Measurement
Query bank expanded to 31 queries
Added commercial, vertical, and infrastructure queries across branded and unbranded intent. Tier breakdown is on the dashboard.
Transparency · In progress
Still being forged.
LuminaForge launched luminaforge.ai in May 2026. This dashboard is live while we collect baselines: Search Console Generative AI access, citation checks, performance checks, and schema coverage. Client engagements receive the same reporting model. Numbers below are early. We publish them anyway.
Last updated · Baseline collection since
Still being forged
We are collecting baselines, not polishing a vanity score.
LuminaForge went live recently. Google AI impressions (Search Console), LLM citations, and entity authority compound over weeks, not on launch day. This page logs what we measure, what we see today, and how it changes over time.
- Dashboard live since
- Citation sample
- 1/246 url (95% range 0% to 2%)
- Query bank
- 41 queries · 3 engines
Indicative. Not a ranking.
last run 2026-11-09
- Google path: Search Console Generative AI reports when the property is in Google's rollout cohort
- Multi-engine path: Gemini, Grok, and Perplexity. A missing key or a failed call is not measured this run.
- Performance and schema sections reflect what is verifiable now (PageSpeed Insights + CrUX field data + live route audit)
Early snapshots are the point. Radical transparency means showing the work before the numbers look impressive.
Performance targets are verified separately below, see Core Web Vitals and live PSI source links.
Forge log
What we shipped to earn citations
What we changed each week and why. Updated as we ship.
- Measurement
Early-phase transparency reframe
Dashboard copy now leads with baseline collection and the readiness scorecard, an honest 0% while entity authority compounds.
- Infrastructure
Gemini refresh quota cap
Refresh script caps at 19 Gemini calls per run on free tier and skips remaining checks after quota exhaustion.
- Entity
Directory submissions prepared
Clutch, G2, and DesignRush submission targets are documented. Profile URLs will flip the readiness scorecard when live.
- Content
GEO pillar content shipped
Published GEO agency pricing guide and home-services GEO pillar with FAQPage schema and internal links to case studies.
- Entity
Founder Person schema + LinkedIn sameAs
Nick Hennemann entity wired on /about with Person JSON-LD, CEO title, and LinkedIn profile in Organization sameAs.
- MeasurementMilestone
First citation snapshot archived
Initial Gemini + Grok baseline run committed to data/transparency/current.json with weekly GitHub Action refresh.
- MeasurementMilestone
Transparency dashboard live
Published /transparency with citation tracking, PageSpeed Insights snapshots, and schema coverage, the same reporting model client engagements receive.
Readiness scorecard
What we can prove today , 4/7
Citations compound over weeks. These infrastructure signals are verifiable now while the citation baseline builds.
Search Console Generative AI report
PendingRolling out in Search Console 2026. Performance, AI Overviews and AI Mode.
Verifyllms.txt (non-Google)
ReadyOptional file for non-Google tools. Google Search ignores llms.txt. Weight 0.
VerifyAI crawlers allowed
ReadyGPTBot, PerplexityBot, ClaudeBot, Google-Extended, and others allowed in robots.ts.
VerifyClutch / G2 / DesignRush profiles
PendingSubmission targets prepared. Profiles pending.
0 of 41 measured checks cited by URL. Indicative.
3 engines
Entity confusion: 3/41
Trend
Snapshot history
27 snapshots from 2026-05-25 to 2026-11-09.
Citation sample
1/246 url (95% range 0% to 2%)
Indicative. A stable share needs 5 or more runs across 3 or more days. Older single runs are not charted as a rate.
LCP (mobile lab)
1.1 s
PageSpeed Insights
Performance
Core Web Vitals on the reference build
Lab scores from PageSpeed Insights mobile (Slow 4G, Moto G Power), not desktop. Verifiable now, independent of citation baselines.
Schema coverage
Structured data deployed across luminaforge.ai
Computed from the live route map, infrastructure we can verify today while citations compound.
- Organization + WebSite
- 1/1
- FAQPage
- 34/34
- BreadcrumbList
- 47/47
- Article
- 33/33
- Service
- 7/7
pages with valid JSON-LD
pages with valid JSON-LD
pages with valid JSON-LD
pages with valid JSON-LD
pages with valid JSON-LD
Query tiers
Citation sample by intent
One blended number hides the story. Branded, commercial, and vertical queries are tracked separately. Unmeasured tiers stay blank.
