SEO agency and SEO digital marketing for brands winning AI Overviews alongside the SERP.

The category has renamed itself. Search Engine Land's May 8 2026 lead headline calls it: SEO's new goal in 2026 is recognition, not rankings. Condé Nast publicly forecasts (May 13 2026) that search will collapse to a single-digit percentage of traffic for many publishers.

The replacement frame is twofold:

• AEO (Answer Engine Optimization): citations inside answer boxes (Google AI Overviews, AI Mode, Bing Copilot)
• GEO (Generative Engine Optimization): citations inside off-Google chat surfaces (ChatGPT Search, Perplexity, Claude, Gemini)

A 2026 SEO programme must run all three: traditional rankings, AEO citations, GEO citations.

Six material changes:

1. FAQ rich results died (May 7 2026)
2. Google formally documented AI Overviews and AI Mode optimisation guidance (Dec 10 2025: no special optimisations needed)
3. Bing Webmaster Tools AI Performance citation report shipped (Feb 2026)
4. Major publishers (Condé Nast) publicly forecast search collapsing to single-digit traffic share (May 13 2026)
5. Search Engine Land reframed the category from rankings to recognition (May 8 2026)
6. Google confirmed spam policies apply to AI Overviews and AI Mode (May 15 2026)

Google's official position (developers.google.com, December 10 2025): no additional requirements to appear in AI Overviews or AI Mode, no special optimisations necessary. No machine-readable files, no AI text files, no markup needed. Performance shows up in Search Console under the standard Web search type.

The practitioner reality (Semrush 200,000-keyword study):

• 82% of desktop AIO triggers on keywords <1,000 monthly searches
• 76% of mobile on the same long-tail informational queries
• Mean AIO contains 11 outbound links
• Only 20-26% overlap between cited links and top-10 organic ranking
• #1 organic result appears in only 46% of desktop AIOs / 34% of mobile

Ranking #1 no longer secures the citation slot. What works: long-form authoritative content, statistic density, expert quotation, primary citations.

Yes, Google confirmed on May 15 2026 that the same spam policies that apply to standard Search apply to AI Overviews and AI Mode.

Implication: cloaking, keyword stuffing, sneaky redirects, scaled content abuse, expired-domain abuse, and site reputation abuse all carry penalty risk in the AI surfaces.

Princeton GEO finding ('keyword stuffing reduces visibility 8%') aligns. The 2026 differentiator is content quality, entity authority, and statistic density, not algorithmic shortcuts.

Google deprecated FAQ rich results effective May 7 2026. From developers.google.com: "As of May 7 2026, FAQ rich results are no longer appearing in Google Search. We will drop the FAQ search appearance, rich result report, and support in the Rich results test in June 2026." Search Console API support ends August 2026.

QAPage is the replacement when a page allows multiple answers.

HowTo structured data was deprecated September 2023 and should be removed entirely.

Still load-bearing for SERP features (26 active schema types per developers.google.com): Article, Breadcrumb, Carousel, Course list, Dataset, Discussion forum, Event, Image metadata, Job posting, Local business, Organization, Product, Profile page, QAPage, Recipe, Review snippet, Software app, Speakable, Subscription/paywalled, Video.

Two reasons:

1. The deprecation is Google-specific to SERP rich-result rendering; LLM extractors (Perplexity, ChatGPT Search, Claude, Gemini, Bing Copilot) continue to parse FAQPage entries for citation candidates.
2. The Q&A structure with author + dateModified per Answer is high-quality machine-readable text regardless of whether Google paints a rich result.

We are progressively migrating leapbuzz Q&A blocks to the QAPage type where a single primary answer fits, retaining FAQPage where a true list of multiple Q&A pairs is the structure. This page uses FAQPage; the Service pages will migrate as the next cleanup pass.

The 2026 baseline @graph for a regulated-sector page:

• Organization (with full sameAs to LinkedIn / Wikidata if available / regulatory registrations)
• Person (named author with credentials, jobTitle, sameAs to LinkedIn, knowsAbout)
• WebSite (with SearchAction)
• WebPage (with lastReviewed + reviewedBy + mainEntity + Speakable)
• Article or TechArticle (with author + dateModified + wordCount + editorialPolicy + correctionsPolicy)
• BreadcrumbList
• Service (with alternateName for keyword variations, hasOfferCatalog, areaServed × Country, Offers with PriceSpecification)
• QAPage where applicable (FAQPage if true Q&A list)

All cross-referenced via @id. Editorial + corrections policy referenced. Author bylined visibly with re-verification date.

For YMYL: every numeric claim needs a primary source in the same sentence, author credentials visible, compliance disclaimers dated.

