Your traffic fell overnight: how to work out why
A sequence for isolating a sudden traffic drop so you can tell a broken tag from a lost ranking before anyone starts changing the website.
The chart dropped and nobody changed anything
Yesterday the analytics chart looked normal. Today it is down by half. The first three people you ask all say nothing was changed, which is true as far as each of them knows.
A sudden fall has far fewer possible causes than a slow decline. That is the good news. The discipline is to identify the cause before you react, because the most expensive responses in this situation are the fast ones. Rewriting pages to recover from what turns out to be a broken tracking tag is a familiar and avoidable waste.
What this costs you and what it does not fix
This process does not restore lost rankings. It tells you what happened. Recovery is a separate decision you make once you know whether anything was actually lost.
No paid tool is required. Analytics, Search Console and your hosting or server logs carry almost everything you need. The expense is calendar time and nerve. Expect half a day of a competent person's attention and a few days of holding the line while others ask for action. Resist deploying changes during the diagnosis. Each change you make removes a variable you were relying on to explain the drop.
When to investigate and when to wait
Investigate immediately when revenue or enquiry volume moved with the traffic. That combination means the loss is real and something is reaching fewer people.
Wait when the traffic chart fell and the enquiries did not. That pattern points at measurement rather than at demand and a day of calm comparison costs you nothing. Wait too if the drop lands on a predictable Indian calendar event. Traffic to B2B sites falls during Diwali week and around major festivals and long weekends every year. A business that compares this week against last week rather than against the same week last year will discover a crisis every October. Do not start an investigation at all if a single low-value page lost traffic while the rest of the site held steady. That is normal movement and chasing it burns a week.
Isolating the cause in order
First confirm the drop is real. Open Search Console and compare clicks for the same period. If Search Console shows normal clicks while analytics shows a collapse then your measurement broke. Check whether a tag was removed in a deployment, whether consent banner settings changed, and whether a filter or a bot exclusion was added to the analytics property.
If the drop is real then segment it. Split by channel first. Organic only, paid only, direct only and all channels at once are four different stories. All channels at once points at the site itself so check uptime, server errors and whether a page template broke. Paid only usually means a card declined, a budget exhausted, or a campaign disapproved. Direct and referral only often means a large referring site changed a link.
If the loss is organic then split by device, by country and by page. A drop on mobile alone points at rendering or speed. A drop confined to one country points at geo-blocking or a redirect rule. A drop on one template points at a change to that template. Then check the obvious self-inflicted causes. A robots.txt disallow, a noindex pushed from staging, an expired certificate, a migration redirect chain, and a CDN rule are responsible far more often than any algorithm update. Only after all of that is clean should you consider a ranking or algorithm change, and you confirm that by looking at positions for your main queries rather than by reading commentary.
Judging the recovery honestly
The metric that matters is clicks to your commercial pages for the queries you held before the drop. Pull a before and after comparison in Search Console at query level. That shows whether you lost visibility or lost measurement.
The metric that flatters is total sessions. Sessions recover when a tag is reinstated or when a bot resumes crawling and neither of those is a business recovery. Average position flatters too because it improves when poorly performing queries stop appearing at all.
The common mistake
The expensive mistake is acting in the first hour. Pages are rewritten, redirects are added, titles are changed and new content is commissioned before anyone has established what changed. When the real cause emerges there is no clean baseline left and you now own a second set of changes to unwind.
There is a second version of the same mistake. Treating every drop as an algorithm update is comforting because it places the fault outside the organisation. It also guarantees nobody checks the deployment log where the answer usually sits.
Write it down before you close the file
When you find the cause record the date, the symptom, the actual cause and the person who confirmed it. Traffic drops repeat and they repeat for the same handful of reasons in the same organisations. A one page history turns the next incident from a week of debate into an hour of checking. Keep a note of planned deployments beside it so the next investigation starts with a list of suspects rather than with three people who are each certain nothing was changed.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.