Why is my traffic down? find the cause before you fix anything

Confirm the drop is real before you act

The first job is proving something happened. Compare the same weekday range year over year rather than week over week, because weekly comparisons make every public holiday look like a penalty. Then check the size and the shape. A ten percent dip that recovers inside a fortnight is noise. A thirty percent fall that holds for three weeks is a signal. A cliff edge on one specific date points to a technical or algorithmic event. A slow slide over months points to competitors, content decay, or a shift in how people search for what you sell. Write the exact date the line changed direction before you look at anything else, because that date is the single most useful piece of evidence you have. Nearly every diagnosis that follows is a matter of asking what else happened on or near that date, and a fuzzy start date makes every later step guesswork.

Rule out tracking and reporting problems first

A surprising share of traffic emergencies are measurement failures. Someone deployed a new template without the analytics tag. A cookie consent banner started blocking the script by default. A filter, a property change, or a bot exclusion rule quietly rewrote the numbers. A subdomain moved and stopped reporting to the same property. Check whether other sources fell at the same moment. If direct, paid and referral all dropped on the same day, you have a tracking problem, not an SEO problem. Cross-check against a source your analytics cannot break: Google Search Console clicks and impressions, or your own server logs. Search Console counts clicks at the search engine, not in the browser, so it survives most tag failures. If Search Console clicks are flat while analytics collapsed, stop the SEO investigation and go and fix the tag. This check takes ten minutes and saves weeks of misdirected effort.

Narrow the drop to a segment

Sitewide drops and localised drops have completely different causes, so segment before you theorise. Break the loss down by landing page group, by country, by device, and by query type. If one directory lost everything and the rest of the site is stable, look at what changed in that directory: a template edit, a redirect, a noindex tag, a content rewrite. If mobile fell while desktop held, look at page experience and rendering. If one country fell, look at hreflang, local competitors, or a market-specific algorithm rollout. If branded queries are flat and non-branded queries collapsed, your brand is fine and your rankings are not. Segmentation converts a vague panic into a specific, testable question, and the specific question is usually answerable in an afternoon. Skipping this step is why so many teams rewrite content that was never the problem.

Check whether your pages are still indexed

Pages that are not indexed cannot rank, and deindexing is one of the fastest and most common causes of a sharp drop. Open the Pages report in Search Console and look at the indexed count against the trend line. A falling indexed count that matches your traffic curve is your answer. The usual culprits are a noindex tag pushed live with a redesign, a robots.txt rule that blocked a whole directory, canonical tags pointing at the wrong URL, a staging environment configuration copied into production, or a server returning errors to Googlebot while serving humans normally. Test a handful of affected URLs with the URL Inspection tool and read what Google actually sees rather than what you expect it to see. These faults are usually one line of code and are reversible within days of being fixed. Our Technical SEO Audit exists mostly to catch this class of problem before it costs a quarter of revenue.

Rule out an algorithm update

If your start date lines up with a broad core update or a spam update, the cause is probably external rescoring rather than a bug on your site. Check the date against Google's published update list and against the industry volatility trackers. Core updates do not punish specific errors; they reweigh what the system considers helpful, trustworthy and well sourced for a set of queries. That means the fix is rarely a technical switch. It is usually about depth, first-hand evidence, author credibility and whether your page genuinely answers the query better than the pages that replaced you. Recovery tends to arrive with a later update rather than immediately after the work, so plan in months. The trap here is a panic rewrite of everything. Identify the specific query clusters that lost ground, study what now ranks, and improve those pages with intent rather than redoing the entire site on instinct.

Audit what changed on your own site

Ask what your own team shipped in the two weeks before the drop. Redesigns, CMS migrations, URL structure changes, plugin updates, CDN configuration, new consent tooling and content pruning are all frequent causes. Redirect chains that lost a hop, internal links that vanished from a new navigation, title tags rewritten by a well-meaning template, pagination changes that orphaned deep pages: all of these cost traffic quietly. Pull a crawl and compare it to an older one if you have it. Compare titles, canonical tags, status codes and internal link counts. If a migration happened, treat this as the first hypothesis rather than the last, because migrations cause more preventable traffic loss than any other single event in SEO. This is also why migration support is worth buying before the launch rather than after the drop.

Look at the results page, not just your rank

Sometimes nothing about your site changed and your position did not move, yet clicks fell anyway. That happens when the results page itself changed. A new ad block, a shopping carousel, an expanded People Also Ask section, a video pack or a featured snippet can each push organic results below the fold and cut click-through rates while average position stays identical. The diagnostic pattern is distinctive: impressions steady, average position steady, clicks falling. Compare click-through rate by query in Search Console over the period. If CTR dropped while position held, your problem is presentation, not ranking. The response is different too. You improve the title and description so the listing earns the click, target queries where the SERP still favours organic listings, or accept the loss on that query and reallocate effort to keyword clusters with a cleaner results page.

Check whether AI answers absorbed the click

The newest cause on this list, and the one most often misdiagnosed. When Google shows an AI Overview, or when buyers ask ChatGPT and Perplexity instead of searching at all, the informational click never reaches you even though your ranking is intact. Two signals point this way: impressions holding while clicks fall on question-shaped queries, and a decline concentrated in top-of-funnel content rather than commercial pages. This is not recoverable through traditional ranking work, because the visit is not being lost to a competitor, it is being answered upstream. The response is to become the source the answer cites. That means quotable structure, clear entity signals and schema, which we cover in optimising for AI Overviews and measuring AI visibility. Start with an AEO Visibility Audit to find out whether you are being cited, mentioned, or ignored entirely.

Seasonality, competitors and the boring answers

Not every drop is a crisis. Demand itself moves. Accounting firms lose traffic after tax season, retailers after the holidays, B2B software every December. Check the same period last year and the year before, and check Google Trends for your head terms. If the whole category fell, your site is fine. Competitors are the other ordinary answer: someone published a better page, earned better links, or entered your category with a bigger budget. Look at who now occupies the positions you lost and read their pages honestly. Content decay is the third: pages with dates, statistics or product details that are now out of date slowly lose ground to fresher work. None of these are emergencies, but all of them need a plan, and understanding how SEO, GEO and AEO fit together helps you choose which plan to run.

Build a fix plan in priority order

Once you have a cause, sequence the work by cost and certainty. Fix measurement and indexing faults immediately, because they are cheap, provable and fast to reverse. Handle self-inflicted technical damage next: redirects, canonicals, templates and internal linking. Then address the SERP and AI layer, where better titles, structured data and quotable content change outcomes without changing rankings. Save content and authority work for last, since it costs the most and pays back over months rather than days. Set a review date and a single metric per fix so you know whether it worked, and resist stacking five changes at once, which makes attribution impossible. Traffic recovery is rarely one dramatic action. It is usually a correct diagnosis followed by three unglamorous fixes applied in the right order, then patience while the engines recrawl and rescore.

Sources

Google publishes a diagnostic path that matches the order used here: debugging Search traffic drops, the reference page on core updates and what they do, and the Search Console performance report documentation you will use to segment the loss.

Common questions