument. The question is what the page offers a reader that nothing else does.
What to do about it. Improve, merge or remove.
Those are the three options and they are all real. Removing is not failure. It is frequently the correct answer, which block eight is about.
The Distinction Is The Diagnosis
One message is about reachability. The other is about worth. They look almost identical in a report and they need opposite responses, which is why treating them as the same message is the commonest mistake in this whole subject.
Reachability problems. Solved by structure.
Internal links, navigation, fewer competing addresses and a sitemap that means something. All of it is about making the page easier to get to and easier to justify visiting.
Worth problems. Solved by content.
Better pages, fewer near duplicates, something on the page that does not exist elsewhere. No amount of linking rescues a page that was read and turned down.
Why confusing them costs so much. The wrong work.
Businesses respond to a worth problem by adding internal links and submitting sitemaps, then conclude search is broken. The page was assessed and declined. Nothing they did addressed that.
The reverse mistake. Rewriting for nothing.
Equally, rewriting a page that was never read wastes the effort. Google has no opinion about it yet, so improving it changes nothing until somebody comes to look.
A quick rule of thumb. Read the verb.
Discovered means found. Crawled means read. If Google only found it, your job is structural. If Google read it, your job is editorial. That single distinction is worth more than most of what gets sold as technical work.
How to be sure which you have. Inspect the address.
The inspection tool tells you what Google holds for that page specifically, which settles it in seconds rather than by inference from an aggregate.
Blocked Or Excluded Deliberately
A substantial share of the not stored list is there because you asked for it. Those entries are the system doing what it was told. Working through them is the clearest example of effort producing nothing.
What belongs in this category. Plenty.
Basket and checkout pages, account areas, internal search results, thank you pages, filtered and sorted views, plus anything else that exists for visitors rather than for search.
Why excluding them is correct. They are not answers.
Nobody searching for anything wants your checkout page. Keeping them out of the index is a decision about what your site offers rather than a limitation.
When to look at this category. When something is in it by accident.
A commercial page carrying an instruction nobody meant to apply, usually left behind after a redesign or a staging environment going live. That does happen and it is worth checking once.
How that mistake looks. Sudden and total.
A whole section vanishing from the stored list at once, rather than pages declining individually. Where a site wide instruction went out by accident, the pattern is unmistakable.
How to check it quickly. Sample the list.
Look at a dozen addresses in this category rather than all of them. If they are all pages you would never want in search, the category is fine and you can stop looking.
Where the instructions themselves are covered. The technical material.
Which instruction to use, where each belongs and how they conflict is a subject of its own and our technical guidance owns it.
Duplicate And Canonical Messages
Where several addresses show similar content, Google keeps one and reports the others as versions of it. Most of the time that is exactly what should happen. Occasionally it tells you something you did not know about your own site.
When it is correct. You intended it.
Where you have pointed several addresses at a preferred version and Google agreed, the reporting is confirming your instruction. Nothing to do.
When it is worth reading. Google chose differently.
Where a different address was picked as the real one, your instruction was overridden or contradicted. That means two pages are more similar than you realised.
What that usually reveals. A near duplicate set.
Two service pages covering the same ground. Perhaps a set of location pages Google has collapsed into one. That is the same finding as block three arriving by a different route.
What to do about it. Differentiate or consolidate.
Give each page a genuine reason to exist or merge them into one better page. Both are legitimate and doing neither leaves the pages competing with each other.
What this costs when ignored. Your own pages competing.
Two near identical pages do not double your chances. They divide the references between them and let Google choose, which is a worse outcome than having decided yourself.
Where the mechanics live. The technical material.
How preference is declared, why it gets ignored and what to do about conflicting instructions all sit in our technical guidance rather than here.
Fix The Cause Then Validate
There is a facility for telling Google you have dealt with something and asking it to check again. It only works if something changed. Pressing it on an unchanged page is one of the more common pieces of theatre in this field.
What validation does. Rechecks the affected pages.
Google revisits a sample and confirms whether the problem has gone. If it has not, the process fails and you have spent the waiting time for nothing.
Why it gets pressed early. It looks like action.
Something has been done, a process is running and a monthly report can say so. That is why validation appears in reporting more often than the changes that would justify it.
The correct order. Three steps.
Establish the actual cause, change the thing responsible, then validate once and record the date. Skipping the middle step is the whole problem.
How long to expect. We will not say.
It depends on the site, the number of pages involved and how often Google visits. Anybody quoting a number of days for this is guessing.
What to do while waiting. Nothing to those pages.
Making a second change mid validation means you will never know which one worked. One change, then patience, which is the discipline the whole cluster keeps asking for.
When The Answer Is Fewer Pages
A site with large numbers of pages that Google has read and declined usually needs fewer and better pages rather than more attention paid to the ones it has. That is rarely what a business wants to hear and it is frequently the finding.
Why more pages stopped working. Volume is not scarce.
Publishing has become easy, so quantity is no longer a signal of effort or of substance. A page that says nothing new competes against everything else that says nothing new.
What the reporting is telling you. A verdict on the strategy.
Where a hundred pages were published and thirty were kept, the site has been assessed rather than misconfigured. That is information about the approach rather than about the setup.
What removing actually achieves. Concentration.
Fewer, stronger pages carry the internal links, the references and the attention that were previously spread across a set of near identical ones.
How to decide what goes. Look at the data.
Pages with no impressions over a long period, no clicks and nothing linking to them are candidates. Our material on pruning covers making that judgement properly rather than by instinct.
What we tell clients. Consolidate before you publish.
Where the indexing gap is widening, adding more pages makes the pattern worse. Everything sits on the Google Search Console guide.
Never fetched
is a queue.
Fetched and
dropped is a verdict.
They sit next to each other in the same report and they need completely different work. Adding internal links to a page that was read and turned down changes nothing. Rewriting a page nobody has visited changes nothing either. Establishing which you have takes seconds.
How we work through it:
If a four thousand line export was attached to somebody's proposal, ask how many of those pages you actually wanted stored.
Every guide.
One tool.
Getting set up and verified, using the reports properly, the performance and indexing data, finding what is broken, finding what is being passed over and turning any of it into work worth doing.
