If you have spent the last few weeks staring at a traffic graph that makes no sense, you are not imagining it. Google’s September 2026 spam update began rolling out on 24 September and did not finish until 8 October, roughly two weeks, against the couple of days its earlier spam updates needed. Ranking volatility has not properly settled since. In mid October the rank trackers lit up again, forums filled with reports of vanishing Discover traffic and collapsing conversions, and Google confirmed nothing.
This post pulls it together in one place: what happened and when, what the volatility tools showed, what site owners were saying, and what is actually worth doing about it. It reflects what was known at 21 October 2026.
The short version: a confirmed, unusually long, global spam update ran from 24 September to 8 October, with three distinct waves of impact. Unconfirmed movement before it and again after it has made the picture messy. Check your own data against the dates below before changing anything.
Summary of Contents
What Happened, and When
Google only confirmed one of the events below, but the volatility around it matters, because it is easy to blame the wrong thing when several movements overlap. All dates are 2026.
Date | What happened | Status |
4 to 13 September | Irregular ranking movement. 4 September was noticeably more volatile than the days around it. Some large sites dropped around 3 and 4 September, then reversed sharply on 13 September. | Not confirmed by Google |
14 to 15 September | Volatility heated up again. UK site owners reported mass movement and traffic drops from the afternoon of 14 September, with pages for established keywords swapped for different pages entirely. | Not confirmed |
23 to 24 September | Tracking tools spiked in the day before the announcement. Google did not tie this earlier movement to the spam update. | Not confirmed |
24 September | Google announces the September 2026 spam update, applying globally and to all languages, and says it may take up to two weeks. | Confirmed |
25 to 27 September | First wave of impact. Friday saw early drops, Saturday 26 September was the peak, and things began to calm on Sunday. | Impact observed during confirmed rollout |
30 September | Second wave. Site owners reported large drops, and Discover and News traffic went missing again. | Impact observed during confirmed rollout |
4 to 6 October | Third wave, widely seen as the final major movement. | Impact observed during confirmed rollout |
8 October | Google confirms the rollout is complete, after about 14 days. | Confirmed |
14 to 19 October | Fresh volatility and a surge in forum chatter, including over the weekend. Some sites reported hours with almost no traffic. Reports of rising manual actions continued. | Not confirmed |
What Made This Spam Update Different
Google described it as a normal spam update, but several details were not normal.
- It was long. This was the fourth spam update of 2026, and the earlier ones finished within a few days. The August update took about two days. This one took about two weeks, which is why it arrived as separate waves rather than one sharp event.
- It was global. It applied to all languages and locations at once, so there was no region that was safe because it was not the US.
- It was not about links. Google indicated it was not aimed at link spam, and did not announce any change to its spam policies. It also did not say what share of queries it affected.
- The guidance was sober. Google’s advice was to review its spam policies and check you comply. It also said recovery can take many months, and that it will run periodic refreshes rather than recovering sites instantly.
Google has not said what the update targeted, so treat what follows as observation rather than fact. Several experienced SEO analysts saw early drops among highly templated, programmatic pages, such as one near identical page per area code or phone number, and among large sites publishing AI generated content at scale.
One analyst reported big drops across many sectors, site types, and countries by 28 September. Commentators also noted that spam updates now seem to hit harder than older ones, and some compared the impact to a core update, though it is not clear that comparison holds.
Measuring the Volatility: What the Tools Show
When rankings move, the first question is whether it was you or everyone. Volatility trackers answer that. The best known is the Semrush Sensor, which monitors daily changes in search results for a fixed set of keywords and boils them into a single score from 0 to 10.
Sensor score | What it suggests |
0 to 2 | Low volatility. Rankings are likely stable. |
2 to 5 | Normal volatility. |
5 to 8 | High volatility. Rankings may fluctuate and positions could be gained or lost. |
8 to 10 | Very high volatility. A high chance of significant ranking changes. |
Semrush Sensor, SERP volatility across all categories, 30 days to 20 October 2026.
The chart above, covering the 30 days to 20 October, tells the story of this update clearly. Volatility sat in the low range until about 23 September, then shot up in the following two days into the very high band, above 8, reaching roughly 9 or higher. It stayed there through the end of September, with a brief dip to around 7 on 2 October, then climbed back to very high for the third wave from 4 to 8 October. From about 9 October it settled into the high range, between roughly 7.5 and 8, before easing to around 5.6 in the middle of the month. It then rebounded to about 7 over the weekend of 18 October, and stood at 6.1 on 20 October, still in the high range, which Semrush describes as position changes for numerous sites. These are approximate readings taken from the chart, not exact figures, and they cover all categories rather than any single industry.
Three things make the Sensor more useful than a headline number. It covers more than 20 categories, so you can compare your own industry against the average, which matters because updates rarely hit every sector equally. It covers mobile as well as desktop. And it has a UK database, so you can check whether the movement you are seeing is visible in the market you actually sell into.
Other trackers are worth checking alongside it, including Mozcast, AccuRanker’s Grump, Sistrix, Advanced Web Ranking, Wincher, Serpstat, Algoroo, Mangools SERPWatcher, CognitiveSEO, Similarweb, Zutrix Tension, and DataForSEO. Each uses a different keyword set and method, so their scores will not match. What you are looking for is agreement. When most of them spike on the same day, something real happened. The three waves described above, 25 to 27 September, 30 September, and 4 to 6 October, were identified exactly this way.
