Data follow-up
August 28, 2026

16.71% of Top-10 URLs Fell Beyond Page 10. What That Number Actually Means.

A new dataset found that 16.71% of tracked URLs that ranked in Google's top 10 fell below position 100 during the August spam update.

That is 1.8× the normal rate in the study's July baseline. It is also not proof that 16.71% of websites were penalized, deindexed, or hit for using AI.

The number is useful because it is specific. It becomes dangerous the moment an SEO team uses it as a diagnosis.

10 min readPublished August 28, 2026

Key takeaways

  • SE Ranking compared 100,000 US keywords across 20 industries, using Aug 17→22 against a Jul 26→31 baseline.
  • The study recorded an 82% relative increase in top-10 URLs that ended up below position 100. It did not identify why those URLs moved.
  • Google confirmed the Aug 18→21 global spam update, but did not confirm a specific target, affected-site rate, or AI-content rule.
  • Read this alongside PikaSEO's 12-property case study: a broad SERP movement statistic and a site-level click study are not competing evidence.

The headline number is real. The panic is optional.

On August 27, Search Engine Land published new data supplied by SE Ranking. During Google's August spam update, 16.71% of URLs that had held a top-10 position fell beyond position 100 for the same tracked keyword. In the study's equivalent July comparison, the rate was 9.2%.

That is an 82% relative increase. It is a sharp signal that the August 18–21 window produced more violent ranking movement than the selected baseline. It also tells us something narrower than the social posts made it sound.

It tells us that a sampled set of URLs moved hard. It does not tell us that Google deindexed them, that every industry took the same hit, or that one production method caused the losses.

Top-ten URLs falling beyond position 100 during the August 2026 spam updateSE Ranking reported 9.2 percent in a July five-day baseline and 16.71 percent during the August spam-update comparison. This describes tracked URL movement, not deindexing or a universal site impact rate.The number that got everyone's attentionSame 100,000 US keywords • positions 1–100 only • SE Ranking dataset0%10%20%9.2%16.71%July baselineJul 26 → Jul 31Spam-update windowAug 17 → Aug 2282% relative increase

Source: SE Ranking data reported by Search Engine Land. The chart reproduces the reported comparison; it is not Google Search Console data.

What SE Ranking measured

This was not a survey of publishers, and it was not a study of Google's full index. SE Ranking analyzed US organic results for the same 100,000 keywords across 20 industries. It compared the movement between August 17 and August 22 with movement between July 26 and July 31, a period without a confirmed Google update.

QuestionWhat the study says
Keyword set100,000 US organic keywords across 20 industries
Update comparisonAug 17 → Aug 22, immediately around the named spam update
BaselineJul 26 → Jul 31, with no confirmed update
Position rangePositions 1–100 only
Top-10 disappearance rate16.71% in August versus 9.2% in July

The study also reported movement across all 20 tracked niches. Its top-10 volatility measure ranged from 74.64% in real estate to 85.55% in fashion and beauty. That range says the update was not confined to one obvious vertical. It does not show which kinds of sites or pages lost.

The useful inference

If a key URL fell from the top 10 to outside the first 100 results between August 17 and 22, that movement belongs in your spam-update investigation. It does not end the investigation.

Four things the study cannot tell you

Whether a URL was deindexed

The tracker stopped at position 100. A URL could be at 101, 1001, or absent from the index entirely. Those are different outcomes.

Why Google moved it

The report did not audit losing URLs or identify site-level policy violations. Correlation with the rollout is not a mechanism.

The site-level impact

One site can lose many keywords, while many unrelated sites each lose one. Keyword movement is not a publisher-impact distribution.

Whether AI content was targeted

Google did not name a target. Community reports on mass-produced AI pages should remain caveated hypotheses, not attribution.

Google's own position is cleaner than most post-update theories. Its rollout notice confirmed the date window, global scope, and completion. It did not name a new policy or a content type. Its spam policies remain the reference point for any audit.

