How to Find Pages Losing Rankings in Google Search Console
A page that stops working entirely gets noticed within a week. A page that slides gently does not, so by the time somebody looks properly the damage is months old. This is about catching that earlier, then ruling out the dull explanations before the dramatic ones.
Decline Is Quieter Than Failure
Nobody misses a page that disappears. What gets missed is the page that loses a tenth of its visibility every month for a year, because no single month looks like a problem. That is how businesses end up wondering where the enquiries went and finding the answer began eighteen months ago.
Why the pattern hides. Every month looks normal.
A modest fall reads as an ordinary fluctuation. It is only when you compare against something distant that the direction becomes visible. Almost nobody does that.
What makes it worse. The tool forgets.
This reporting holds a limited history, so the comparison that would reveal a long slide is the one that stops being available. Businesses discover the problem exactly when they can no longer see its beginning.
Why it matters commercially. Compounding.
A page losing visibility slowly is losing enquiries slowly, every month, in a way that never triggers a phone call to anybody. The cost accumulates quietly.
What this page is for. Two things.
Finding it in your own data, then working through the causes in an order that does not waste your time. Block six is that order and updates come last in it.
What it is not for. Blaming an update.
That is the conclusion everybody reaches first and the least likely to be right. Our guide to telling whether a drop came from an update covers that question properly.
Comparing Two Periods Properly
Everything here depends on a comparison. A badly built one manufactures declines that never happened. Getting this right takes a minute and saves you from investigating arithmetic.
Equal lengths. Non-negotiable.
Three months against the previous three months. Never one calendar month against another, since the difference in days alone can produce a fall that looks like a finding.
One country. Applied first.
Before you read a single figure. Our performance report guide covers why unfiltered figures blend your market with noise.
Page level, then query level. In that order.
Find which pages fell, then look at which searches they lost. The second step tells you whether you lost a position or lost a whole subject.
Look at clicks and impressions together. Both matter.
Impressions falling means you are appearing less. Clicks falling with impressions steady means something changed about how your result is chosen rather than where it ranks.
Exclude the last few days. They are incomplete.
The most recent data is still being counted, so including it in a comparison makes the current period look worse than it is. That alone has convinced plenty of businesses they were in trouble.
Long enough to be real. Three months minimum.
A fortnight against a fortnight will show you noise and let you interpret it as a trend. That is how false emergencies begin.
Concentrated Or General
Before anything else, establish whether the fall is confined to a few pages or spread across the site. Those two shapes have completely different causes and mixing them up sends the investigation in the wrong direction immediately.
Concentrated. A few pages or one section.
Almost always traceable. Something happened to those pages, to their addresses or to the subject they cover. The cause is usually findable in an afternoon.
General. Everything, roughly evenly.
Points at something site wide, something technical or something external. A redesign, a migration, a server problem or a broad change in how the results page looks.
Why the distinction matters. It selects the questions.
A concentrated fall asks what changed about those pages. A general fall asks what changed about the site or the market. Nobody should be asking both at once.
The version people misread most. One big page falling.
Where a single high traffic page slides, the site total falls and it looks general. Reading at page level rather than in aggregate prevents that mistake.
What to do with a general fall. Check the site first.
Recent releases, redesigns, plugin changes and hosting moves, in that order. Something you did is considerably more likely than something Google did.
Rule Out The Boring Explanations
Most reported declines are not declines. They are artefacts of how the comparison was built. Or predictable seasonal movement. Or the consequence of something the business did itself. Ruling those out first is what separates diagnosis from storytelling.
The comparison itself. Check it twice.
Unequal periods, a seasonal boundary in the middle, an unfiltered country view or a date range including a partial week. Any of those produces a fall from nothing.
Seasonality. Your own trade.
A roofer in July and a roofer in November are different businesses. If the same dip happened last year, it is your market rather than your website.
Something you changed. The most common real cause.
A redesign, a page retired, a menu simplified, an address changed without a redirect or a developer tidying something up. Ask what was released in that period before looking anywhere else.
A reporting delay. It happens.
This reporting has been late by weeks before now. Stale figures can look exactly like a fall, so check whether the data is current before reacting to it.
What is left after all that. A genuine finding.
Whatever survives those four checks is worth investigating properly. Only then is an external cause a reasonable hypothesis.
What A Real Decline Looks Like
A genuine decline has a shape. It persists across several comparison periods, it shows in impressions rather than only in clicks and it survives every check in the previous block. Recognising the shape stops you chasing fluctuations.
It persists. Across periods.
Down this quarter, down last quarter, down on the same quarter a year ago. One period down is weather and three is a climate.
Impressions fell too. The key signal.
If impressions fell, you are appearing less often, which is a ranking or a relevance change. If only clicks fell, the ranking held and something else changed.
The queries narrowed. A useful clue.
A page appearing for forty searches last year and twelve now has lost coverage of a subject rather than a place in a list. That is a bigger finding than a position slipping.
It is not uniform. Usually.
Real declines rarely affect everything equally. Some searches hold and others vanish. The pattern in what went tells you what the cause was interested in.
What position tells you here. Less than you think.
We use it as supporting evidence rather than as the measure. Our material on measuring performance explains why that figure is the most misread of the four.
Common Causes, In Order
Five causes, ordered by frequency rather than by drama. Note where updates sit, which is at the end rather than the beginning.
The content went out of date. The commonest by far.
The page was written four years ago, the subject moved and somebody else has written the current version. Nothing was done to you. The page simply stopped being the best answer.
Somebody else improved. Ordinary competition.
A competitor published something better or rewrote what they had. Positions are relative, so you can decline without changing at all.
Something changed on your site. Self inflicted.
A redesign that removed content, internal links lost in a menu rebuild, a migration handled badly or a page slowly edited into something vaguer than it was.
The results page changed. Around you.
More appearing above the ordinary results means the same position delivers fewer visits. Real, commercially painful and not a ranking loss.
A change in ranking systems. Last, plus rarely provable.
Possible and hard to establish. Most changes are never announced, so no named update means nothing either way. How often Google changes things covers that. the updates guide owns the subject.
Catch It Early Or Not At All
The whole problem with gradual decline is that it is cheap to fix early and expensive to fix late. A page slipping for two months can be updated in an afternoon. A page that has been slipping for two years has usually lost the subject entirely.
Export monthly. The only real defence.
Because the tool holds a limited history, your own records are the only way to see a long trend. This is the habit in this whole cluster with a deadline attached, since the data you do not save is gone.
Look quarterly. Not weekly.
Weekly looking produces false alarms and encourages reacting to noise. A proper comparison every three months catches everything that matters in time.
Keep a list of what changed. Dates included.
Releases, redesigns, new pages, retired pages. Half of diagnosing a decline is knowing what happened and when. Nobody remembers a year later.
Watch your best pages specifically. Not just the total.
Pick the handful of pages that actually generate enquiries and track those individually. A site total can look stable while the pages that matter are sliding.
Where to go next. The other finding.
If impressions held and clicks fell, pages being passed over is the guide you want. Everything sits on the Google Search Console guide.
Most declines
are a comparison,
a season or
something you did.
An update is the conclusion everybody reaches first and the least likely to be right. Working through the dull explanations takes an hour, costs nothing and settles most cases, which is considerably better than paying for a recovery plan aimed at the wrong problem.
The order we work in:
If a supplier explains your decline by naming a dated update without evidence, remember that most changes are never announced at all.
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.
