Algorithm Updates 2024-2026: The Most Important Changes

Between 2024 and 2026, Google rebuilt its ranking system more profoundly than in the years before. The Helpful Content System moved into core ranking, three new spam policies were added, an internal documentation leak exposed thousands of ranking signals, and with AI Overviews, search behavior shifted away from the classic click. For B2B sites in the DACH region, this translates into concrete tasks: each of these events had direct consequences for title tags, content structure, internal linking, schema markup, and the question of how content can still become visible in generative answers at all. This article puts the most important changes in chronological order and derives what to do on-page.

Why this wave of updates changed everything on-page

Earlier algorithm updates could often be understood as an isolated turn of a single dial. The 2024 to 2026 wave is different because it hits three levels at once: site-wide quality assessment, the evaluation of individual signals, and the way search results are delivered in the first place.

Google itself provided the clearest evidence. With the March 2024 Core Update, according to Google, 45 percent less low-quality, unoriginal content appeared in search results, more than the 40 percent originally expected (international, rollout completed April 19, 2024). A reduction of this magnitude shifts the visibility landscape for every site, including sites that were not directly demoted: whoever remains in the cleaned-up index competes for different positions.

At the same time, the SERP itself is changing. According to an international analysis of more than ten million keywords, AI Overviews were triggered for 15.69 percent of queries in November 2025, after the share had peaked at 24.61 percent in July 2025 and had still been at 6.49 percent in January 2025 (international). The consequence for on-page work is twofold: classic ranking factors remain relevant but are no longer enough, because a growing share of answers is generated above the organic results.

In Austria, the market structure amplifies this effect. In May 2026, Google's search engine market share in Austria was 81.87 percent, ahead of Bing at 9.01 percent and DuckDuckGo at 2.75 percent (Austria). If you build B2B visibility in Austria, you are effectively optimizing almost exclusively for a single system. Every Google algorithm change therefore hits the Austrian market almost unfiltered.

Timeline: the most important updates from 2024 to 2026

According to its Search Status Dashboard, Google rolled out nine core updates, eight spam updates, and one Discover update between March 2024 and October 2026. The table lists the start and completion of each rollout, and each update links to its entry in the dashboard.

When it launched, the February 2026 Discover Core Update applied only to English-speaking users in the US. Google announced that it would later expand the update to all countries and languages but gave no date for this.

Two events that shape this article are deliberately missing from the table: the Content Warehouse API Leak in May 2024 and the expansion of AI Overviews are not algorithm updates. Both do show, however, which signals Google captures and how the SERP is changing, so they belong here as context. For on-page work, three points each set their own focus:

• March 2024 Core Update (site-wide quality): Google integrated the Helpful Content System into core ranking and introduced three new spam policies at the same time. The focus is on site-wide content quality, originality, and whether content is created primarily for users or for rankings.
• May 2024, Content Warehouse API Leak (context, not an update): Internal documentation became public and confirmed that engagement signals and a stored authority attribute are part of the evaluation. The on-page focus shifts to click quality, compelling titles, and a structure that keeps users on the page.
• Core updates from August 2024 onward (consolidation): After three core updates in August, November, and December 2024, five more followed from March 2025 to May 2026. Of these five, the December 2025 Core Update took the longest at 18 days. The on-page focus remains on well-sourced content, freshness, and E-E-A-T, now in the context of an increasingly AI-driven SERP.

The terms need to be kept apart. A core update is a broad re-evaluation of the entire relevance and quality logic and does not appear as a notification in Search Console. A manual action, by contrast, is triggered by a reviewer and reported. The 2024 spam policies work through both channels: algorithmically in ongoing ranking and as the basis for manual actions.

The March 2024 Core Update in detail

The March 2024 Core Update is the pivot point of the entire wave. Two things happened at once.

First, the Helpful Content System, previously a separate site-wide signal, was integrated into core ranking. Quality is therefore no longer treated as a downstream filter but as part of the central evaluation across multiple systems. On-page, this means a single strong page can be held back by a weak overall environment on the same domain. Content quality is a property of the site, not just of the individual page.

Second, Google introduced three new spam policies: Expired Domain Abuse, Scaled Content Abuse, and Site Reputation Abuse (international). Each one targets a specific pattern:

• Scaled Content Abuse: Generating many pages with the primary purpose of manipulating rankings rather than helping users. The policy applies regardless of whether the content was created through automation, by humans, or by a combination of both. Mass-rewritten platitudes are therefore a violation even when humans were involved.
• Site Reputation Abuse: Third-party content published without close editorial oversight by the host site, primarily to exploit the domain's authority, colloquially known as parasite SEO. For this policy, Google set a separate enforcement date: it took effect on May 5, 2024 (international).
• Expired Domain Abuse: Buying expired domains to misuse their legacy authority for low-quality content.

