D2 Technology · Search OS

Website Health, SEO, AEO and GEO need one evidence model — not four disconnected dashboards.

D2 Search OS is a multi-client website health and search visibility operating system that separates observed performance, controlled benchmarks and deterministic diagnostics before turning them into actions or measured outcomes.

Direct answer

What does Search OS actually control?

Search OS gives D2 and authorized client users one operating layer for website health, technical search evidence, SEO performance, AEO readiness, controlled GEO observations, actions and later outcome measurement. The important design choice is that these signals do not become one opaque visibility score: the system keeps the evidence class, source, scope and missing-data state visible.

Core model: Observe what happened → benchmark what can be repeated → diagnose what can be checked deterministically → act only inside the declared evidence boundary.

Open Search OS Authenticated app · noindex

Observed performance

First-party or provider-reported evidence such as Google Search Console, GA4, Website Health telemetry and imported provider performance data.

Controlled benchmarks

Repeatable search and AI observations run against declared prompts, engines, locations or tracked queries so the comparison boundary remains visible.

Deterministic diagnostics

Search Auditor and AEO checks calculated from website structure and declared rules rather than inferred from a black-box visibility score.

Search intelligence layers

From uptime and HTML truth to query performance, answer readiness and AI visibility evidence.

Website Health

First-party synthetic checks plus privacy-minimized browser telemetry for availability, HTTP state, response timing, Web Vitals and browser-side errors.

Technical Search Auditor

Robots, sitemap, redirects, titles, descriptions, H1, canonicals, indexability, internal links, hreflang, JSON-LD, content hashes and technical issue detection.

SEO performance

Google Search Console query, page and query-to-page history combined with GA4 landing-page and conversion evidence where configured.

Keyword & authority intelligence

Tracked rankings, keyword intelligence, competitor coverage and backlink/referring-domain evidence from configured search-data providers.

AEO readiness

Deterministic page and query-to-page checks for direct answers, intent alignment, structure, evidence, entity clarity, FAQ, schema, links, freshness and expertise signals.

GEO intelligence

Controlled buyer-intent prompt benchmarks across configured AI engines with mention, citation, cited URL, source-domain and engine-coverage evidence.

Actions & prioritization

Deterministic opportunities from measured conditions such as low CTR, striking-distance queries, traffic declines, technical issues, conversion gaps and GEO visibility gaps.

Change → Outcome

Declared website changes can be measured against scoped pre-change and post-change windows without silently attributing unrelated site-wide metrics to one action.

Operating architecture

Managed site + search sources + controlled providers → Search OS → actions and outcomes.

The current product architecture combines first-party site monitoring, Search Auditor evidence, Google search and analytics sources, configured search-data providers and controlled AI engines behind a native Search OS API and Supabase-based auth, RLS, Vault, history, queue and scheduler layer.

Site evidence

Synthetic Website Health, D2 Monitor telemetry, Search Auditor and declared site-change events.

Search & analytics evidence

GSC, URL Inspection, GA4, rankings, keyword intelligence, backlinks, PageSpeed/CrUX where configured.

Controlled AI evidence

Prompt-engine observations for ChatGPT, Perplexity, Claude and Gemini where agency provider credentials are configured.

Privilege boundary

Supabase Auth, RLS and encrypted agency credential storage keep tenant access and provider secrets outside public browser state.

Action layer

Measured conditions become explicit opportunities, fixes and operating priorities instead of an unsupported success probability.

Outcome layer

Scoped changes can be revisited against later search, analytics, AEO and GEO evidence when the relevant data exists.

Operating loop

Collect → Normalize → Classify → Diagnose → Prioritize → Measure.

Collect

Ingest site telemetry, crawler evidence and configured provider data.

Normalize

Map URLs, queries, pages, prompts, providers and time windows into comparable records.

Classify evidence

Keep observed, controlled and deterministic evidence visibly separate.

Diagnose

Turn technical, SEO, AEO, GEO and measurement gaps into explicit findings.

Prioritize

Rank actions from measured conditions and declared operating rules rather than opaque success predictions.

Measure

Connect scoped changes to later evidence while preserving missing or unresolved data states.

AEO · Answer Engine Optimization

AEO is treated as explainable readiness — not as a mysterious AI score.

