Quick answer: Google confirmed a spam update rolling out globally from roughly August 18 to 21, 2026. Do not change your site during a rollout. Instead, hold a matched baseline, wait for the rollout to be marked complete, gather 14 days of post-rollout data, and only then decide whether anything about your site — as opposed to the competitive set, seasonality, or your tracking — actually changed.
What Google confirmed, and what it did not
Google announced the update on its Search Status Dashboard, describing a spam update rolling out across all languages and regions with a completion window of a few days. Search Central's spam policies documentation is the reference for what the term covers: scaled content abuse, site reputation abuse, expired-domain abuse, cloaking, deceptive behaviour, and similar categories.
What Google did not say matters just as much. It did not publish a list of affected sites, describe which specific policies weighted most heavily, promise a recovery timeline, or characterise this as a core or helpful-content update. Any advice built on those specifics is invention, and following invented specifics is how sites end up worse off after an update than during it.
Why the first instinct is usually wrong
The reflex after a visible ranking drop is to change something immediately. Three problems with that. Rankings genuinely fluctuate during a rollout and often partially revert once it completes. If you change several things at once, you cannot attribute the recovery or the further decline to any of them. And in a spam update, a meaningful share of movement is competitors shifting position in the same result sets — nothing about your site changed at all.
The 14-day structure below exists to separate observation from action, which is the only way to learn anything from an update.
Days 1 to 3: lock the baseline
Before analysis, freeze your comparison data so it cannot shift under you.
- Export Search Console performance for the 28 days before the rollout start, at page and query level, with the date window recorded.
- Export the equivalent GA4 organic-landing-page report for the same window.
- Screenshot your Search Console indexing and page-experience summaries as they stand now.
- Record any of your own changes deployed in the four weeks before the rollout — releases, redirects, template edits, content pruning, migrations.
- Note external events: seasonality, a pause in ads, PR coverage, an outage.
That last two bullets resolve more "update" investigations than any ranking tool. Sites frequently attribute their own deployment to Google.
Days 4 to 7: segment before you conclude
Aggregate numbers hide the answer. Compare matched windows — equal length, equal weekday composition — and segment along four axes:
By page
Rank pages by absolute impression and click change. If the decline concentrates in a definable group — a blog category, a location-page set, syndicated or thin templated pages — that concentration is your lead. If every page dips proportionally, look at tracking and seasonality first.
By query group
Group queries by intent: brand, commercial, informational, local. Brand queries holding steady while commercial queries fall is a different problem from everything falling together. Remember that Search Console thresholds low-volume query data, so query totals will not reconcile exactly with page totals.
By country and device
A rollout described as global still lands unevenly. Restrict to your actual market before drawing conclusions; a drop driven entirely by irrelevant international impressions is not a business problem.
By surface
Separate standard organic results from local pack visibility. Local pack and Business Profile performance are governed largely by a distinct system, so confirm which surface moved before you diagnose. Business Profile insights, not Search Console, are the record for local pack interactions.
Days 8 to 11: honest policy review
Only now, with a confirmed and concentrated pattern, read Google's spam policies against your own site — and read them as a sceptical outsider would.
- Scaled content abuse. Have you published large volumes of pages with little original value, generated or otherwise, primarily to capture search traffic?
- Site reputation abuse. Are third parties publishing on your domain with minimal oversight — coupon sections, sponsored posts, hosted subfolders?
- Thin templated pages. Do your location or service-variant pages differ only by a swapped place name?
- Link practices. Have you bought, exchanged, or scaled links in ways that would be hard to defend?
- Expired-domain or redirect history. Did the domain, or anything redirected into it, previously host unrelated content?
- Deceptive elements. Anything shown to crawlers but not users, or interstitials that block the content promised in the result?
A "no" on all six, plus a concentrated drop, points toward competitive displacement rather than a policy problem — which changes the response from remediation to improving the affected pages on their merits.
Days 12 to 14: decide with rules, not mood
Write the decision rules down before you look at the last week of data, so the data does not talk you into something.
- Confirmed policy issue, concentrated drop: remediate the specific issue, document the date, expect a slow reassessment with no guaranteed timeline.
- No policy issue, concentrated drop: improve the affected pages substantively — depth, evidence, internal links, intent match. Treat it as a content programme, not an emergency.
- Uniform drop, no concentration: audit tracking, tagging, canonicals and indexing before touching content.
- Movement inside normal variance: do nothing. Record the observation and return to the roadmap.
- Manual action notice in Search Console: different process entirely — remediate fully, then file a reconsideration request.
Change one category of thing at a time and leave enough time between changes to attribute the result. Our reporting dashboards exist largely to make this kind of before-and-after attribution possible without re-exporting everything by hand.
Anti-panic controls
Three controls prevent the most expensive update mistakes. First, a written freeze rule: no content or technical changes during a confirmed rollout except genuine bug fixes. Second, a single owner for the investigation, so five people do not each change something. Third, a change log with dates, so that in November you can still tell what you did in August and what happened next.
Check the fundamentals while you wait
The waiting period is a good time to verify the technical and content fundamentals that matter regardless of any update.
Where to get help
If the pattern is concentrated and you are not sure whether it is a policy issue or displacement, that distinction is worth getting right before spending on remediation. Our SEO services include update investigations built on matched-window evidence, and you can contact us for a free consultation to talk through what your data actually shows.
Updates reward patience more than they reward speed. The sites that recover well are almost always the ones that measured first.
