How to Read Your Rankings During a Google Update

When Google rolls out an update, your rank tracker suddenly looks alarming. Keywords jump, drop and jump back, sometimes within a day. The short answer: during a rollout, don't judge any single check. Compare against the week before the update began, and wait until Google says it's finished before you act.

Google's September 2026 spam update is a good example. It started on 24 September, applies globally and in every language, and Google said it may take up to two weeks. Search Engine Roundtable reported a first wave of drops a day or two later and a second wave around 30 September. If you checked your rankings on any one of those days, you saw a different story.

Start with Google's own record

Before reading anything into your numbers, confirm there actually is an update. Google lists every confirmed ranking update on its Search Status Dashboard, with:

• the update's name, such as "September 2026 spam update"
• the date and time it started
• a note when the rollout is complete

That page is the only source I treat as fact. SEO news sites are useful for spotting volatility early, but "people are seeing big swings" is not the same as Google confirming an update.

What to compare against

The most common mistake I see is comparing today against yesterday. During a rollout, both days sit inside the noise.

Use the week before the update as your baseline

Pick your last check from before the update started and compare against that. If you track weekly, that's simply the check before the start date. It shows where you stood before Google changed anything.

Look at many keywords, not one

One keyword moving tells you very little. Twenty moving in the same direction, in the same week, tells you something. I look at how many keywords moved more than five places, not at the biggest single drop.

For the brands I track myself, 33 keywords had a position both before this update started and after its first waves. 25 of them moved two places or less. One dropped five or more places, and three rose five or more. That is what a normal update looks like for most sites: a lot of small shuffling and a few real movers.

When a drop is worth acting on

I wait for three things before I change a page:

1. Google has marked the rollout complete. Until then, positions are still settling.
2. The new position has held for at least a week after that.
3. The drop shows up on several related keywords, not just one.

If all three are true, look at what replaced you. The pages now ranking above you tell you more than any guess about what the update targeted. A spam update, for example, mostly removes pages that break Google's spam policies. If your page dropped, it's often because the pages around it moved, not because yours was flagged.

What not to do mid-rollout

• Don't rewrite pages that dropped on day two. Many come back by the end of the rollout.
• Don't delete content in a panic. Removing a page is permanent; a ranking dip usually isn't.
• Don't disavow links on a hunch. A drop during a spam update is rarely fixed by cutting links you can't connect to it.
• Don't trust a single day's chart. Mark the update's dates on your chart, so later you remember why that week looked strange.

How I handle this in Presence Scout

I build Presence Scout, so this is how my own tool handles it. When Google confirms an update, a notice appears on the Search rankings pages while it rolls out, and each keyword's position chart shades the update's dates. Months later, the shaded band still explains a dip that would otherwise look like a problem.

Google updates also reach AI answers. AI Overviews and AI Mode draw on Google's results, so it is worth checking both together. I wrote about the difference between SEO, AEO and GEO and about tools that track Google AI Overviews if you want to go further.

The rule I follow is simple: during an update, watch, note the dates, and wait. Once Google says it's done and the dust has settled for a week, compare against where you started. That's when the numbers mean something.