Paste your Search Console performance export and this tool finds the exact day your traffic broke, then tells you whether you lost rankings or lost clicks. Those two look identical in a traffic graph and need completely different fixes, which is why so much recovery work goes in the wrong direction. The analyzer also matches your break date against every confirmed Google update since 2020. Your file is read in the browser and never uploaded.
Free tool
Search Console drop analyzer
Paste your Search Console performance export and the tool finds the exact day your traffic broke, works out whether you lost rankings or lost clicks, and matches the date against every confirmed Google update since 2020. Nothing is uploaded. The file is read in your browser and never leaves your machine.
- Open Search Console, then Performance, then Search results.
- Set the date range to the last 16 months and make sure the Date tab is selected.
- Click Export, choose CSV, and open the file called Dates.csv.
- Paste the contents below, or drop the file onto the box.
The analyzer smooths weekly seasonality with a seven day rolling average, then compares the 28 days before and after each candidate date to find the largest sustained change. It reads only the Date, Clicks, Impressions and Position columns.
What the analyzer is actually doing
Being transparent about the method matters, because the output is only trustworthy if you can judge whether the method suits your data.
- It smooths weekly seasonality with a centred seven day average. Why: Raw daily clicks swing hard between weekdays and weekends. Without smoothing, every Monday looks like a recovery and every Saturday looks like a drop. Centring the window keeps the detected date accurate rather than reporting it several days late.
- It compares the 28 days before and after every candidate date. Why: A real break is a sustained shift, not a bad week. Testing every day and keeping the largest sustained change finds the break rather than the noise.
- It reads clicks, impressions and average position together. Why: The relationship between those three is the diagnosis. Clicks alone cannot tell you whether you fell in the results or stayed put and lost the click.
- It requires an 8 percent sustained change before calling it a break. Why: Below that, normal fluctuation and small seasonal shifts produce false positives.
The four verdicts and what they mean
| Verdict | Measurement (the pattern behind it) | What it usually is | What it needs |
|---|---|---|---|
| Genuine ranking loss | Clicks, impressions and position all move against you | An algorithm update, technical fault or manual action | Diagnosis, then a narrow fix, then wait for reassessment |
| You kept rankings, lost clicks | Clicks fall, impressions hold, position does not worsen | An AI Overview or SERP feature taking the click | Format and query targeting, not quality work |
| Reporting change | Impressions fall, clicks hold, position improves | A Search Console measurement change | Nothing. Compare clicks, not impressions |
| Sustained decline, no break | Gradual slope with no step on any date | Content decay, lost links or competitors improving | Content refresh and competitive work |
The second row is the one worth dwelling on. It is the most common misdiagnosis I see in 2026, and the standard response to it makes things worse: rewriting a page for quality usually strips out the original details that earn an AI citation. The full pattern is covered in impressions stable, clicks falling.
The fourth row has no break date because there is nothing to match against an update. A slope means the loss accumulated over months rather than arriving on a day, which rules out an algorithm event and points at the causes in the seven causes of a drop with no update.
Getting the right export
- Open Search Console, then Performance, then Search results. Why: The Search results report is the one with the full 16 month history. Discover and News are separate reports with their own patterns.
- Set the date range to the last 16 months. Why: The analyzer needs at least 40 days and works better with a full year, because it compares 28 day windows on both sides of every candidate date.
- Switch on all four metrics before exporting. Why: Without impressions and average position, the tool can find the break but cannot tell you whether it was a ranking loss or click displacement.
- Export as CSV and open the file called Dates.csv. Why: The export contains several files. Queries and Pages are useful separately, but the Dates file is the daily time series this tool reads.
If you exported to Google Sheets instead, copy the Dates tab including its header row and paste that. Tab-separated data is handled the same way.
Reading the result carefully
Three situations where the headline verdict deserves a second look.
- Your site is seasonal. Why: The analyzer compares adjacent periods, not year on year. A genuine seasonal trough can register as a break. Check the same weeks last year before accepting it.
- You had more than one drop. Why: Only the largest sustained break is reported. If you suspect two events, run the tool again on a narrower date range covering the second one.
- A confirmed update matched your date. Why: That is correlation. Sites also drop during rollouts for technical reasons, and roughly one in five cases I audit turns out that way. Confirm with the 60 point audit before committing to a plan.
Frequently asked questions
Is my Search Console data uploaded anywhere?
No. The file is parsed in your browser using JavaScript and nothing is transmitted or stored. You can disconnect from the internet after loading the page and the analyzer will still work.
How much data does it need?
At least 40 days of daily rows, and preferably a full 16 month export. It compares 28 day windows either side of each candidate date, so a short range leaves little room to establish a baseline.
Why is the break date slightly different from what I expected?
The tool reports the day the sustained shift begins, which is often a day or two before most people notice it. During a rollout that runs 12 to 24 days, effects can also land on any day inside the window rather than on the announcement date.
Can it detect more than one drop?
It reports the largest sustained break in the range you give it. To find a second event, export or trim the data to a narrower window covering that period and run it again.
What if it says no significant break was found?
Either the drop predates your export, it is smaller than normal fluctuation, or the loss is in conversions rather than traffic. The analyzer will tell you if it sees a gradual decline instead of a step, which points at decay rather than an algorithm event.
Does it work with GA4 data?
Only partially. It can read any CSV with a date column and a clicks or sessions column, but without impressions and average position it cannot distinguish a ranking loss from click displacement, which is most of the value.
Got the break date but not the cause?
This tool works from your daily totals. A full diagnosis needs the page and query level data, which is where the scope of the loss becomes clear. Send your Search Console access and within 48 hours I will tell you which page groups were hit and what caused it. Free, no obligation.
Get free recovery audit