My Website Traffic Dropped: A Diagnostic Checklist
Before you diagnose a traffic drop, prove it happened. A good share of them are measurement faults, and a growing share are clicks disappearing while rankings hold perfectly steady.

Your website traffic dropped, and the instinct is to start looking for the cause or to call whoever handles your SEO. Resist it for twenty minutes, because a meaningful number of traffic drops are not drops at all, and the ones that are fall into five categories with completely different responses.
Work through these layers in order. Each one rules out a whole class of explanation.
Layer one: did it actually happen?
Before anything else, cross check two independent sources. Open your analytics and open Search Console and compare the same date range.
If analytics sessions fell but Search Console clicks held steady, you have a measurement fault, not a traffic fault. That changes everything about what you do next. The usual culprits:
- The tracking code stopped firing. A theme update, a plugin conflict or a tag manager container that got republished. Open your own site and check the realtime report. If you do not appear, nothing is being recorded.
- A consent or privacy setting changed. Consent handling has tightened and a misconfiguration can stop a large share of visits being recorded while the visits themselves continue happening.
- You are comparing across a reporting change. Any year on year comparison that spans a platform migration or a change in data retention is comparing two different measuring instruments. The drop is in the ruler.
- Duplicate or wrong property. Two tracking codes on one page, or a report pointed at a property that only covers part of the site.
- The data is not in yet. Analytics and Search Console both lag. A drop in the last two or three days is often nothing at all.
There is a second measurement issue almost nobody mentions, and it cuts the other way. A growing share of requests to any website now come from automated crawlers rather than people, and the volume has risen sharply. If your historical numbers included a lot of that, some of what looks like a decline is a filtering change or a crawler losing interest, and none of it was ever a customer. Worth knowing before you spend a quarter trying to win back traffic that was never human.
Layer two: has demand changed?
Compare against the same period last year, not against last month. Then check your branded searches separately from everything else in Search Console, using a query filter on your company name.
If non branded is flat and branded is down, something changed in how many people know about you, and that is not a search problem. If both are down by a similar amount in the same month last year as well, it is seasonal. If demand in your sector fell, no amount of optimisation brings it back and the correct response is a different one.
Layer three: rankings or clicks?
This is the layer that has changed most, and getting it wrong leads to months of work on the wrong problem. In Search Console, display clicks, impressions and average position on the same chart.
| What the three lines do | What it means | Where to go next |
|---|---|---|
| Impressions down, position down | You lost rankings | Layer four or five |
| Impressions steady, position steady, clicks down | You are still there and fewer people are clicking | Read the rest of this layer |
| Impressions up, clicks flat | Normal in 2026, often not a problem at all | Check your titles, then leave it |
| Everything down on one section only | Something specific broke there | Layer four |
| Everything down across the whole site at once | Core update or site wide quality judgement | Layer five |
If you are in the second row, go and run your main searches yourself and look at what is above you. Google now writes AI answers at the top of many results pages, and readers who get the answer there do not click. Pew Research Center measured in July 2025 that a search carrying one of these produced a click on a normal result around 8 per cent of the time against 15 per cent without. Amsive analysed roughly 700,000 keywords in April 2025 and found an average click rate decline of about 15 per cent where these summaries appeared, worse on non branded queries and worse again where a featured snippet sat alongside.
That is a real revenue problem and it is not a ranking failure. The response is different: commercial pages that still earn clicks, your own brand searches, and channels that do not depend on somebody choosing a link.
Layer four: did something break?
If rankings genuinely fell on specific pages or sections, work down this list. All of it is checkable in an afternoon.
- Manual Actions report. Thirty seconds, and it either rules out a whole category or redirects the entire investigation.
- Page Indexing report. A sudden rise in excluded or not indexed pages dates the problem for you.
- robots.txt and noindex. A deployment that carried a staging setting into production is depressingly common and removes pages within days.
- Redirects. After any redesign or migration, check that old addresses still land somewhere correct, in one hop, with the right status.
- Server health. If the site was slow or down when the crawler visited repeatedly, crawling drops and so does everything downstream.
- The sitemap. Still in place, still valid, still listing pages that return a normal response.
- Certificate and security. An expired certificate or a compromised site can remove you from results quickly and visibly.
- What changed on the page. Use an archive of your own site to see the previous version. Content lost in a redesign is a frequent and easily missed cause.
- Lost links. If pages that pointed at you were removed or redirected, the pages they supported can slip.
- Two of your own pages competing. Filter Search Console by the query, look at the Pages tab, and see whether the ranking page keeps changing. If so, consolidate.
Layer five: was it an update or an enforcement?
Find the exact date the decline started in Search Console, then check whether a Google update was running that week. A date match is a lead rather than a verdict, and the distinction between two situations matters a great deal.
A core update is not a penalty. Nothing was done wrong in the sense of a rule being broken. The whole site is being judged differently, and recovery takes substantial, broad improvement over six to twelve months. Tweaks do not work. Google folded its helpful content system into core ranking in March 2024, so there is no separate event to wait out any more.
An enforcement is different and two policies now matter more than the lists acknowledge. Scaled content abuse covers producing large volumes of pages with little added value, which is exactly the output of a cheap content programme. Site reputation abuse covers third party content hosted on your domain to borrow its standing, and the policy was tightened in late 2024 to apply whether or not you were involved in making it. If you host sponsored posts, coupon sections, or a subfolder somebody else runs, check there first. Neither of these shows up as a manual action in every case, which is why they get missed.
What to do in the first hour
- Cross check analytics against Search Console for the same dates. Measurement fault or real.
- Check Manual Actions.
- Pull sixteen months with clicks, impressions and position on one chart, and find the exact start date.
- Decide whether it is site wide or one section. That single fact halves the work.
- Run your three most valuable searches yourself and look at the whole screen, not just your position.
- Write down the date, the shape and the scope before you change anything. Fixing things in a panic destroys your ability to tell what worked.
Because the results page changed around you. AI summaries, ads, a map pack and other features can all take clicks without taking your position. This is now one of the most common traffic drops there is and it is frequently misdiagnosed as a ranking loss.
They measure different things. Search Console counts clicks from Google; analytics counts sessions that your tracking recorded, from everywhere. They will never agree exactly and that is normal. What matters is whether they agree about the direction.
A technical fault, days to a few weeks once fixed. A lost ranking to a specific competitor, a few months. A core update, six to twelve months with substantial improvement and no guarantee. A policy enforcement, as long as it takes to remove the cause and be reassessed.
Fix anything clearly broken, yes. Beyond that, no. Rewriting pages, changing titles and restructuring navigation all at once in week one leaves you unable to tell which change helped. Diagnose first, then change one thing at a time.
Want this checked on your own site?
Send the URL and we will tell you what we would do first, in writing, before anybody talks about a retainer.
- A written audit within 48 hours
- No call required and no obligation
- If search is the wrong spend for you, we will say so
Free site audit
Your URL and what you want more of. That is all we need to start.
Request received!
Thanks! We have your request. Check your inbox within 48 hours for your written proposal. Nothing else needed from you.



