The AI Visibility System.
Five anchors. One question each. The framework I run with clients to move a brand from “ranked on Google” to “recommended by AI.”
Why five anchors.
Search isn’t one surface anymore. A brand has to be crawlable for Google, parsable for entity graphs, citable by LLMs, and recommendable by AI agents — all at once. The system breaks that work into five questions any team can answer, in order.
Each anchor is a clear job. Each job has signals you can measure. Together they make a checklist you can actually run.
Visibility.
Can they find you at all?
Before AI can recommend you, it has to know you exist. That sounds obvious. Most brands fail here anyway.
Visibility is the floor. Crawlable, indexable, surface-able. Not just on Google — across every surface that feeds the answer layer: Bing for ChatGPT, Google for Gemini, public web for Perplexity, plus the secondary sources (Reddit, YouTube, GitHub, news) that LLMs lean on.
If any of those can’t reach your content cleanly, you’re invisible to anything downstream. Fix Visibility first, always.
- Robots.txt and crawl budget
- Sitemap coverage & depth
- Indexation across Google, Bing
- Render & JS parity
- Core Web Vitals + LLM crawl agents
- Block patterns hiding key pages
- Orphaned content with no internal links
- Slow or broken render for AI bots
- Duplicate or canonical conflicts
- Missing OpenGraph & metadata
Structure.
Can they parse what you mean?
Crawling tells a machine you exist. Structure tells it what you are. Most sites stop at the former.
Modern AI search runs on entities, not strings. Your brand, your products, your services, your people — each is a node, with attributes and relationships. If those nodes aren’t clearly defined in your markup and content, you’re guessing what the model decides you are.
Schema.org isn’t optional anymore. Neither is consistent entity treatment across your own pages, your knowledge panel, and your third-party profiles.
- Schema coverage & validity
- Entity consistency across surfaces
- Information architecture & topic clusters
- Internal linking as semantic graph
- FAQ, HowTo, Organization markup
- Missing or broken Organization schema
- Inconsistent NAP across the web
- Page templates without semantic anchors
- Author & expertise markup gaps
- Knowledge graph mismatches
Trust.
Who vouches for you?
AI doesn’t trust your homepage. It triangulates from everywhere your brand shows up — and weights the rest of the web more than your own claims.
Trust is the layer that decides whether your content gets cited at all. LLMs preferentially surface entities with strong authority signals: third-party mentions, expert authorship, citations in reputable sources, consistent presence on platforms the model already trusts.
Owned content sets the table. Earned signals decide if the model serves it.
- E-E-A-T signals across surfaces
- Author profile & topic expertise
- Brand mentions (linked & unlinked)
- Citations in trusted sources
- Wikidata / knowledge graph presence
- Anonymous content with no author
- Missing or thin About / Team pages
- Gaps in third-party profile coverage
- Weak press / podcast / interview footprint
- Outdated Wikipedia / Wikidata entries
Recommendation.
Will it hand you back?
This is the new conversion. Not “did they click” — but “did the AI name you when someone asked.”
Recommendation is where the content work actually pays off. We write for retrieval: clear answers, specific examples, structured comparisons, the kind of writing an LLM can lift a passage from and cite. We earn placement in the sources the AI already trusts.
The output target isn’t a page-one ranking — it’s showing up in the answer box, in the AI summary, in the recommended list. Different surfaces, same job.
- Answer-shaped content density
- Comparison & “best of” coverage
- Reddit, YouTube, podcast presence
- How brand appears in ChatGPT / Gemini / Perplexity
- Citation patterns & source distribution
- Generic content with no specific claim
- Missing TL;DR & answer summaries
- Thin coverage of buyer-intent queries
- Outdated comparison content
- Weak presence in cited 3rd-party sources
Measurement.
Are you actually moving?
If you can’t see your AI visibility week-over-week, you can’t fix it. We measure the things that matter, on a cadence.
Traffic is a lagging indicator now. Brand mentions in AI answers, share of citations, presence in featured surfaces — those are the leading ones. The dashboard we build tracks all five anchors over time, so wins are obvious and regressions get caught early.
Visibility is a system, not a campaign. Measurement is what makes it operational.
- AI answer share of voice
- Citation count by LLM surface
- Branded prompt coverage
- Entity recognition accuracy
- Traditional rank as supporting signal
- Weekly scorecard email
- Monthly board-ready review
- Live dashboard for client teams
- Quarterly playbook refresh
- Slack/Teams alerts on movement
One framework. Five anchors. One sequence.
Score the brand against all five anchors. Get a baseline number for every surface.
Rank fixes by impact × difficulty. Visibility & Structure always come first.
Ship the fixes — content, schema, links, off-site presence — with clear owners.
Earn citations in the sources LLMs trust. Plant the brand in answer-shaped content.
Weekly scorecard, monthly review. Visibility becomes a number you can move.
Tools, not magic.
The framework is the strategy. These are the instruments we play it with — combined with custom queries against ChatGPT, Gemini, and Perplexity to track answer share weekly.
Screaming Frog · Sitebulb · Ahrefs Site Audit
For the Visibility layer: full-site crawl, render parity, log analysis.
Schema.org validator · Wordlift · Diffbot
For the Structure layer: build, validate, and connect the entity graph.
Muck Rack · Ahrefs · Brand24
For the Trust layer: track mentions, citations, third-party authority signals.
Custom prompt suite · Profound · Otterly
For Recommendation + Measurement: weekly AI answer share across surfaces.
Surfer · Clearscope · custom prompts
Answer-shaped briefs and content QA against the retrieval target.
Looker Studio · Notion · custom Slackbots
The weekly scorecard, the monthly review, the alerts when something moves.