Search OS checks page structure and query-to-page readiness with deterministic rules. A failed check should be explainable as an exact gap that can become a fix; it should not be presented as proof that one AI engine will or will not cite the page.

Direct answer

Does the page answer the intended question clearly enough to extract without reconstructing the whole page?

Intent alignment

Does the page actually satisfy the query intent instead of merely containing related keywords?

Section structure

Are headings and information blocks organized into inspectable answer units?

Evidence

Are factual or performance claims supported by visible source or first-party evidence?

Entity clarity

Are products, services, organizations and concepts named consistently enough to resolve what the page is about?

FAQ / Q&A

Are high-intent questions answered directly where FAQ structure is useful?

Schema

Does structured data represent the page truth without overstating eligibility, results or relationships?

Internal links

Can users and crawlers move from the answer to supporting services, evidence and deeper knowledge?

Freshness

Does the content expose meaningful update signals where recency changes the answer?

Author / expertise evidence

Is the page connected to a credible author, reviewer, organization or first-party operating context?

GEO · Generative Engine Optimization

Measure comparable AI visibility evidence without inventing a universal rank.

Controlled prompts and matched engine runs can answer narrower questions such as whether the brand is mentioned, whether a response cites a source, which URL is cited and which source domains recur. Search OS keeps those observations scoped to the benchmark that produced them.

Prompt coverage

Which controlled buyer-intent prompts have comparable observations?

Brand mention coverage

How often the managed brand is mentioned within the declared benchmark set.

Citation coverage

How often responses cite a source under the declared prompt and engine scope.

Cited URLs

Which URLs are cited in the controlled observations.

Source domains

Which domains recur as answer sources across comparable runs.

Engine coverage

Where the brand or owned sources appear across configured engines on matched prompts.

Gained / lost citation pairs

Changes on retested prompt-engine pairs rather than across incomparable runs.

Observed AI evidence

Provider-reported AI visibility imports remain separate from controlled benchmarks when available.

Actions & Change → Outcome

The operating goal is not another score. It is a traceable next action and a measurable later state.

Low CTR

Search demand exists, but the result may need title, description, intent or offer review.

Striking-distance query

A query already has measurable search visibility and may justify focused page-level improvement.

Traffic decline

Observed performance changed enough to trigger diagnosis before assuming the cause.

Technical issue

Search Auditor found a structural condition that needs an explicit fix or review.

Conversion gap

Search traffic exists but downstream GA4 behavior or conversion evidence is weaker than the operating expectation.

GEO visibility gap

A controlled benchmark identifies a prompt/engine gap that can be investigated without calling it a universal AI rank.

Measurement sequence

Declared change

28-day pre-change baseline

28D

90D

Outcome windows can include the metrics relevant to the declared change scope. If a page change has no related GEO prompt, Search OS should not manufacture a site-wide GEO attribution for that page.

Claim boundaries

Search intelligence is useful only if the system refuses to turn incomplete evidence into stronger claims.

AEO readiness ≠ AI ranking

A deterministic AEO score describes declared structural checks. It does not prove ranking inside ChatGPT, Claude, Perplexity, Gemini or another engine.

Controlled GEO benchmark ≠ universal visibility

Prompt and engine observations are scoped experiments, not a complete model of every user, engine or future answer.

GSC impressions ≠ AI impressions

Search Console evidence belongs to Google Search surfaces covered by GSC and is not silently relabeled as generative-engine exposure.

Provider-reported evidence ≠ deterministic diagnostics

Observed performance, controlled benchmarks and crawler-derived checks remain separate evidence classes even when displayed together.

Missing evidence ≠ zero

Missing CrUX, observed AI, provider, query-page or conversion evidence remains missing instead of becoming a zero score.

Data confidence ≠ probability of success

Freshness and source health can describe data quality without predicting that a recommendation will work.

Change correlation ≠ causal proof

Change → Outcome windows support scoped comparison, but they do not automatically prove one change caused every movement in traffic, ranking, conversion or AI visibility.

Technical completeness ≠ rich-result eligibility

Schema and structured-data heuristics are diagnostics; they are not guarantees of search features or answer-engine inclusion.

Monitoring architecture ≠ uptime guarantee

Website Health can measure checks, incidents and recovery evidence, but architecture alone does not establish an SLA.

Product capability ≠ guaranteed growth

Search OS provides measurement, diagnostics and operating controls. It does not guarantee traffic growth, rankings, citations, conversions or revenue.