One technical clarification matters here: the three spam policies were announced in the same post as the core update but technically rolled out together with the March 2024 Spam Update, which started at the same time. What counts in practice is the consequence: in March 2024, the quality track and the spam track were merged.

The Content Warehouse API Leak and its on-page signals

In May 2024, internal Google documentation that had previously been left in a code repository by mistake became public. According to iPullRank, the analysis covers 2,596 modules with 14,014 attributes (international). The leak does not confirm ranking weights, but it does show which signals Google captures and stores in the first place. Three findings are central to on-page work.

NavBoost appears by name 84 times in the documentation and is described as a click-driven system that amplifies, promotes, or demotes rankings (international). This establishes that aggregated click behavior is part of the evaluation. The on-page takeaway: the title tag and the displayed snippet decide the first click and at the same time feed a system that evaluates click quality over time. A compelling, precise title is therefore a ranking-relevant asset, not a cosmetic detail.

Second, the leak confirms a stored attribute called siteAuthority, which is derived from a quality signal and applied in the evaluation (international). This contradicts earlier public statements that Google does not use a site-wide authority value. In practice, it backs up what the Helpful Content System already suggests: domain-wide quality and topical consistency pay off for every single page.

Third, and most important for day-to-day work: the existence of these engagement and authority signals shifts the focus. If you optimize titles, content structure, and internal linking, you optimize for the initial relevance assessment and at the same time for what users do after the click. A clear heading hierarchy, a fast answer to the search intent, and useful internal links to further content extend dwell time and reduce bounces back to the SERP.

AI Overviews: from classic ranking to citability

The most consequential shift is not algorithmic but structural: the answer is increasingly generated in the SERP itself. This measurably changes the economics of the click.

An international analysis of 300,000 keywords, half with and half without an AI Overview, found that the presence of an AI Overview correlates with a 58 percent lower average click-through rate for the top-ranking page (international, December 2023 compared with December 2025). A second, independent survey by the Pew Research Center confirms the direction: when an AI summary was present, users clicked on a result in only 8 percent of cases, compared with 15 percent without a summary, which corresponds to a relative reduction of 46.7 percent (US, via Search Engine Journal).

The strategic consequence is called Generative Engine Optimization (GEO): when clicks become rarer, being mentioned in the AI answer itself becomes the goal. The academic foundation comes from the Princeton GEO study (KDD 2024). It shows that targeted GEO methods can increase visibility in generative answers by up to 40 percent (international). The framing matters: GEO does not replace classic on-page optimization but extends it with the question of whether a piece of content is easy for an AI system to extract and cite.

Putting GEO into practice on-page

The Princeton study is so relevant because its most effective methods are classic on-page disciplines, just with a different goal. The findings translate into concrete measures that also remain good SEO practice.

• Embed statistics with sources: Concrete numbers with a named source make a paragraph citable for AI systems. A vague claim without evidence is picked up less often than a precise, sourced statement.
• Use quotes and entities deliberately: Clearly naming people, organizations, and technical terms makes machine attribution easier. Entities give the model anchor points it uses to assign your content to a topic.
• Front-load extractable definitions: A self-contained definition of one to two sentences directly below the matching heading is both a featured snippet candidate and a preferred building block of generative answers.
• Put the answer before the context: The core statement belongs at the start of the section, with the reasoning after it. AI systems prefer to extract clearly delimited passages that make sense on their own.

These levers match good E-E-A-T practice and work in classic search as well as in AI answers. They are the most direct on-page response to the falling click-through rate when an AI Overview appears.

On-page consequences of the updates: title, structure, schema

Taken together, the updates add up to a clear on-page requirements list. Four areas stand out.

• Title tags as an engagement signal: Because the leak confirms click evaluation via NavBoost, the title is no longer just a keyword field. It has to match the search intent, be precisely worded, and invite the click, because the subsequent click quality feeds into the evaluation.
• Heading hierarchy as an extraction grid: A clean, sequential structure of one H1 and logically nested H2s and H3s helps both classic crawlers and AI systems capture and classify individual passages. Headings should describe what the section answers instead of carrying vague buzzwords.
• Schema markup as JSON-LD: Structured data makes content machine-readable. JSON-LD is Google's preferred format and, according to the Web Almanac, rose from 34 percent of pages in 2022 to 41 percent in 2024 (international). That makes it the fastest-growing of the three markup formats and the obvious standard for new implementations.
• Content structure for helpful content signals: Content has to visibly answer a specific user question, provide its own data or analysis, and back up its claims. That is exactly what the integration of the Helpful Content System into core ranking rewards as quality.