There are credible reports that mass-generated SEO content may have been affected. Search Engine Journal's coverage gathers examples and explicitly labels them anecdotal. That is the appropriate confidence level until Google or a URL-level study can show more.

Why this does not contradict our August case study

PikaSEO's 12-property August spam update case study examined 5.3 million Search Console impressions and found that no property lost more than 10% of daily clicks during the August 18–21 window. The biggest observed cliffs began on July 30, weeks before the named update.

At first glance, that looks incompatible with a 16.71% URL drop rate. It is not. These datasets sit at different levels of the system.

SE Ranking's question

Across a fixed keyword set, how often did a URL that was top 10 on one date end up outside the top 100 five days later?

Our case study's question

Across 12 known properties, what happened to organic impressions and clicks through the official rollout window?

A tracker can catch a large amount of URL churn across a representative query corpus while a specific group of sites stays broadly stable. It can also catch keyword replacement where the winning domains receive enough traffic elsewhere that their sitewide graphs barely move.

The better reading is additive: the new dataset says this was a disruptive SERP window. Our data says a timestamp alone was a poor explanation for the losses we investigated. Both are reasons to do more diagnosis, not less.

Read the sequence, not a single headline

Start with our rollout and diagnosis guide. Then use the 12-property case study to separate correlation from a real traffic break. This report adds a third layer: keyword-level disruption in a large external dataset.

The 48-hour diagnosis after a severe ranking drop

Treat a top-10 to beyond-100 movement as urgent, but not self-explanatory. This is the order we would use before deleting pages, disavowing links, or blaming a writing tool.

  1. 1

    Confirm the movement in more than one dataset

    Compare Search Console query and page data, analytics landing-page sessions, and a small hand-checked keyword set. A tracker anomaly, tracking-location shift, or reporting bug is not a penalty.

  2. 2

    Segment the loss before you name it

    Split branded and non-branded queries, templates, directories, countries, devices, and search appearances. A single failing page family is a very different problem from a sitewide drop.

  3. 3

    Rule out the boring technical explanations

    Check indexability, robots rules, canonicals, redirects, status codes, template releases, hosting incidents, and hacked-content signals. Run a Technical SEO Audit on both a winner and a loser.

  4. 4

    Match facts to policy language

    A large page set is not automatically scaled-content abuse. Find the purpose, duplication pattern, user value gap, and production process that support a specific policy concern.

What to audit if your footprint resembles the losers

If the loss clusters around template-driven pages, resist the tempting fix: rewrite 20 introductions and wait. Google's policies care about the purpose and utility of a production system. That means the audit has to inspect the system too.

Content inventory

Group pages by template and ask what distinct information each one contributes. Location pages need local facts. Comparison pages need a real methodology. Tool pages need working functionality and interpretation, not a title, a form, and generic FAQs.

Generation and review loop

Document how pages are proposed, sourced, generated, checked, and retired. AI use is not the red flag. Shipping thousands of pages that no person can explain, validate, or maintain is.

Search-to-page promise

Check the query, snippet, and landing page together. A page that wins because it implies a calculator, local answer, product comparison, or expert review must deliver that thing clearly.

Use the Helpful Content Checker to make the review consistent, then make the final call with an editor or operator who understands the subject. A score is a prompt to investigate. It is not a policy verdict.

Bottom line: a severe movement rate is a starting gun

The August spam update produced a real and unusually sharp pattern in SE Ranking's dataset: 16.71% of tracked top-10 URLs fell beyond position 100, versus 9.2% in its July comparison. That deserves attention.

But no serious SEO operation should turn that statistic into a universal impact rate, a deindexing claim, or an AI-content verdict. The pages that moved are a clue. Your own dates, page groups, technical state, and user value are the evidence.

The move this week

  • Mark August 18–21 on every organic-performance chart.
  • Export URLs that lost the most non-brand clicks and group them by template.
  • Read the case study before deciding the rollout caused your decline.

Frequently asked questions

Sources and next checks