Algorithm Updates Checklist: Diagnose, Respond and Recover

What counts as an algorithm update?

An algorithm update is a change to a search engine’s ranking, retrieval, spam detection or result presentation systems. Google distinguishes broad core updates, spam updates, changes to individual ranking systems, feature changes and smaller adjustments that may never be announced.

A core update broadly recalibrates how results are assessed. It is not a penalty directed at one website. Google describes the process as reassessing content against its goal of presenting helpful, reliable and people-first results. Ranking systems also address areas such as language understanding, freshness, links, reviews, local results and spam.

According to the Google Search Status Dashboard, confirmed updates available by August 10, 2026 included the March 2026 core update, March 2026 spam update, May 2026 core update and June 2026 spam update. Unconfirmed volatility should not be labeled an update merely because monitoring tools or practitioners detect movement.

The first 48 hours: algorithm update triage checklist

1. Confirm timing. Compare the first date of sustained change with the official rollout dates. Account for reporting delays and the full rollout window.
2. Annotate the data. Record the update, migrations, releases, template edits, tracking changes, promotions and outages on one timeline.
3. Validate measurement. Check analytics tags, consent behavior, conversion definitions, Search Console ownership and reporting filters.
4. Inspect Google Search Console. Check manual actions, security issues, indexing reports, crawl statistics and affected search types.
5. Segment the decline. Separate brand from nonbrand, mobile from desktop, countries, directories, templates, query intents and rich or generative appearances.
6. Check demand and SERP composition. Look for seasonality, new competitors, news results, shopping modules, local packs, video, forums or AI answers taking more space.
7. Preserve evidence. Export query and page data before interfaces and comparison windows change.
8. Avoid reflexive site-wide edits. Fix confirmed defects immediately, but wait for rollout completion before interpreting ordinary volatility as a stable outcome.

The goal is not to prove that an update happened. The goal is to isolate what changed, where it changed and which explanation best fits the evidence.

Algorithm update diagnostic matrix

Build a defensible loss analysis

Use equal comparison periods that account for day-of-week patterns and seasonality. In Search Console, analyze Web, Image, Video, News and relevant generative reporting separately. Google recommends breaking traffic changes down by search type, date, query, page, device and country. Add business dimensions such as product category, margin, lead quality and customer stage.

Create cohorts for pages that lost, gained and remained stable. Stable pages provide an internal control: if one template fell while another on the same domain held, a universal domain-level explanation is less convincing. Compare each cohort for intent, update date, author expertise, original evidence, internal link depth, freshness need, media, structured data eligibility and indexation state.

Use third-party volatility tools only as corroboration. The Semrush Sensor, for example, measures movement across fixed keyword sets on a 0 to 10 scale. It can show unusual SERP turbulence, but it cannot establish why an individual website lost traffic. Likewise, rank trackers should use stable locations, devices and result settings. Otherwise, localization and feature changes can look like ranking changes.

Choose the response by update and failure type

Broad core update

Compare affected pages with results now satisfying the same intent. Look beyond keywords: assess whether the winners provide a clearer answer, stronger first-hand evidence, better task completion, more appropriate format or a more coherent topical resource. Avoid deleting useful pages simply because they declined during a core rollout.

Spam update

Audit against Google’s spam policies. High-risk patterns include scaled low-value production, scraping, expired-domain abuse and publishing third-party material primarily to exploit the host site’s ranking signals. AI assistance is not inherently prohibited. Producing large volumes of minimally useful pages for rankings is the underlying risk.

Feature or presentation change

If rankings remain stable but clicks decline, inspect the live results. AI answers, local packs, shopping results, discussions, video and rewritten snippets can reduce clicks. The remedy may be better answer formatting, distinctive data, stronger brand demand or targeting queries where users still need a site visit.

Manual action or technical incident

These require direct remediation rather than core update speculation. Resolve the stated manual action, security issue, indexing directive, redirect or server defect, document the fix and validate representative URLs.

Content and topical architecture recovery

Start with pages responsible for meaningful impressions, revenue or qualified leads. Rewrite only after defining the intended user task. Add original comparisons, expert interpretation, constraints, methodology, examples and decision rules that cannot be reproduced by paraphrasing the current results. Remove unsupported certainty and make commercial relationships visible.

Map the topic as a hub-and-spoke system. A durable algorithm updates hub can link to core update diagnosis, spam policies, traffic-drop analysis, Search Console segmentation, content consolidation and AI visibility measurement. Each spoke should answer a distinct intent. Merge pages competing for the same query, redirect obsolete equivalents when appropriate, and preserve unique sections that still earn links or satisfy long-tail needs.