Strengthening E-E-A-T on-page after the quality updates

The 2024 to 2026 quality updates turned experience, expertise, authoritativeness, and trustworthiness from a soft recommendation into a practical necessity. E-E-A-T is not a single score, but it is approximated algorithmically through numerous signals. On-page, you can make it concrete.

• Author signals: Named authors with verifiable qualifications, a linked About Us page, and a consistent entity across the site. Anonymous mass content is exactly the pattern Scaled Content Abuse targets.
• Proof of experience: Your own project figures, practical examples, and original assessments lift a text above interchangeable boilerplate. This is the experience component that generic AI content typically does not deliver.
• Freshness and originality: Sourced, current data instead of outdated platitudes. The merging of quality and spam signals penalizes unoriginal, recycled content site-wide.
• Evidence and sources: Link key statements to reputable primary sources. This increases trustworthiness for users, for Google, and for AI systems, which prefer to cite sourced statements.

For B2B in the DACH region, this is an advantage: subject-matter depth, real references, and traceable authorship are the basis of trust here anyway and match the E-E-A-T requirements exactly.

Common mistakes in dealing with the updates

When sites react to algorithm updates, the same mistakes keep recurring, and they cost or give away visibility.

• Confusing a core update with a manual action: If you wait in vain for a Search Console notification after a traffic drop, you lose time. A core update does not announce itself; only a manual action shows up there.
• Hoping for a quick recovery: After an algorithmic demotion, there is no reconsideration request. Re-evaluation only happens in a later core update, which can take months.
• Treating symptoms instead of causes: Optimizing individual pages helps little if the site-wide quality problem persists. Since the HCS integration, quality is a property of the entire domain.
• Ignoring GEO or treating it as everything: Both extremes do harm. If you ignore AI Overviews, you lose visibility; if you neglect classic rankings, you lose the base from which AI systems cite at all.
• Overinterpreting the leak: The API leak shows which signals exist, not how they are weighted. If you derive exact ranking recipes from it, you are building on speculation.

Recovery strategy and measuring success

The recovery path depends on the type of hit. With a manual action, Search Console names the cause: identify the violation, remove or substantially improve the affected content, then submit a reconsideration request. With an algorithmic demotion from a core update, the only remedy is to raise content quality site-wide and wait for the re-evaluation in the next relevant update.

In both cases, the diagnostic core is a content audit. Thin content, redundant template pages, and unsourced mass-produced texts are identified and then deleted, merged, or enriched with evidence and your own examples. This pruning of unhelpful pages is especially important after the HCS integration, because weak pages drag down the site-wide assessment.

After this wave of updates, you need to track several metrics in parallel:

• Rankings and visibility: Positions for the most important keywords, ideally tracked over time around update dates to spot correlations.
• Click-through rate by position: Since AI Overviews push CTR down, the click-through rate from Search Console, not just the raw position, is an early indicator of SERP changes.
• AI visibility: Whether and where content appears as a source in AI Overviews and other generative answers becomes a metric in its own right alongside classic rankings.
• Engagement after the click: Dwell time and bounce behavior as a proxy for click quality, which the leak confirmed as relevant to the evaluation.

Realistically, recovery is not a quick process. A manual action can be lifted within weeks after a successful reconsideration request; recovering from a core update can take several months until the next relevant update.

Further reading and conclusion

The 2024 to 2026 wave of updates follows a consistent pattern: Google evaluates quality site-wide, demonstrably factors in engagement signals, and increasingly delivers answers directly in the SERP. This sets a clear on-page course for Austrian B2B sites. First, treat quality as a domain property and consistently prune weak pages. Second, align titles, heading hierarchy, and internal linking with click quality and dwell time, because the leak confirmed these signals.

Third, treat GEO as an extension of on-page work: sourced statistics, clear definitions, and JSON-LD schema so that your content gets cited in generative answers. Since Google handles more than four out of five searches in Austria, each of these measures has a direct effect on business-relevant visibility. Sensible next steps are a content audit for thin and mass-produced content, a review of title and heading structure against search intent, and a targeted expansion of evidence, author signals, and structured data.