Sphere Partners

SEO · GEO · LLM Findability

/dev/seogeollm · noindex · 2026-05-28

How discoverable is sphereinc.com to Google, AI Overviews/Perplexity/ChatGPT, and LLM agents — three independent layers, audited together so we can move them in lock-step.

Site at a glance · 949 live URLs

/blogs424/case-studies149/services115/events78/platforms43/whitepapers34/industries24/solution24/accelerators10/dev8/about-us6/design-system4/insights4/podcast2/1/ai-product-engineering1/authors1/blog-progress1/changelog1/claude-31/claude-figma-guide1/clutch1/comply-ai1/contact-it-company1/eu-ai-act-consulting1/legacy-modernization1/legacy-modernization-claude-21/miami-ai-development1/newform1/privacy1/search1/solar-revenue-loss-calculator1/solutions1/sphereiq1/terms1/thank-you1/thank-you-download1/thank-you-webinar1

SEO

GAP5 / 8 pass

Classic search engines — Google, Bing. Foundation strong, structured data thin.

Lighthouse marks
LCP≤ 2.5sunknownN/Aweb.dev
INP≤ 200msunknownN/Aweb.dev
CLS< 0.1unknownN/A
Canonicalevery pagepageMeta() / withCanonical()OK
OG cardsunique per route138 colocatedOK
Sitemapauto + per-section13 indexedLEAD
Organization + WebSite JSON-LDsite-wideemitted in layoutOK
Article JSON-LD on blogsall posts424OK
Service JSON-LD on /services/*all 1140MISS
FAQPage on services/industries≥ 1 / page3MISS
BreadcrumbListnested routes0MISS
hreflangif multi-localeEN-onlyN/A
Site map — findability by SEO
  • /services114 URLs
    OKcanonicalMISSService schemaMISSFAQOKOG
  • /blogs424 URLs
    OKcanonicalOKArticle schemaOKAuthor sameAsOKOG
  • /case-studies148 URLs
    OKcanonicalGAPArticle schemaOKclient mentionOKOG
  • /industries24 URLs
    OKcanonicalMISSService schemaMISSFAQOKOG
  • /solution22 URLs
    OKcanonicalMISSService schemaGAPFAQOKOG
  • /platforms43 URLs
    OKcanonicalMISSSoftwareApp schemaOKOG
  • /accelerators10 URLs
    OKcanonicalMISSProduct schemaOKOG
  • /events78 URLs
    GAPEvent schemaOKOrg organizer
  • /whitepapers34 URLs
    GAPArticleOKgated CTA
  • /insights4 URLs
    OKhub pageGAPlong-form
Who leads · technique
  • Globant — Service IA mirrored in /llms.txt → consistent retrieval surface
  • Thoughtworks — 3 segmented sitemaps in robots.txt (main / Radar / careers)
  • Persistent — Schema-marked services + canonical entity names in JSON-LD
Outlook
6 months
  • AI Overviews on ~48% of tracked queries (was 31% Feb-25) SQ
  • 43% of all searches end zero-click; 93% under AI Mode DA
  • Top-10-ranked pages: AIO citation share dropped 76% → 38% Ahrefs
2 years
  • Classic SEO = qualifier (top-20 to be eligible), not the driver
  • Schema combo Service + FAQ + HowTo + Breadcrumb = table-stakes on /services/*
  • INP regresses as React/Next ship more interactivity — must budget per route
Actions for Sphere
  1. Emit Service + FAQPage + BreadcrumbList JSON-LD on every /services/* and /industries/* (helpers already exist in src/utilities/jsonLd.ts — just wire them).
  2. Add a real-user web-vitals beacon (web-vitals/attribution → /api/vitals) so LCP/INP/CLS show up here instead of "unknown".
  3. Target 2,500+ words on ~10 cornerstone /insights/* pages (RAG, agentic AI, healthcare AI, etc.) — 1.6× citation lever.

GEO

MISS0 / 8 pass

Generative engines — AI Overviews, Perplexity, ChatGPT Search, Gemini. Almost no surface area today.

