From Rankings to Recommendations: The New Search Playbook
AI-generated answers have not made SEO obsolete. They have expanded its remit. Brands must now earn four outcomes: indexation, citation, recommendation and conversion.
That distinction matters because visibility increasingly occurs before, or without, a website visit. The strongest strategy is not to replace SEO with a separate collection of “GEO hacks.” It is to build authoritative, technically accessible content and measure how that content performs across search results, AI answers and subsequent customer behaviour.
The most important recent developments
In May 2026, Google published its clearest guidance yet on optimising for generative search. Google confirmed that AI Overviews and AI Mode rely on its existing search index, ranking systems, retrieval-augmented generation and “query fan-out", through which one question can trigger searches across multiple subtopics. Google’s conclusion is unambiguous: foundational SEO remains the basis for generative-search visibility. Google’s generative AI optimisation guide
Google also rejected several fashionable tactics. It says llms.txt does not help visibility in Google Search, special AI markup is unnecessary, content does not need to be divided into artificial "chunks", and creating pages for every imagined query variation may cross into scaled-content abuse.
The measurement environment is improving, however. In June, Google began testing dedicated Search Console reports showing generative-AI impressions, cited pages, countries, devices and visibility over time. Google Search Console announcement
Bing moved even further toward citation-level measurement. Its AI Performance dashboard reports citations, cited URLs, visibility trends and samples of the “grounding queries” used to retrieve a publisher’s content across Bing, Copilot and selected partner experiences. Bing AI Performance announcement
Meanwhile, OpenAI confirms that publishers must allow OAI-SearchBot to include their content in ChatGPT summaries and snippets. ChatGPT referral links carry utm_source=chatgpt.com, while the separate GPTBot control governs potential training use. OpenAI’s publisher guidance
The strategic implication is simple: SEO and GEO increasingly share an operating system, but they require a broader scoreboard.
SEO remains the eligibility layer
A page cannot become a reliable AI source if search and retrieval systems cannot access, interpret and trust it.
The technical priorities remain familiar:
Maintain clean indexation, canonical URLs, XML sitemaps and accurate lastmod values.
Ensure important content is present in crawlable HTML and isn’t hidden behind fragile JavaScript interactions.
Preserve snippet eligibility; Google requires a page to be indexed and eligible to appear with a snippet before it can support AI Overviews or AI Mode.
Use IndexNow for important additions, changes and removals where Bing and participating services matter.
Audit crawler permissions separately. Googlebot, OAI-SearchBot and GPTBot serve different purposes and should not be treated as one control.
Keep product feeds, Google Business Profiles, Bing Places and other authoritative business records accurate.
Every organisation should maintain a crawler-governance register showing what is allowed, what is blocked, why the decision was made and who owns it. This is now a communications and data-governance issue, not merely a technical SEO setting.
Non-commodity information is the new content advantage
AI systems can reproduce generic advice almost instantly. Publishing another competent summary of widely available information gives a search engine little reason to cite the brand behind it.
High-value content contains something that cannot easily be reconstructed from everybody else’s pages:
Original research, benchmarks or customer data
First-hand experience and documented case studies
Named expert analysis
Proprietary frameworks or decision tools
Clear comparisons based on disclosed criteria
A defensible point of view
Current facts with visible dates, sources and methodology
Google’s 2026 guidance explicitly prioritises “non-commodity content” and contrasts first-hand insight with pages that merely restate existing material. Its February Discover update similarly emphasised original, timely, locally relevant and in-depth content while reducing sensational and clickbait material. Google’s February 2026 Discover update.
The practical test is not “Did we cover the keywords?" It is “What information would disappear from the web if we did not publish this page?”
Optimise for questions, not just keyword strings
AI search frequently decomposes a broad request into several related retrieval tasks. A prospective customer asking for a recommended crisis communications adviser might implicitly be asking about industry experience, response times, geographic coverage, leadership credentials, methodology, evidence and price.
Content architecture should therefore cover the complete decision problem. Build authoritative topic clusters around the questions customers ask before, during and after a decision. Connect those pages through descriptive internal links and maintain one clear purpose for each page.
This does not mean manufacturing hundreds of near-duplicate long-tail pages. Google specifically warns against producing pages for every possible fan-out query. The objective is comprehensive topical coverage with genuine differentiation, not synthetic scale.
Make important claims easy to extract and verify
You don’t need to write robotic prose for AI systems. However, there is a need to make important information precise and self-contained.
Strong answer passages should:
Give the direct answer near the beginning of a relevant section.
Use explicit names rather than ambiguous pronouns.
Include dates, jurisdictions, units and limitations where they affect accuracy.
Support material claims with primary sources.
Use descriptive headings, lists and comparison tables when they improve comprehension.
Separate fact from company opinion, prediction or recommendation.
This is not artificial “content chunking". It is disciplined executive communication. Bing specifically recommends clear headings, tables, evidence and reduced ambiguity because those elements help systems reference information accurately.
FAQ sections should only be used for real follow-up questions. Adding repetitive questions to every page will not compensate for thin underlying content.
Treat entity consistency as reputation infrastructure
An AI system may encounter a brand through its website, news coverage, review sites, social and video platforms, databases, business listings and third-party commentary. Contradictory names, product descriptions, executive biographies or location details make the entity harder to interpret.
