How to recover SEO traffic you have lost

Correction, 23 September 2026: a matching update date does not rule out a technical fault. The canonical advice now also distinguishes duplicate pages from their preferred URL.

Start by finding out which pages lost the traffic and on what date. Every useful next step depends on those two facts, and guessing at them is how a recovery turns into six months of unrelated changes.

Open Search Console, set the date range to sixteen months, and compare the four weeks after the drop with the four weeks before it. Sort by clicks lost. Read the page list before you read anything else.

Use the shape of the drop to choose checks

The date and affected URLs help narrow the investigation. They do not prove a cause. Google’s diagnostic guide covers technical faults, ranking changes, security issues and changes in demand.

Start with these four checks.

Something went noindex. A staging flag shipped to production, a plugin setting flipped, a new robots rule blocked a folder. This is the first thing to rule out because it is the fastest to check and the fastest to fix. In Search Console, look at Pages, then at the excluded reasons. “Excluded by noindex tag” appearing on pages you expect to rank is your answer.

The URLs changed. A redesign, a new permalink structure, a migration between platforms. If the old URLs now 404, the rankings had nowhere to go. If they redirect to the homepage instead of to the matching page, Google treats that as a soft 404. The effect is close to the same.

A core update landed. Compare the drop with Google’s published update history. An overlap is a reason to investigate the update. It does not rule out a technical fault on your site. Check indexing and recent releases before deciding what to change.

Someone else got better. Search your own target queries. If two new results sit above you that did not exist last year, you are looking at a content problem. The work is to be more useful than those pages on the question the query asks.

Check the four fast things first

Before any content work, confirm the site is servable.

Fetch the page as Google sees it, using the URL Inspection tool, and read the rendered HTML. Check that the canonical names the intended preferred URL. Duplicate pages can correctly name another URL. Confirm robots.txt does not block the folder. Confirm the page returns 200 and not a 200 that renders an error message, which is the failure mode a status code cannot show you.

Then read the pages that lost the most

Take the top ten losing URLs and read them as a reader would. Three questions matter. Does the page answer the query it ranks for. Is the answer still true. Would anything on the page make someone leave.

Dates matter here. A page that says “the best tools for 2022” is being compared against one that says 2026, and the reader chooses before they read either.

What recovery actually looks like

Fixing a noindex flag can return traffic within days of the next crawl. Recovering from a core update takes a content cycle, and the change has to be real. Republishing the same article with a new date does nothing, because the date was never the problem.

Set the expectation before you start. If the cause is competitive or algorithmic, you are rebuilding a page’s usefulness, and that is measured in months.

One thing to avoid: making twelve changes in the same week. If the traffic returns you will not know which one did it, and you will carry all twelve into the next site.

If you want a second pair of eyes on a drop, the SEO work we do starts with the same diagnosis before anything gets changed.