Seven engines run the 2026 citation-tracking baseline:

1. Google AI Overviews (in standard Google Search)
2. Google AI Mode (standalone chat, launched March 5 2025)
3. Bing Copilot (with Microsoft AI Performance report, public preview Feb 2026. only first-party AI-citation analytics from a search engine)
4. ChatGPT Search (OAI-SearchBot)
5. Perplexity (PerplexityBot + Perplexity-User)
6. Claude (ClaudeBot + Claude-User + Claude-SearchBot)
7. Gemini app

Our citation-tracker polls all seven on a recurring prompt-set and reports citation share, sentiment, and competitor share-of-voice.

Microsoft launched the AI Performance report in Bing Webmaster Tools in public preview February 2026. Surfaces which URLs are cited in Copilot and Bing AI summaries and how citation activity changes over time.

As of May 2026, this is the only first-party AI-citation analytics available from a search engine. Google has not shipped an equivalent for AI Overviews or AI Mode.

The practitioner implication: Bing Webmaster Tools is now a mandatory tool in any 2026 SEO programme, not a Google-tool afterthought.

Three-tier policy:

• Tier 1 (training-bots, your choice): GPTBot, Google-Extended, CCBot, anthropic-ai. Block if you do not want training use.
• Tier 2 (inference / citation bots, NEVER block): OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, ClaudeBot, Claude-User, Claude-SearchBot, Google-Extended for AI Overviews, Gemini bots. Allow these explicitly.
• Tier 3 (standard crawlers): Googlebot, Bingbot. Allow with infrastructure-path disallow (/api/, /admin/).

The single most common mistake: blocking PerplexityBot but not Perplexity-User, which still ignores robots.txt on real-time user queries and produces citations the publisher cannot see in their bot logs.

llms.txt is a proposal by Jeremy Howard (published Sept 3 2024, llmstxt.org). As of May 2026: no formal endorsement from Google, OpenAI, Anthropic, or Perplexity. Adoption is grassroots (FastHTML, nbdev, Answer.AI, fast.ai). None of the AI engines have committed to reading it.

Google's December 2025 AI-features guidance explicitly says no new AI text files are needed.

Implementation costs ~30 minutes and signals practitioner currency. Expecting traffic or citation lift from llms.txt alone is unsupported by evidence. We ship it on every leapbuzz client site as a no-cost hygiene move, not as a lever.

Thresholds unchanged from 2024 (web.dev):

• LCP ≤ 2.5s is good
• INP ≤ 200ms is good (INP is the stable FID replacement, observes all interactions)
• CLS ≤ 0.1 is good

Measured at the 75th percentile of real users. No new metric in beta for 2026. Bing supports the data-nosnippet HTML attribute (October 2025) for paywalled or promotional content hiding without ranking impact.

CWV remains a ranking signal but its weight is diluted by the AI Overviews shift: a fast page that has nothing AI-citable will lose to a slower page with stronger entity, schema, and content signal. Our audit fixes CWV alongside the deeper plays, not as the only play.

The Aggarwal et al. paper (Princeton, arxiv 2311.09735, GEO: Generative Engine Optimization) reports GEO methods can boost visibility by up to 40 percent in generative engine responses.

The strongest individual levers:

• Statistic Addition: numerical density throughout content
• Quotation Addition: direct quotes from authoritative sources

Keyword stuffing is the inverse, reducing visibility by 8 percent. Domain variation is a caveat: lift is not uniform across query types.

Our content briefs apply Princeton patterns by default:

• Statistic density: 1+ verified numerical claim per 200 words
• Expert quotation: 1+ quoted authority per 500 words
• Primary-source citations (.gov,.edu, Google's own properties, named research firms)
• Declarative authoritative tone (low hedge density)
• Self-contained chunks of 50-150 words extractable as citations without context
• BLUF (Bottom Line Up Front) in the first 100 words of every page

Brand-mention to AI-citation correlation is the strongest correlated signal in 2026 AI search. Search Engine Land May 4 2026: "Why brand authority beats topical authority in AI search."

Practitioner mechanics:

• Digital PR via Featured.com (revived HARO brand since April 16 2025, sold from Cision to Featured.com after the December 9 2024 shutdown)
• Founder-led content on LinkedIn (Thought Leader Ads amplification)
• Guest contributions to authoritative publications in your vertical
• Podcast appearances
• University or association membership signals

The goal is not just backlinks but contextualised brand co-mentions: "leapbuzz, the AI-native consultancy based in Singapore" rather than just "leapbuzz" alone.

