Published
Google Update or Bad Data? How to Diagnose a Sudden Ranking Drop
In late September 2026, many SEO teams saw rankings fall across their tracked keywords. Google's Search Status Dashboard showed a spam update starting on September 24. At the same time, rank trackers were receiving poisoned SERPs and partial query results from Google. Some drops were real. Many were bad data.
Before you report a drop to a client or start rewriting pages, spend ten minutes on one question: did the rankings change, or did the data?
Step 1: Check whether an update was running
Google lists confirmed ranking updates on its Search Status Dashboard. Recent ones:
| Update | Started | Duration |
|---|---|---|
| September 2026 spam update | September 24, 2026 | About 14 days |
| August 2026 spam update | August 18, 2026 | About 3 days |
| June 2026 spam update | June 24, 2026 | About 2 days |
| May 2026 core update | May 21, 2026 | About 12 days |
| March 2026 core update | March 27, 2026 | About 12 days |
| March 2026 spam update | March 24, 2026 | Under 1 day |
| December 2025 core update | December 11, 2025 | About 18 days |
A drop during a confirmed update is more likely to be real, but not certain. September 2026 shows why: the spam update and the worst of the poisoning overlapped almost exactly.
Step 2: Look at how the drop is spread
Real updates and data problems leave different patterns:
| Signal | Real ranking change | Data problem |
|---|---|---|
| Next check | The new position holds | The old position comes back |
| Which keywords | Related keywords or pages move together | Unrelated keywords drop in the same collection run |
| Competitors | Some sites gain what you lost | Unfamiliar sites appear, often social or off topic |
| Search Console | Clicks and impressions move | Clicks and impressions stay steady |
| Ranking URL | Your URL moves or changes | The URL vanishes or shows as google.com |
The single most useful signal is the next check. Nozzle reports that most poisoning spells last one SERP, and that partial query results answered the full query again on the next check 61.8% of the time.
Step 3: Check the keyword again
Run the keyword again, a few minutes later if you can. If your tool allows it, use a fresh session or a different collection route. If the usual sites return, the first reading was bad data. Discard it rather than averaging it into your history.
Step 4: Look at the page the tracker saw
If your tool stores the full result list or the raw HTML, open it and compare it with the keyword's history:
- Are most of the usual domains missing? That points to a poisoned page.
- Do most results ignore part of the query? That points to a partial query page.
- Does the ranking domain show as google.com? The tracker may not be resolving Google's /goto links.
- Did the tracker check fewer pages than usual? A deep check that stopped early can make a page ranked 25th look like it dropped out of the top 30.
Step 5: Check Search Console
Search Console reports what real searchers saw. If clicks and impressions for the page are steady over the same days, the drop in your tracker is probably not real. If they fell too, look at which queries and pages lost them.
Keep two known Search Console shifts in mind when comparing dates. In September 2025, impressions fell for many sites after Google removed num=100. And Search Console data arrives with a delay of a day or more, so very recent days can look low.
Step 6: Compare with what you see
As a last check, search the keyword yourself using the tracker's location and device. Our guide on why tracked rankings don't match what you see in Google lists what to match so the comparison is fair.
Building this into your tracking
If you run your own rank tracking on a SERP API, most of these checks can run automatically:
- Keep each keyword's usual domains from recent checks.
- Flag any check where fewer than half of them appear, or where most results ignore part of the query.
- Fetch flagged checks again before saving them.
- Store the location, device, check time, and request ID with every observation, so you can audit a strange result later.
- Mark confirmed Google updates on your charts.
Our rank tracker build guide shows the storage side, and the guide to silent scraper failures covers monitoring data quality in general.
Frequently asked questions
How do I know if a ranking drop is from a Google update?
Check Google's Search Status Dashboard for an update running at the time, then confirm that the new position holds on later checks and that Search Console shows a matching change in clicks and impressions. A drop that disappears on the next check is usually a data problem.
Why did my rankings drop in late September 2026?
Google's September 2026 spam update ran from September 24 to about October 8. At the same time, Google was returning poisoned and partial query results to many automated rank checks. Some drops from that period were real, and many were bad data. Compare with Search Console to tell them apart.
Should I trust Search Console or my rank tracker?
For trends over days or weeks, Search Console is the better reference, because it reflects what real searchers saw. A rank tracker is better for daily positions at a specific location and device, as long as its data is checked for poisoned and partial results.
What are signs of bad rank tracking data?
Unrelated keywords dropping in the same collection run, unfamiliar or unrelated sites appearing in the results, a ranking domain shown as google.com, and positions that return to normal on the next check.