Two cautions. First, a volatility score tells you the search results moved. It does not tell you your site was affected, and your own Search Console data is the only real test of that. Second, in mid October several commentators warned that rank trackers were struggling and were less reliable than usual, which is part of why the latest movement is hard to read. If you want to understand why third party data has limits, our post on whether Google is limiting third party tools from crawling beyond the first page is a useful background read.
What Site Owners Were Saying
The mood on SEO forums, particularly WebmasterWorld, and in discussions on LinkedIn and X was mixed, and it moved with the waves. The themes below come mainly from forum posts, so treat them as anecdote. People whose traffic fell are far more likely to post than people whose traffic held.
- Large traffic drops. Site owners reported falls ranging from around 20 percent to 80 percent in search traffic, with one news site reporting revenue down by roughly four fifths. Others reported flat or normal traffic, and some saw recoveries between waves.
- Discover and News disruption. This was the most consistent complaint. Discover traffic appeared and vanished within the same day, spiked for a few hours then faded, or disappeared for days. Publishers also reported stale stories in News and Top Stories, including content months old.
- Conversions collapsing. Several people saw normal looking traffic but almost no sales or leads, with one suspecting bot traffic and linking the change to the spam update. That is a hypothesis, not a finding, but it is a reminder to check conversion quality as well as volume.
- Quality complaints. Posters described spam on page one, weak AI Overview answers, and low quality AI translated pages in Google News. Analysts also noted AI translated pages from a large forum platform surging again in several European countries.
- UK specific movement. In mid September, UK site owners described mass movement of their own and results that looked poor, which is worth knowing if the UK is your main market.
One more point of context. Industry commentators noted that general SEO chatter had been unusually calm earlier in September, and wondered whether that reflected waning interest as search clicks decline. That calm did not last.
Is It Over?
Officially, yes. Google said the rollout was complete on 8 October. Unofficially, the movement in mid October, with tools spiking, Discover and News volatility, and weekend reports of sites losing traffic for hours at a time, suggests something is still shifting. Some in the community expect a core update to follow, and unconfirmed updates have in the past sometimes preceded one. That is speculation, and nobody outside Google knows. Commentators also reported a continuing rise in manual actions, which are penalties applied by people rather than algorithms, so check for those separately.
The practical point is that you should not treat 8 October as a clean finish line. Any analysis of your own data should allow for movement before 24 September and after 8 October.
What You Can Do
The right response depends on whether you were hit. Work through these in order.
- Check whether you were actually affected. In Search Console, compare clicks and impressions across the windows that matter: before 24 September, 25 to 27 September, 30 September, 4 to 6 October, and since 8 October. Look at the UK separately from other countries, and look at Discover and News separately from web search, because they behave differently and are often confused.
- Find out where the change happened. Segment by page type, folder, and template. A drop across your whole site means something different from a drop confined to one set of pages. If the damage is concentrated in a single template, that is your lead.
- Audit against the spam policies. Google told site owners to review them, so do exactly that. Be honest about templated pages built mainly to capture searches, content produced at scale with little added value, however it was produced, third party content hosted on your site to borrow its authority, and old domains repurposed to inherit ranking power. Also check Search Console for manual actions.
- If you were hit, fix the cause and be patient. Improve, consolidate, or remove the thin or near duplicate pages rather than tweaking them. Google has said recovery can take many months and depends on later refreshes, so a lack of movement in the first few weeks does not mean the fix has failed. Resist the urge to change things every few days.
- If you were not hit, treat it as a warning. An update that punishes scaled, templated content is a good prompt to look at your own. Pages that survived this round are not guaranteed to survive the next. Content that has quietly decayed is also worth checking, because it is often the same pages.
- Fix your measurement before you need it. Annotate your analytics with update dates, filter bot traffic, and make sure conversion tracking works. Good reporting also looks beyond rankings.
What Not to Do
- Do not rewrite or delete large parts of your site in a panic before you know where the problem is.
- Do not trust a single tracker. One chart spiking is noise. Several agreeing is a signal.
- Do not assume a drop was the spam update because the dates roughly line up. Seasonality, tracking errors, and unconfirmed movement either side of it can all look the same.
- Do not chase Discover traffic as if it were stable. This update showed again how quickly it can disappear and return.
If You Operate in More Than One Market
The update was global and covered every language, so international sites should check performance country by country rather than relying on a combined figure. In particular, be honest about any translated content produced at scale. Machine translation is not a problem in itself, but pages translated in bulk with no local adaptation sit uncomfortably close to the templated, scaled content this update appeared to target.
Our guides to translation versus localisation and international SEO in 2026 explain what good looks like, and our signs your international SEO is underperforming is a good first diagnostic.
If you would rather have specialists work out whether this update hit you and what to do about it, our SEO agency team can do exactly that, separating real impact from noise across every market you operate in. So, if your traffic changed between 24 September and now and you are not sure why, then get in touch today to have our team diagnose what happened and prioritise the fixes.
Facing a Challenge?
Get help from international SEO professionals. We can make your website visible when matters most. A new game plan to boost your online presence when your potential customers need you.
Get Started
SEO for SaaS: The SEO strategies for SaaS companies you should be using