Look for three signals:

• AEO + GEO fluency, not just legacy SEO. Most Singapore agencies still optimise for Google rankings as the end-state; the 2026 buyer needs citations on AI Overviews, AI Mode, Bing Copilot, ChatGPT Search, Perplexity, Claude, and Gemini.
• Regulated-sector content discipline. MAS FAA-N03 (effective March 25 2026), FAA Section 36, ASIC RG 234 (updated November 2025), and OAIC Privacy Act all apply to digital content.
• Senior practitioner involvement on the account.

Our leadership team brings over 50 combined years:

• Siddharth Surana, Founder & CEO, 18+ years (ex-Havas Regional CDO, ex-Media360 COO)
• Sundeep Surana, Managing Director, 16+ years
• Ratnakar Nemani, Operations Director, 11+ years, Google Ads Certified
• Nitesh Sanghvi, Search & Social Director, 11+ years, Google Ads & Google Analytics certified

We work with Travel Guard Singapore and have anonymised proof: 6x digital sales growth for a regional banking institution across six APAC markets over seven quarters.

Multi-market sites get an additional layer in the audit:

• Hreflang implementation review (broken hreflang is the most common multi-market issue: ~40% of audited sites have it)
• International schema (Service.areaServed properly populated, Place node for "Worldwide" where applicable)
• Per-market Search Console and Bing Webmaster Tools verification
• Per-market citation-tracking baseline (prompt-set varies by market: queries differ across SG, MY, AU, US, CA)
• Regulatory disclosure per market: MAS FAA-N03 (SG), ASIC RG 234 (AU), OAIC Privacy Act (AU), FTC Endorsement Guides (US), Quebec Law 25 (CA), PDP Commissioner Malaysia + Bank Negara rules (MY)

The audit produces a per-market remediation plan, not a one-size-fits-all global plan.

Three patterns: audit-only, embedded strategist, or full takeover. We do not bid against the incumbent in pitches or steal accounts mid-contract.

Yes, when the incumbent contract is up. The handover process: 1-2 week diagnostic on programme state, formal handover of Search Console, Bing Webmaster Tools, analytics access, parallel-run period, full operational responsibility at the agreed transition date.

A senior practitioner from the leadership team is on every account. Strategic direction set by Siddharth Surana, Sundeep Surana, Ratnakar Nemani, or Nitesh Sanghvi. Team based in Singapore and India. One team, two time zones, no white-labelling.

Written findings covering:

• Technical SEO baseline (CWV scores mobile/desktop, site speed, indexing health, sitemap freshness)
• Schema audit (FAQ rich-results cleanup, QAPage migration, @graph/@id pattern, Organization sameAs coverage, Speakable applicability, Article completeness)
• robots.txt AI bot policy (3-tier review with specific changes recommended)
• Entity-graph coverage (Wikidata presence, LinkedIn entity, Crunchbase, sameAs ladder)
• Content audit against Princeton GEO patterns
• Citation-tracking baseline across 7 AI engines
• Competitor share-of-citation analysis
• Per-market regulatory disclosure review
• 90-day execution plan with prioritised work

Delivery: 2-3 weeks. Cost: fixed. Document yours regardless. Founder participates.

Three triggers.

1. The category has renamed itself. Search Engine Land (8 May 2026) called it 'recognition not rankings'; Condé Nast has forecast search traffic collapsing to single-digit share. If your team is still optimising for blue-link rankings as the end state, the playbook is misaligned.
2. FAQ rich results died on 7 May 2026 and your schema has not migrated to QAPage (and HowTo has been deprecated since September 2023, yet remains in most accounts we audit).
3. Your team has no AI-citation baseline across the 7 AI engines (Google AI Overviews, AI Mode, Bing Copilot, ChatGPT Search, Perplexity, Claude, Gemini). If you cannot answer 'are we being cited?' in any executive meeting, the visibility programme has not started.

fixed, 2-3 weeks. Findings document yours regardless of next steps.

SEO is not a cost-per-impression channel; it is a 6-18 month compounding investment.

SEO deserves the next dollar when:

• Your bottleneck is sustained organic visibility and AI-search citation share, where paid spend cannot keep buying the same audience
• Your bottleneck is brand recognition in AI Overviews (62 percent of citations come from pages NOT in the organic top-10, per Princeton-grade research)
• Your bottleneck is trust signals for financial-services content where Google's helpful-content updates penalise thin content harder

If your bottleneck is urgent demand capture this quarter, paid search wins. SEO does not deliver in 90 days. The audit reads your timeline plus your funnel position.

Three layers.

1. Business outcomes. Organic revenue contribution (Search Console organic impressions + sessions + assisted conversions in GA4), citation share across the 7 AI engines (Bing AI Performance report URL-level data plus our monthly prompt-set polling), brand-mention coverage from earned-media outreach (Featured.com revived HARO since April 16 2025) and founder-led LinkedIn content.