Lighthouse marks
Share of Model (SoM)tracked weeklynot measuredMISSFrase
AI Overview presence≥ industry medianunknownN/A
FAQPage answers (1-3 sentence)every solution page3MISS
HowTo schema on guides≥ 51MISS
Answer-first paragraphs (60w)cornerstone pagesnarrative-first todayGAP
Named-entity density (client, metric, year)every case-studypartialGAPstudy
Third-party citations (Reddit/HN/dev.to)monthly seednone trackedMISS
YouTube case-study walkthroughs≥ 12 / yrunknownGAP
Press / earned-media coveragetrackedmanualGAP
Site map — findability by GEO
  • /services114 URLs
    GAPanswer-firstMISSFAQMISSHowToGAPcite-worthy stat
  • /blogs424 URLs
    GAPanswer-firstOKnamed entitiesOKArticle
  • /case-studies148 URLs
    OKnamed clientGAPquant metricGAPArticle
  • /industries24 URLs
    MISSFAQGAPvertical-specific stats
  • /solution22 URLs
    GAPFAQMISSHowTo
  • /insights4 URLs
    GAPlong-formMISSoriginal research
  • /accelerators10 URLs
    MISSProduct schemaGAPdemo asset
  • /whitepapers34 URLs
    OKcite-worthy dataGAPgated
  • /podcast2 URLs
    GAPtranscriptMISSEpisode schema
Who leads · technique
  • Globant — Canonical product names (CODA, Enterprise AI) repeated identically site-wide → entity lock-in
  • Thoughtworks — Technology Radar — biannual artifact ChatGPT cites for trend queries
  • McKinsey Digital — Original research with named clients + quant outcomes — model-cited at high frequency
Outlook
6 months
  • ChatGPT ≈ 17% of global digital queries; 250–500M weekly searches DA
  • Gemini overtook Perplexity globally Jan 2026 (~29% lead in AI traffic) SERanking
  • 85% of brand mentions in AI answers come from third-party pages (Reddit/YouTube/press) Mindcentrix
2 years
  • GEO folds back into content strategy — every paragraph answers one prompt
  • AI Citation Source Index becomes a vendor metric (Similarweb / 5W already publishing) 5W
  • Third-party mentions outweigh owned-site SEO as primary lever
Actions for Sphere
  1. Build /answers/* — short Q&A pages, 60-word lead answer + FAQPage schema. Seed with the 30 prompts we lose to competitors.
  2. Re-frame every case-study lead with named client + quant outcome in the first sentence. Add Article+Result schema.
  3. Weekly Share-of-Model run: 30 prompts × 4 engines (ChatGPT/Claude/Gemini/Perplexity), parse citations, store in Payload, chart trend.

LLM

MISS2 / 8 pass

Agent-readable surface — llms.txt, AI bot policy, MCP. Currently invisible at the protocol layer.

Lighthouse marks
/llms.txtpresentmissingMISSllmstxt.org
/llms-full.txtpresentmissingMISS
robots.txt names AI botsGPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Applebot-Extendedno — wildcard onlyGAP
MCP endpointmcp.sphereinc.comnoneMISSMCP
Sitemap index (XML)present13 sub-sitemapsLEAD
Organization sameAs (LinkedIn, X, FB, YT)all socials4 platformsOK
Structured services feed (JSON)/api/services.jsonnoneMISS
RSS/Atom for /blogs and /insightspresentunknownGAP
OpenAPI / public API docsif any public APIN/AN/A
Site map — findability by LLM
  • llms.txt0 URLs
    MISSfileMISSservices taxonomyMISSgovernance clause
  • robots.txt0 URLs
    OKexistsGAPnames AI botsOKsitemap link
  • /sitemap.xml13 URLs
    OKindexOKper-sectionOKlastmod
  • JSON-LD entities0 URLs
    OKOrganizationOKWebSite + SearchMISSService
  • MCP server0 URLs
    MISSendpointMISStools
  • Machine-readable feeds0 URLs
    GAPRSSMISSJSON servicesMISSJSON case-studies
Who leads · technique
  • Globant — /llms.txt mirrors studio taxonomy (Enterprise / Digital Evolution / GUT / AI Studios)
  • Persistent — /llms.txt ships governance: attribution clause + training-data restriction + contact
  • Slalom — /llms.txt with structured firm overview (offices, partnerships, service areas)
Outlook
6 months
  • llms.txt adoption ≈ 10.13% of 300k domains audited — top decile if shipped Presenc
  • AI crawlers = 22% of all bot traffic; fastest-growing segment HUMAN
  • Anthropic crawl-to-referral ratio 20,583:1 — they read heavily, never refer back Cloudflare
2 years
  • B2B service sites expose MCP endpoints — list_services, get_case_study, request_intro
  • Either llms.txt gets IETF RFC, or MCP-native surfaces win (bet on MCP — interactive > static)
  • Agent traffic exceeds human traffic on B2B sites before generic e-commerce (Cloudflare projects 2027 generally) ALM
Actions for Sphere
  1. Ship /llms.txt — generate from Payload globals (company description + service taxonomy + industry map + case-study index + contact + attribution clause). Persistent-style.
  2. Explicit AI bot policy in robots.ts: Allow GPTBot/ClaudeBot/PerplexityBot/Google-Extended/OAI-SearchBot/Applebot-Extended; Disallow /admin /api /next /preview.
  3. MCP server stub at /api/mcp (or mcp.sphereinc.com) with 3 tools: list_services, get_case_study, request_intro. Even a stub leapfrogs every named competitor.

If we ship in one quarter — priority order

  1. /llms.txt + robots AI policy — 1 day. Top-decile globally, costs nothing, hits all three pillars.
  2. Service + FAQPage + Breadcrumb JSON-LD on every /services/* and /industries/* — 1 week. Helpers exist; just wire and seed copy. 2.3× citation lever.
  3. Real-user web-vitals beacon — 2 days. Turns the SEO “unknown” rows into measured numbers.
  4. /answers/* Q&A layer (50 pages, 60-word lead + FAQPage) — 3 weeks. The GEO surface we currently lack entirely.
  5. MCP server stub — 1 week. Pure differentiation — none of EPAM/Globant/Thoughtworks/Endava/Slalom expose one today.
  6. Share-of-Model tracker — 1 week. Without measurement, GEO work is invisible to the business.