Product access & governance

The public product page is indexable. The client application is not.

Client and agency access

super_admin — full D2 agency control

agency_admin — client, site, user and integration operations

client_admin — operate assigned client sites

client_viewer — read-only access to assigned client data

Credential and data boundary

Tenant data is protected by Supabase Auth and RLS

Agency provider credentials are stored server-side or in encrypted Vault storage

Saved secret values are not returned to the browser

The authenticated Search OS application remains noindex, nofollow, noarchive

Related knowledge

Use the product page as the entity. Use Methodology, Insights and Services for the surrounding evidence and delivery context.

D2 Technology

See how Search OS sits beside Commerce Control and the D2 n8n Production Stack.

Explore

Evidence & Measurement Methodology

Truth → Formula → Rule → Workflow → Outcome and the claim boundaries used across D2 systems.

Explore

Insights

Commerce and Automation knowledge designed around direct answers, source evidence and operational decisions.

Explore

Discuss Search OS

Scope a managed website, required integrations and the visibility questions the system should answer.

Explore

FAQ

Search OS, SEO, AEO, GEO and evidence boundaries.

What is D2 Search OS?

D2 Search OS is D2 Group's multi-client website health and search visibility operating system. It combines first-party monitoring, technical Search Auditor checks, Google Search Console and GA4 evidence, SEO intelligence, deterministic AEO diagnostics, controlled GEO benchmarks, actions and Change-to-Outcome measurement in one operating model.

Is Search OS an SEO audit tool?

It includes technical and AEO auditing, but its scope is broader than a one-time audit. Search OS also stores historical performance, provider data, website health, actions, incidents, changes and outcome windows so the system can support recurring operations.

Does Search OS monitor websites hosted only on Vercel?

No. The product architecture is hosting-provider independent. The monitored site can be on Vercel, Cloudflare, a VPS, WordPress, Shopify, Webflow or another platform as long as the required public and integration boundaries are available.

What data does Search OS use for SEO?

Depending on configuration, Search OS can use Google Search Console, URL Inspection, GA4, tracked rankings, keyword and backlink intelligence, Search Auditor technical evidence, PageSpeed and CrUX. Missing providers remain missing rather than being fabricated.

How does Search OS handle AEO?

AEO is handled as a deterministic readiness diagnostic. Current checks include direct answer quality, intent alignment, section structure, evidence, entity clarity, FAQ, schema, internal links, freshness and author or expertise evidence. Query-specific AEO can connect a real search query to its ranking or landing page.

How does Search OS measure GEO or AI visibility?

Search OS can run controlled buyer-intent prompts against configured AI providers and store comparable mention, citation, cited-URL, source-domain and engine evidence. These runs are controlled benchmarks, not universal AI rankings.

Does a higher AEO score guarantee more AI citations?

No. AEO readiness and GEO observations answer different questions. AEO describes structural readiness under declared checks; controlled GEO runs observe selected prompts and engines. Neither is a guaranteed future citation outcome.

What is Change → Outcome in Search OS?

A website change can be recorded with a declared scope and compared against a 28-day pre-change baseline and later 7-day, 28-day and 90-day snapshots where the required data exists. Page-level changes do not automatically inherit unrelated site-wide metrics.

Does Search OS use AI to invent recommendations?

The product can integrate AI providers for controlled GEO observations, but deterministic opportunities are derived from measured conditions and explicit rules. Evidence classes remain separated so AI output is not silently treated as source truth.

How are client accounts separated?

The current architecture uses organizations, clients, sites, Supabase Auth and row-level security. Roles include super admin, agency admin, client admin and client viewer, with privileged credentials kept server-side or in encrypted Vault storage.

Is the Search OS application itself indexable?

No. The authenticated application at search.d2group.co is intended for authorized users and is configured noindex. This public D2 Group technology page is the canonical public explanation of the product and its evidence model.

Does Search OS guarantee rankings, traffic, citations or revenue?

No. Search OS is a measurement, diagnostics and operating-control system. Any performance conclusion still depends on the underlying evidence, implementation quality, market conditions, platform behavior and the scope of the change being measured.

Search OS scope

Need one operating view for a managed website's technical health, search performance, AEO readiness and GEO evidence?

Discuss Search OS →