Corrections are part of the editorial product.

Before: Direct effects on training, citation timelines and automatic Wikidata eligibility were suggested.

After: Both editions distinguish verifiable identity, documented markup, Wikidata eligibility and measured outcomes. Fictitious profiles and guaranteed timelines were removed.

Before: Quotas and visibility figures were still presented as rules; the English translation retained outdated advice.

After: Templates are identified as editorial choices. Unverified figures and universal quotas were removed, and sources and translation were updated.

Before: Citation lifetimes and a universal advantage for established content were asserted without measurement.

After: The guide describes fact-based maintenance, genuine revision dates and monitoring without a citation guarantee.

Before: Unsupported recommendations remained about JavaScript rendering, ranking signals, IndexNow timelines and Copilot citation selection.

After: Bingbot Edge rendering and the AI Performance report are now sourced. Ranking, timing and citation guarantees were removed; receipt, indexing and citation are distinguished.

Before: Thresholds of 10 to 20 articles and a universal advantage over PageRank were presented as established facts.

After: The page distinguishes the documented Google News system, informal SEO terminology and RAG research. Unsupported thresholds and causal claims were removed.

Before: A length of 3 to 6 lines, chunk boundaries and a FAQPage advantage were presented as universal rules.

After: The page explains that retrieval implementations vary, there is no magic length, and FAQPage guarantees neither a rich result nor a citation.

Before: Timelines, length preferences, citation factors and clicks or queries in the generative AI report were asserted without sufficient evidence.

After: The answers distinguish editorial practices from documented rules and describe the specific experimental report: impressions only, with access limited to some properties.

Before: Outdated Anthropic tokens were used, and an English page described PerplexityBot as a training crawler.

After: The pages distinguish ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot and Perplexity-User using the current documentation.

Before: A citation multiplier was attributed to an unidentifiable study.

After: The causal claim was removed. The page now distinguishes extractable formatting, correlation and evidence.

Before: Markup and positions 8 to 15 were presented as documented selection factors.

After: The page follows the official rule: no special structured data is required for Google generative search features.

Before: A direct effect of markup on AI citations was implied.

After: The scope is limited to documented uses: describe visible content and support compatible features, without a citation guarantee.

Before: Microsoft Clarity and non-public behavioural signals were linked to rankings.

After: The page states that Clarity has no documented ranking benefit and should not be installed for SEO.

Before: Unsubstantiated French launch dates and coverage rates were published.

After: Unverifiable dates and metrics were corrected or removed.

Before: A Seer study was interpreted as a CTR increase for cited pages.

After: The page correctly describes the observed CTR decrease on queries displaying an AI Overview.

Before: The French desktop market share was overstated.

After: The value was corrected using the period and source actually consulted.

Before: Ten English articles were served with French language settings and metadata.

After: The locale is now derived from the path, with reciprocal hreflang and English navigation.