Why is my website losing traffic?

Short answer

It depends on the shape of the loss. A cliff on a date you can name is usually something that changed: a deploy, a migration, an update. A slow slope over a year is usually something that did not change while everything around it did. Date it and segment it before you fix anything.

When a site owner asks me this, the first thing I want to see is not the site. It is the graph. Sixteen months of Search Console clicks and impressions, by week. The shape of the loss tells you more about the cause than anything on the pages themselves.

A cliff or a slope

A cliff is a drop of thirty per cent or more inside a week or two, on a date you can point to. Cliffs have causes you can usually find: a site release, a redirect map that went wrong, a robots.txt change, a noindex left on from staging, a canonical pointing at the wrong place, or a Google update rolling out that week. I have written up the first six things I check when traffic drops overnight, because the order matters and the boring causes are the common ones.

A slope is harder. Traffic down ten or fifteen per cent a quarter, for a year, with no single date to blame. Nobody panics, so nobody investigates, and by the time someone asks the question the site has lost half its search traffic. Slopes are almost never one thing. They are the site standing still while competitors improve, the search results changing shape, and content ageing out.

Rule out the things that are not losses

Before you diagnose anything, check that the loss is real.

Seasonality. Compare year on year, not month on month. A dip every November is a pattern, not a problem. I wrote about telling a seasonal dip from a real problem.
Tracking. If analytics shows a drop and Search Console does not, the tracking changed, not the traffic. Consent banners and tag changes cause this constantly.
Branded versus non-branded. If branded search fell, that is a demand problem, not an SEO one. Separate the two before going further.

The external causes

Some of what looks like a lost ranking is fewer clicks on the same ranking. SparkToro’s analysis of US Google searches found that 68.01% ended without a click between January and April 2026, up from 60.45% in 2024, with AI Overviews appearing on more than a fifth of searches. If your impressions are flat and your clicks are falling, you may be holding your position on a results page that simply sends fewer people anywhere.

Then there are Google’s updates. Google says that several times a year it makes significant, broad changes to its ranking systems, with smaller unannounced ones in between. If your drop lines up with a confirmed update, that is a strong lead. It is still a trigger rather than a cause: the update changed how something already true about your site was weighted.

Segment before you guess

A sitewide drop and a drop in one section are different problems. Split the loss by page type, by directory, by query group and by country. Ask which pages lost the most clicks and whether they share a template.

When the loss is concentrated in one template, the cause is usually technical or structural and fixable. When it is spread evenly across the whole site, it is more often a quality or authority judgement, which is slower to reverse.

Look at which queries moved, too. If you lost informational queries and kept commercial ones, the results pages for those queries may have changed shape. If you lost the commercial ones, that is more urgent.

The causes I find most often

Across the recovery work I have done, the honest ranking is not what most articles suggest. In order:

Something the site did to itself. A redesign, a migration, a plugin change, a content cleanup that removed the wrong pages.
A results page that changed. More ads, a map pack, AI answers, or aggregators moving in above you.
Competitors that got better. Not glamorous, very common.
An algorithm update. Real, but less often the whole story than people assume.
A penalty. Rare, and Google tells you in Search Console when it is a manual one.

What not to do

Do not start changing things before you know the cause. Every change you make while diagnosing muddies the evidence, and you lose the ability to tell what helped.

Do not delete content in a panic. Pruning removes the wrong pages more often than the right ones.

And do not assume it will come back on its own. Sometimes it does. Usually the slope continues.

How long it takes to find

For a cliff, the cause is usually identifiable within a day or two of looking at the right data, because something specific happened on a specific date. For a slope, it can take a week of segmenting before a pattern shows, and the answer is often several smaller causes rather than one. Either way, the investigation is cheap compared with a quarter spent fixing the wrong thing.

If you can name roughly when it started, my SEO recovery work starts by dating and segmenting the loss, then establishing the cause before anything gets touched. If the loss is not recoverable, I’ll tell you that rather than bill you to chase it.