Organisations should establish a canonical fact set covering their name, positioning, products, leadership, locations, credentials and key claims. That information should remain consistent across owned properties and legitimate external profiles.
Relevant structured data remains useful for rich results and entity clarity, but it is not a special GEO switch. Google explicitly says there is no unique schema required for generative search.
Digital PR also has a larger role. Credible media coverage, expert contributions, independent reviews and original research create corroboration beyond the corporate website. The aim is not to manufacture mentions. It is to become a source that other trustworthy sources choose to reference.
Prepare for multimodal and agentic discovery
Search visibility is no longer limited to body copy. Google can surface images, video, products and local-business information inside generative experiences. Important visual material should have descriptive surrounding text, accessible alternatives and consistent entity information.
Agent readiness is the next emerging layer. OpenAI recommends descriptive ARIA roles, labels and states so its browser agent can understand interactive elements. Google is also directing businesses toward accessible DOM structures, clear interfaces and emerging commerce protocols.
For transactional websites, test whether an automated agent can identify a product, compare options, understand availability and complete key steps without encountering ambiguous controls. Accessibility improvements increasingly benefit both people and software agents.
Replace the traffic-only dashboard
Traffic remains commercially important, but it no longer describes the entire discovery journey. Similarweb reported that 95% of ChatGPT users also used Google in both September 2025 and May 2026, suggesting that AI and conventional search are currently complementary behaviours rather than clean substitutes. Similarweb’s 2026 analysis
A useful search scorecard should include:
Traditional rankings, impressions, clicks and conversions
AI citations and cited pages
Brand mentions, recommendation frequency and competitive share
The prompts and topics generating those appearances
Accuracy and sentiment of the surrounding answer
Branded-search and direct-traffic movement
AI referral quality, assisted conversions and revenue
Indexation, freshness and crawler health
Do not interpret a single prompt test as a stable ranking. A 2026 research preprint found substantial citation variation when identical queries were repeated across generative platforms. AI visibility should therefore be sampled repeatedly and reported as a trend or range.
A practical 90-day priority plan
During the first 30 days, audit indexation, crawler permissions, sitemaps, IndexNow, canonicalisation, business listings and analytics attribution. Establish a repeatable set of customer prompts across Google, Bing/Copilot, ChatGPT and other platforms relevant to the audience.
During days 31–60, identify high-value topics where the organisation possesses real evidence or expertise. Improve those pages with original information, direct answers, expert attribution, primary sources, clear update dates and useful visual material.
During days 61–90, connect content, PR, subject-matter experts and analytics into one discovery programme. Benchmark citations, mentions, branded demand, referrals and conversions. Review inaccurate AI answers as reputation issues and trace them back to missing, unclear or contradictory source material.
The strategic conclusion
The search contest is no longer simply about securing the highest link. It is about becoming the source an answer system can retrieve, the authority it can cite, the brand it can recommend and the destination a customer eventually chooses.
SEO provides the technical and authority foundation. GEO adds attention to extractability, corroboration, cross-platform visibility and indirect customer journeys.
Organisations that separate the two into competing disciplines will create duplicated work. Those that integrate search, content, PR, data governance and measurement will be much better placed for the next phase of discovery.
———
About the Author
Ana Maria Gardiner is a senior communications executive, board-level adviser and founder of Lighthouse PR. With extensive experience providing strategic counsel to multinational organisations and leadership teams, her expertise spans corporate reputation, public relations, marketing communications, crisis management and high-stakes communications.
Throughout her career, Ana Maria has led and implemented communications strategies for organisations including JPMorgan, Coca-Cola, ExxonMobil, Siemens Energy, HEINEKEN, Carrefour, Lexus, Franklin Templeton, BNP Paribas, Sungrow, XTB, Bitget, EssilorLuxottica and Pfizer. Her experience encompasses projects across the Middle East, North Africa, and Central and Eastern Europe, covering a broad range of industries and business environments.
She currently works closely with senior executives and board-level decision-makers, advising Lighthouse PR clients on reputation management, strategic positioning, communications risk management, and the development of responses to sensitive situations and crises.
Ana Maria holds a bachelor’s degree in Political Science and a master’s degree in European Affairs.
About Lighthouse PR
Lighthouse PR is an independent public relations and strategic communications consultancy headquartered in Bucharest, working with organisations across Romania, Central Europe, and South-Eastern Europe. The agency provides senior-led counsel to companies operating in financial services, energy, manufacturing, technology and cybersecurity, transport and logistics, retail, and FMCG.
Its portfolio of services includes media relations, corporate communications, reputation management, crisis communications, crisis preparedness and response, stakeholder and investor communications, social media and influencer management, B2B communications, risk assessments, business continuity planning, resilience framework development, media buying, corporate events, and SEO and website design services.
Lighthouse PR holds ISO 9001 and ISO 27001 certifications and is the exclusive representative for Romania and the Republic of Moldova of Eurocom Worldwide and Crisis Communication Network Europe, two international networks that strengthen the agency’s capacity to manage communications projects and crises with regional and international dimensions.
Through its consultancy model, Lighthouse PR places senior expertise at the heart of every client relationship, from strategy development and reputation management to the implementation of communications programmes and the coordination of responses in sensitive situations.
Lighthouse PR: Clear. Concise. Convincing.