2. Attribution caveats stated up front. AI Overviews trigger primarily on long-tail informational queries (82 percent of desktop AI-Overview triggers happen on keywords under 1,000 monthly searches, per Semrush). Only 20-26 percent overlap between AI-Overview citations and top-10 organic ranking. The #1 organic result earns only a 46 percent desktop AI-Overview citation rate. Ranking is no longer the citation pipeline.

3. The bets. Schema migrations shipped, content pieces published with Princeton GEO patterns, brand-mention placements earned.

Two finance-grade adjustments.

First, fully-loaded cost. SEO investment includes content production + technical SEO + schema engineering + citation-tracking tooling + brand-mention earned-media work; not just management fees.

Second, payback shows up in two waves. Ranking lift and AI-citation lift at 90-180 days; then compounding organic revenue at 12-18 months as content earns links and entity authority.

Defensible benchmarks: positive net contribution by month 12 is excellent for small-to-mid-sized businesses; by month 18 healthy for enterprise financial-services where E-E-A-T amplification slows the climb. Beyond 24 months requires either reframing content scope (probably under-investing in production volume) or rebuilding the entity graph (Organization sameAs depth, Person schema with credentials).

Eight things 90 days before launch.

1. Technical SEO baseline: Core Web Vitals (Largest Contentful Paint ≤ 2.5s, Interaction to Next Paint ≤ 200ms, Cumulative Layout Shift ≤ 0.1 at the 75th percentile real users), sitemap.xml, hreflang for multi-market, mobile-first indexing.
2. Schema architecture in @graph format with cross-referenced identifiers (Organization with sameAs, Person with credentials, WebSite with SearchAction, WebPage with mainEntity, Article or TechArticle with author + dateModified + wordCount + editorialPolicy + correctionsPolicy, BreadcrumbList, Service, QAPage where applicable. Note: FAQPage rich results died 7 May 2026, though FAQPage schema remains useful for AI-engine extraction).
3. robots.txt 3-tier AI bot policy (training-bots: block your choice; inference-bots: NEVER block (OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, ClaudeBot, Claude-User, Claude-SearchBot, Google-Extended, Gemini bots); standard crawlers: allow with infrastructure paths disallowed).
4. Entity graph build (Organization sameAs across LinkedIn, Wikidata if eligible, Crunchbase).
5. Content production pipeline at 4-8 long-form pieces per month, each applying Princeton GEO patterns (statistic addition, expert quotation, primary citation, authoritative tone).
6. Citation-tracking baseline poll across the 7 AI engines.
7. Bing Webmaster Tools verification plus AI Performance report integration (the only first-party AI-citation analytics from a search engine as of February 2026).
8. YMYL compliance review on every page (in Singapore: MAS FAA-N03 for financial-services; ASIC RG 234 for Australia; OAIC Privacy Act; FTC Endorsement Guides for the US) with a named author byline and a reviewed-date visible.

SEO is not a CPM channel; it is a content + schema + technical + entity investment that compounds over 6-18 months. Realistic expectations: technical SEO + Core Web Vitals fixes take 4-12 weeks depending on engineering bandwidth. Schema rebuild takes 2-6 weeks. Content programme producing 4-8 long-form authoritative pieces per month takes 90 days to start showing AI Overview citations and 180 days to show ranking lift on competitive terms. Brand-mention building via digital PR + founder-led content takes 6-12 months to compound into measurable AI-citation share lift. The audit gives you the 90-day execution plan; the managed engagement runs that plan for 12 months minimum to fairly assess the programme.

Look for three signals. First, AEO + GEO fluency, not just legacy SEO. Most Singapore agencies still optimise for Google rankings as the end-state; the 2026 buyer needs citations on AI Overviews, AI Mode, Bing Copilot, ChatGPT Search, Perplexity, Claude, and Gemini. Second, regulated-sector content discipline. MAS FAA-N03 (effective March 25 2026), the Financial Advisers Act Section 36, ASIC RG 234 (updated November 2025), and OAIC Privacy Act all apply to digital content. Third, senior practitioner involvement on the account. Our leadership team brings over 50 combined years across SEO, paid social, and AI optimisation (Siddharth Surana, 18+ years Founder and CEO; Sundeep Surana, 16+ years Managing Director; Ratnakar Nemani, 11+ years Operations Director, Google Ads Certified; Nitesh Sanghvi, 12+ years Search and Social Director, Google Ads & Google Analytics certified). We work with Travel Guard Singapore and have anonymised proof: 6x digital sales growth for a regional banking institution across six APAC markets over seven quarters.

Every engagement is scoped to the data, industry, and market. We don't publish standard rates because no two engagements are the same. Talk to us about your specific challenge and we'll come back with a scoped proposal.