Why Google Rankings Suddenly Dropped

Updated September 2026 · Written and maintained by the Progression Agency strategy team

A ranking drop is a change with a date attached, and the fastest route to an answer is to establish the shape of the change before theorizing about causes. This page works through the diagnosis in elimination order — your own deployments first, because self-inflicted causes outnumber penalties — then index status, algorithm updates, lost links, search results pages that changed around you, seasonality, manual actions and competitor improvement, with what not to do while you are still looking.

Establish what actually changed before explaining why

A ranking drop is a change, and changes have dates. Almost every wasted week in this situation comes from theorizing about causes before establishing precisely when it happened and what moved.

Three questions settle most of it. Did rankings fall for everything, or for specific pages? Did they fall on one day or decline over weeks? And did traffic fall by the same proportion as rankings, or did one move without the other?

Those answers point in different directions. A single-day drop across the whole site suggests an algorithm update or a site-wide technical fault. A gradual decline on specific pages suggests competitors improving or content going stale. Rankings holding while traffic falls usually means something changed in the search results page rather than in your position.

Rule out your own deployments first, because they are the most common cause

Before considering anything Google did, establish what you did. Self-inflicted ranking losses are more frequent than algorithm penalties and are usually fixable in an afternoon.

The recurring ones are mundane. A staging site pushed live with its robots directives intact, blocking crawling entirely. A noindex tag left on a template. A redesign that changed every URL without redirects. A plugin update that broke rendering for crawlers while looking fine to you. Internal links removed during a navigation cleanup. Content trimmed during a redesign because it looked long.

Get the date of the drop, then look at what shipped in the week before it. Version control, deployment logs and plugin update histories answer this faster than any analysis of search trends.

Check index status before anything else

If pages have left the index, ranking analysis is pointless. The index coverage report will say whether Google can see and is choosing to keep your pages, and will usually name the reason.

Verify the pages render for a crawler

A page that requires JavaScript to display its content may render perfectly for you and return nearly nothing to a crawler. Fetch the raw HTML and confirm the words are actually in it.

Confirm redirects survived

URL changes without redirects are among the most damaging and most reversible causes. Every old URL should return a permanent redirect to its closest equivalent, not to the homepage.

Look for accidental blocking

A robots directive or a meta tag from a staging environment is the single most common cause of a total, sudden disappearance.

Algorithm updates are real, but they are over-diagnosed

Google makes frequent changes, several of which are announced and named. When a drop coincides exactly with a confirmed update, that is meaningful evidence. When it does not, attributing it to an update is usually a way of avoiding a harder answer.

The distinguishing feature of an update-driven drop is that it affects a category of content rather than individual pages, and that it happens to competitors too. If everyone in your space moved on the same day, the cause is external. If only you moved, it is not.

There is no fast recovery from an update, and anyone offering one is describing a tactic that will cause a second problem. Updates change what the system rewards; the response is to improve against the new standard, and recovery typically arrives at a subsequent update rather than immediately.

Losing links is invisible until you look for it

Rankings supported by links decline when those links disappear, and nothing notifies you. Sites are redesigned, articles are unpublished, directories close, and pages that linked to you quietly stop existing.

This produces a gradual decline rather than a cliff, which makes it easy to misattribute to content or to an update. The check is direct: compare your referring domains now against six and twelve months ago, and look specifically at whether the pages that linked to your most affected URLs still do.

Recovering a lost link is often possible and rarely attempted. A site that redesigned and broke the link will usually restore it when told, particularly if the page still references you in the text.

Check whether the result page changed rather than your position

It is now common for a page to hold its ranking and lose most of its traffic. This is not a ranking problem and treating it as one produces no improvement.

The causes are structural changes to the search results page. An AI-generated summary answering the question directly. More ads above the results. A features block, a map pack, or a video carousel pushing the first organic result well down the page. A competitor winning a rich result that draws the eye.

The way to tell is to compare impressions against clicks over the same period. Impressions steady with clicks falling means your position held and the page around you changed. That reframes the problem: the response is to target queries where a summary cannot substitute for a visit, and to compete for the features now occupying the space.

Seasonality, and other declines that are not declines

Some drops are arithmetic. A business whose demand is seasonal will see traffic fall every year at the same point, and month-over-month comparison will present that as failure.