Branded
1/18 url (95% range 1% to 26%)
3 queries · Does AI know LuminaForge exists?
Definitional
0/30 url (95% range 0% to 11%)
5 queries · Unbranded education queries: GEO/AEO category ownership.
Commercial
0/48 url (95% range 0% to 7%)
8 queries · Money queries: agency selection and pricing intent.
Location
0/60 url (95% range 0% to 6%)
10 queries · Metro-intent queries: city hubs and near-me agency selection.
Vertical
0/54 url (95% range 0% to 7%)
9 queries · Industry-specific GEO: home services, SaaS, B2B.
Comparison
0/18 url (95% range 0% to 18%)
3 queries · Decision-stage queries: SEO vs GEO vs AEO.
Infrastructure
0/18 url (95% range 0% to 18%)
3 queries · Technical GEO signals: llms.txt, schema, crawlers.
Citation log
Not cited in sample
LuminaForge was not in the source list for the logged runs. The table shows each check.
Alternative sources
Who appeared when LuminaForge was not cited
Aggregated from engine source lists on not-cited checks, including wrong-entity matches we are actively disambiguating.
| Domain | Appearances | Engines | Example query |
|---|---|---|---|
| searchenginejournal.com | 15 | Grok | LuminaForge Answer Engine Optimization |
| perplexity.ai | 11 | Grok | What is Generative Engine Optimization? |
| semrush.com | 9 | Grok | What is Answer Engine Optimization? |
| luminaforge.comEntity confusion | 8 | Grok | LuminaForge GEO agency |
| arxiv.org | 7 | Grok | What is Generative Engine Optimization? |
| ahrefs.com | 7 | Grok | What is Generative Engine Optimization? |
| originality.ai | 7 | Grok | What is Generative Engine Optimization? |
| surferseo.com | 7 | Grok | Generative Engine Optimization agency llms.txt |
| developers.google.com | 6 | Grok | What is the difference between SEO, GEO, and AEO? |
| searchengineland.com | 6 | Grok | What is the difference between SEO, GEO, and AEO? |
Baseline collection
41 queries · 3 engines · 1/246 url (95% range 0% to 2%)
Gemini, Grok, and Perplexity are sampled only when a key is set. A missing key or a failed call says not measured this run. Indicative API answers, not the consumer apps.
Showing 5 commercial queries, expand for the full 41-query bank.
| Query | Gemini | Grok | Perplexity |
|---|---|---|---|
| How much does GEO cost | not measured this run | Not cited in sample 0/6 url (95% range 0% to 39%) | not measured this run |
| GEO agency pricing | not measured this run | Not cited in sample 0/6 url (95% range 0% to 39%) | not measured this run |
| Best GEO agency 2026 | not measured this run | Not cited in sample 0/6 url (95% range 0% to 39%) | not measured this run |
| Generative Engine Optimization agency United States | not measured this run | Not cited in sample 0/6 url (95% range 0% to 39%) | not measured this run |
| What does a GEO agency do | not measured this run | Not cited in sample 0/6 url (95% range 0% to 39%) | not measured this run |
AI crawler policy
Access policy is part of GEO
GEO is not only schema and blog posts. It is whether AI systems can access and cite your site. LuminaForge publishes its crawler policy and measures citations publicly.
What LuminaForge allows
We invite AI crawlers on the public marketing site. Admin routes and gated client previews stay blocked, everything else is readable for citation checks.
- GPTBot
- OAI-SearchBot
- ChatGPT-User
- ClaudeBot
- Claude-Web
- PerplexityBot
- Perplexity-User
- Google-Extended
- Applebot-Extended
- CCBot
- Bytespider
- Amazonbot
Signals vs full blocks
Many sites mix Content-Signal lines with Disallow rules. They are not the same thing and GEO agencies should know the difference.
| Signal / rule | Means |
|---|---|
| search=yes | Search and retrieval OK Content may be indexed and used to answer queries, similar to how search engines build an index for results and AI summaries. |
| ai-train=no | Training opt-out (when supported) Signals that content should not be used to train foundation models. This is an emerging standard, and not every crawler honors Content-Signal yet. |
| Disallow: / | Full block for that bot The crawler cannot fetch your pages at all, which usually prevents real-time citation in RAG, not just training. Stronger than ai-train=no. |
Methodology
How we measure
What counts as cited (multi-engine snapshots)
A URL citation is a luminaforge.ai link in the engine's source list. A text mention names the business without that link and is counted separately. Position means place in the source list, not a Google rank. The published figure is k out of n runs with a 95% Wilson range, labeled indicative.