Improve retrieval with concise definitions, descriptive headings, explicit relationships between entities and answer-first passages that remain accurate when extracted. Cover likely query fanout naturally, including what changed, whether a site was penalized, when to act, how to measure recovery and when expert help is justified.

Refresh on evidence, not an arbitrary publishing calendar. Maintain update logs, statistics pages, comparison assets and original datasets that create natural link demand. Expert contribution programs can strengthen accuracy when contributors are real, identified and substantively involved.

Technical SEO, crawling and indexation controls

Run a representative crawl of losing, stable and winning templates. Verify status codes, robots directives, rendered content, canonicals, hreflang, pagination, structured data and internal links. Compare XML sitemaps with canonical indexable URLs. Accidental noindex directives, canonical drift and client-side rendering failures can produce update-like declines.

Use server log files to determine whether important pages are being crawled, whether bots encounter repeated errors and whether crawl activity shifted toward faceted or duplicate URLs. Improve crawl prioritization by controlling parameter traps, consolidating duplicates, repairing redirect chains and linking important pages from relevant hubs. Do not block resources required to render meaningful content.

Canonical discipline matters during consolidation. A canonical is a hint, not a substitute for consistent redirects, internal links and sitemap signals. Ensure every signal names the preferred URL. For large sites, monitor indexable page counts by template rather than relying on one domain total.

Structured data must match visible content and supported eligibility rules. Adding markup cannot rescue an unhelpful page, and fabricated reviews or hidden claims create policy and trust risks.

AI Overviews, AI Mode, Bing Copilot and ChatGPT

Generative search can change both retrieval and click behavior. Google’s guidance continues to emphasize foundational SEO: crawlable pages, indexable content, useful information, clear structure and accurate claims. In June 2026, Search Console added dedicated generative-AI performance reporting for AI Overviews, AI Mode and generative Discover features, giving teams a clearer way to separate visibility across these experiences.

Optimize for answer absorption without making the page disposable. State definitions and procedures clearly, then offer assets that justify a visit: interactive tools, complete datasets, downloadable methods, nuanced comparisons, local availability, expert evaluation or a transaction. Use consistent entity names and source important numerical claims.

Pew Research found AI summaries in 58% of observed Google searches in its March 2025 browsing analysis, with lower link-clicking when summaries appeared. A 2026 preprint reported source clicks in roughly 1% of AI Overview visits, but that figure is not an official Google metric and should not be generalized without considering the study design. Measure impressions, citations, assisted conversions, branded search and qualified visits instead of treating clicks as the only outcome.

Authority recovery and controlled growth

Audit whether lost pages also lack corroborating authority. Use link-intersect analysis to find legitimate publications, associations and resource pages that cite several competitors but not your asset. Reclaim broken links and convert accurate unlinked brand mentions through polite outreach. Digital PR works best when attached to verifiable data, expert analysis, useful tools or timely public-interest findings.

Test titles and intent positioning in controlled cohorts. Change one comparable group, preserve another and evaluate click-through rate, ranking, conversions and query mix over a defined period. A more clickable title that attracts the wrong audience is not an improvement. Avoid changing titles, templates, internal links and body copy simultaneously because attribution becomes impossible.

Risk and reward: aggressive guest-post networks, rented authority sections and expired-domain strategies may produce short-term visibility, but they create substantial spam-policy and durability risk. Do not use hacked links, cloaking, doorway pages, fabricated reviews, deceptive redirects or schema that contradicts visible content.

Seek specialist support when losses affect critical revenue, span several templates, involve migrations or internationalization, or cannot be separated from technical defects. Buyers should ask for cohort analysis, technical evidence, policy awareness and measurable priorities, not a guaranteed recovery date.

Recovery measurement: proven, consensus and uncertain

Proven from official guidance: core updates are broad reassessments rather than site-specific penalties. Google advises analyzing drops by search type, date, query, page, device and country. Improvements may become visible within days, but broader site reassessment can take months, and recovery is not guaranteed.

Practitioner consensus: the most useful diagnosis compares losing pages with stable internal controls and current winners. SEO teams generally wait for a rollout to finish before judging noncritical changes, prioritize technical certainty first and remediate affected cohorts rather than rewriting an entire domain.

Still uncertain: volatility tools cannot identify the exact system or signal involved. Practitioners reported intent-specific shifts after the May 2026 core update and unconfirmed volatility around July 11 and August 1 to 3, but these reports are anecdotal. They do not establish another confirmed update.

Use a recovery scorecard with nonbrand impressions, median position by cohort, indexable canonical pages, click-through rate, conversions, revenue or lead quality, AI citations and branded demand. Review weekly during volatility and monthly for durable trends. A successful recovery restores business outcomes and qualified visibility, not merely the previous count of ranking keywords.