Compare against the same period a year earlier before concluding anything. If the year-over-year figure is flat or up while month-over-month is down, nothing is wrong. This single check eliminates a meaningful share of ranking panics.

The other false alarm is measurement. Rank trackers report position from a specified location and device, and changing either changes the number. Personalization, a different data center, or a tracker that silently switched defaults will all produce apparent movement that no visitor experiences.

Compare year over year, not month over month

Seasonal businesses will otherwise diagnose a problem every year at the same time.

Check the tracker’s location and device settings

A default change produces a drop that exists only in the report.

Look at a range of positions rather than one keyword

Single-keyword movement is noise; a distribution shift is signal.

Verify in an incognito window from the relevant location

Personalized results differ from what the tracker reports and from what you normally see.

Manual actions are rare, obvious, and worth ruling out

A manual action is a human decision by a reviewer, and it is reported directly in Search Console. It is the one cause you can confirm rather than infer, which is why it is worth checking early even though it is uncommon.

The usual triggers are unnatural links, thin or scraped content, and cloaking. If you inherited a site or worked with an agency whose methods you could not see, this is more likely than it would otherwise be.

Recovery requires fixing the underlying issue and submitting a reconsideration request that describes what was wrong and what changed. Partial fixes are routinely rejected, and the review takes weeks.

Competitors improving is the least dramatic and most common gradual cause

If nothing broke, no update landed, no links were lost and no penalty exists, the remaining explanation is usually that someone else got better.

This is diagnosable by looking rather than theorizing. Take the queries that fell, look at who now occupies the positions you held, and compare directly: what does their page cover that yours does not, how recently was it updated, what links to it, and does it match the query’s intent more precisely than yours.

The uncomfortable version of this finding is that your page was never especially good and was ranking because the alternatives were worse. That is a normal state of affairs and it corrects itself the moment a competitor invests.

What not to do while diagnosing

The instinct to act quickly causes more damage than the original drop in a meaningful number of cases. Several common responses make recovery harder.

Do not rewrite or delete pages before you know why they fell — you will lose whatever signal remains and remove the ability to compare. Do not disavow links reflexively; the tool is for confirmed manual actions and can cause harm otherwise. Do not redesign the site as a response. Do not buy links to compensate. And do not change many things simultaneously, because you will never learn which one mattered.

The correct posture is to fix confirmed faults immediately, and to change one substantive thing at a time thereafter, allowing enough time between changes to attribute the effect.

How long recovery takes, honestly

Recovery time depends entirely on cause, and any single figure quoted without that qualification is meaningless.

A technical fault — a blocking directive, a broken redirect — typically recovers within days to a few weeks of being fixed, as pages are recrawled. Lost links recover as replacements accumulate, which is months. Competitor displacement recovers when your page is genuinely better, which is also months. Algorithm updates frequently do not recover until a subsequent update reassesses the site, which can be a quarter or more.

The practical implication is that the diagnosis determines the timeline, so it is worth spending a day getting it right rather than a month acting on a guess.

Redirects, feeds, and why rankings move

Three unrelated causes of the same symptom.

A 301 moved permanently response tells a search engine that a URL has moved for good and that the new location should inherit its ranking signals. It is the single most important instruction in a site migration, and the two ways it goes wrong are chains, where one redirect points to another that points to a third, and blanket redirects that send every old URL to the homepage rather than to its closest equivalent. The second is the most common cause of a redesign losing traffic.

The FYP algorithm — the recommendation system behind a for-you feed — works differently from search in a way that matters for planning: it predicts what a given viewer will watch and be satisfied by, using their history and the video’s performance with similar viewers. That means a business account’s reach is decided per video rather than by follower count, which is why consistency of subject matters more than consistency of schedule.

Rankings that drop across a whole site rather than on specific pages usually indicate a site-level cause — a technical change, a manual action, or a link profile problem — while rankings that drop on individual pages usually indicate that something better was published. The diagnostic order is: check for a manual action in Search Console, check whether anything changed technically, then look at what now ranks.

Reference videos

Search and measurement fundamentals relevant to the diagnostics above.

Frequently asked questions