Google Search Console (primary for Google AI)
Google Search Console Generative AI performance reports show impressions when your URLs appear in AI Overviews, AI Mode, and generative Discover features. The Search Analytics API has no generative type (Google docs updated 2026-08-11: web, image, video, news, discover, googleNews). LuminaForge stores those rows from a Search Console CSV export, not from invented AI_OVERVIEW API rows. Official docs: https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
How we evaluate third-party SEO advice
LuminaForge aligns Search & AI Visibility with Google's official guidance: foundational SEO for Google Search and generative AI features, Search Console as first-party measurement, and honest labeling of third-party citation snapshots. Source: https://developers.google.com/search/docs/fundamentals/third-party-seo
One run is not truth
Third-party AI answers change between runs, days, and apps. A stable share needs at least 5 runs across at least 3 days. Fewer runs stay incomplete. Overlapping ranges are reported as no clear change.
Third-party engines tracked
Supplementary API samples: Gemini, Grok, and Perplexity. These are not the consumer apps, and they do not replace Search Console for Google AI Overviews or AI Mode. A missing key is not measured this run.
Gemini free-tier cap
Gemini free tier allows roughly 20 requests per day. The refresh caps live Gemini checks at 19 per run. Checks past the cap, or a failed call, are not measured this run.
PSI lab data vs CrUX field data
PageSpeed Insights reflects lab conditions on a throttled mobile profile, good for regression testing. CrUX field data reflects real Chrome users over 28 days (good / needs improvement / poor distributions). We publish both when available.
Entity confusion
Wrong-domain lumina* sources (unrelated GitHub repos, luminaforge.com, and similar names) are tracked separately as entity confusion. That is a sign the brand is not yet distinct in AI retrieval.
Citation sample
URL citations are reported as k out of n valid runs, with a 95% Wilson range, and labeled indicative. Text mentions are separate. One query on one engine is one run. This does not replace Search Console Generative AI impressions.
llms.txt (non-Google)
Optional brief at /llms.txt for tools that choose to read it. Google Search ignores llms.txt. It has zero weight as a Google or Google AI signal.
AI crawler access
Third-party AI crawlers (GPTBot, ClaudeBot, PerplexityBot) are explicitly allowed for non-Google citation paths. Google AI Overviews and AI Mode use the Search index via Googlebot. See robots.txt on this site.
FAQ
Questions about the dashboard while we collect baselines.
Plain answers about how the work actually runs.
6 questions
What does the LuminaForge transparency dashboard show?
Citation checks across Gemini, Grok, and Perplexity for a defined query bank, Core Web Vitals from PageSpeed Insights, CrUX field data when available, and structured-data coverage computed from the live route map. A missing key is not measured this run. Client engagements receive the same reporting model.
Why does the dashboard say not measured this run?
LuminaForge only publishes a citation number after a real check. If Gemini, Grok, or Perplexity has no key, or the call fails, that engine stays blank. We do not copy an older answer forward and call it this week's result.
Why does Grok show the wrong LuminaForge?
LuminaForge.ai launched recently. AI engines often conflate similar names, such as unrelated GitHub repos, luminaforge.com, or other Lumina-branded projects. We track these as entity confusion in the alternative sources panel and resolve them with a named founder entity, sameAs profiles, directory listings, and consistent NAP.
How often is citation data updated?
LuminaForge runs the conversational query bank on a weekly refresh cycle (Mondays UTC via GitHub Action), publishes snapshot deltas, and expands the bank as new content ships.
Can I get this dashboard for my own website?
Yes. When Search & AI Visibility is in scope, LuminaForge includes citation tracking and performance reporting, including the early baseline period while numbers compound. Scope is custom after we understand the need.
Why publish performance and citation data publicly?
LuminaForge dogfoods its own methodology. Public dashboards make every claim auditable, including before the citation numbers look good. The same standard applies to client work.
Ready when you are
Start with a process review.
Book a process review or request a free AI Visibility Snapshot. We respond within one business day. No pressure, no jargon deck.