Generative Engine Optimization (GEO) & AI Visibility: The Complete Guide
See how ChatGPT, Gemini, Perplexity, and Claude describe and recommend your brand, uncover the sources shaping AI answers, optimize your content, and monitor AI crawlers on your site.
Updated Nov 19, 2026 · 22 min read · ElixBrand Product Team
When buyers, journalists, or analysts research your category today, many of them ask an AI assistant before they ever open a search results page. There is no page two in an AI answer: the assistant either recommends your brand, mentions a competitor instead, or leaves you out altogether. ElixBrand's Generative Engine Optimization (GEO) workspace monitors how leading AI engines answer your buyers' questions, shows which third-party websites and articles drive those answers, helps you improve your own pages and publish citation-ready content, and tracks live AI crawler visits on your website. This guide walks through every tab, metric, and limit in plain language, followed by step-by-step playbooks.
1. What Generative Engine Optimization (GEO) is for
Traditional SEO measures where your website ranks in a list of blue links. Generative Engine Optimization measures whether AI assistants know your brand, trust your sources, and recommend you when someone asks a real question.
The AI Visibility workspace brings monitoring, competitive intelligence, citation PR, content readiness, and crawler analytics into one place to answer four questions PR, marketing, and SEO teams face every week:
- How often do ChatGPT, Gemini, Perplexity, and Claude recommend our brand for discovery, comparison, and buying questions—and where do rivals outrank us?
- Which third-party publications, review sites, forums, and blogs are AI engines citing when they build those answers?
- Are our own web pages structured, readable, and technically open to AI search bots—and what articles should we publish to close visibility gaps?
- Which AI bots are actually crawling our website right now, are they verified or spoofed, and are they respecting our robots.txt rules?
Where to find it
Open a project and choose AI Engine Visibility (Radar icon) under Insights & Intelligence in the left sidebar. Do not confuse it with GEO Intelligence (Globe icon), which maps geographic country-level media coverage. AI Visibility requires a Pro, Business, or Enterprise plan.
2. First-time setup: 1-Click Auto-Setup, Starter Packages & the Prompt Wizard
Before you monitor your first prompt, AI Visibility opens on a focused onboarding screen with no extra tabs to distract you. Once your first monitored prompt is saved, all six workspace tabs appear immediately.
You have three ways to create your first set of prompts:
- 1-Click Auto-Setup: Generates up to 2 tailored prompts for each monitored keyword in your project (capped at 50 prompts in one batch), saves them on ChatGPT and Gemini with Live Web Search enabled, and starts running verification audits right away.
- Instant Starter Packages: Three one-click prompt packs tailored to your primary monitored keyword—Direct Recommendation (top-of-funnel discovery), Competitor Head-to-Head (mid-funnel comparison), and Reputation & Reviews (bottom-of-funnel buying checks)—saved and queued automatically.
- Custom Prompt Wizard (2 steps): Choose a monitored keyword, describe your niche or category, set target Country (US, UK, Canada, France, Germany, Spain, Morocco, UAE, or Global), optional Metro/City, Language (English, French, Spanish, German, Arabic, or All), and Funnel Stage (TOFU, MOFU, or BOFU). Click Suggest Competitors or add rivals manually, select your target AI engines, keep Live Web Search (RAG) switched on, generate prompts, uncheck any you do not want, and save.
Configure your Website URL and Brand Aliases first
In Project Settings, add your official Website URL, owned domains, product names, and brand or competitor aliases. AI Visibility uses those identifiers to credit your brand accurately, separate owned domains from third-party citations, generate your llms.txt file, and verify your site's AI crawlers.
3. How audits work: engines, web search grounding & cadence
Each monitored prompt is sent to the AI engines you choose on a recurring schedule. ElixBrand captures the full natural-language response, every cited URL, and how your brand was positioned—even on runs where your brand was completely omitted.
Four AI engines are live today, with a fifth prepared for rollout:
- ChatGPT (OpenAI) & Google Gemini: Core engines selected by default on every new prompt.
- Perplexity AI (Beta) & Anthropic Claude (Beta): Available to enable on any monitored prompt or compare side by side in the Sandbox. Beta badges indicate that model responses or availability may fluctuate more often.
- Google AI Overviews: Listed in the engine picker as Coming Soon and disabled by default until enabled for the workspace.
- Live Web Search (RAG): Enabled by default so ChatGPT, Gemini, and Claude browse the live web with regional context just like a real user (Perplexity always searches the live web). Turning Live Web Search off tests what the model knows strictly from its internal training memory.
Tracking both live web-grounded answers and multiple AI engines reveals a crucial split: you might lead on ChatGPT thanks to strong press coverage while disappearing on Perplexity because competitors dominate recent community threads.
How every answer is graded
On scheduled and manual audits, an AI judge reads the response to detect your brand and aliases, record a 1-based rank only when the answer is an ordered list, classify the mention (Direct Recommendation, Comparison, Alternative, or Passing Mention), score brand-specific sentiment, write a short framing summary, flag toxic language, and discover competing brands automatically. If the judge is ever unavailable, a rule-based fallback ensures no audit is lost. Default scanning is Daily (every 24 hours) or Weekly (every 7 days), with on-demand runs available every 5 minutes per prompt.
4. Overview tab: KPIs, Executive Briefing & breakdowns
The Overview tab is your command center. Use the timeframe selector at the top right (Today, Last 7, 30, 90, or 365 days; defaults to 30 days) to update every KPI card, chart, competitive table, and ranking matrix at once.
- Executive Briefing: A collapsible leadership card summarizing your posture as Market Leader (70%+ visibility), Strong Contender (40–69%), Needs Attention (1–39%), or Omission Risk (0%). Expand it to read the Executive Synthesis alongside three pillars—Where you win, Omissions & threats, and Priority actions—plus a direct Draft GEO Article button. Briefings cache for 24 hours and can be refreshed manually by editors and admins.
- Four Headline KPIs: Visibility Score (% of successful audit runs that mention your brand), Citations Index (total source links cited across successful runs), Response Stability (how consistently prompts produce the same cited or omitted outcome), and Brand Sentiment (% of all successful runs with positive sentiment toward your brand).
- AI Engine Comparison & Recommendation Tone: Donut charts comparing visibility across each active engine and splitting successful runs into Positive, Neutral, and Critical brand tone.
- Funnel Stage & Regional SOV: Compare visibility across TOFU (Discovery), MOFU (Comparison), and BOFU (Commercial) prompts, plus Regional SOV by Country—and a third Metro/City SOV list whenever you monitor city-specific prompts.
- Share of Voice (SOV) Shifts Timeline: Plots your daily presence percentage against up to 3 tracked competitors over a rolling window (up to 90 days) with an overall shift badge. Days with no completed audits show explicit gaps rather than fabricated lines.
Competitor Matrix, Rival Dossiers & Prompt Ranking Grid
Below the overview charts, two deep-dive modules show exactly who is beating you and on which questions.
- Head-to-Head Competitor Recommendation Matrix: Switch between Rankings (sortable table), Engine Matrix (a 2D heatmap showing where each rival displaces you per engine), and Cards view. Filter by High Threat, Contested, or Advantage, or search for any discovered competitor—even rivals you did not list manually during setup.
- Competitor Dossier Drawer: Click any rival row or card to inspect their displacement rate, target win rate, average rank gap, per-engine breakdown, top cited domains, and exact displaced prompts (clicking any displaced prompt jumps directly to that prompt's audit logs). Use Copy Battlecard or .md to export a ready-to-share Markdown sales and PR battlecard.
- Rival Ranking Grid: Maps every tracked prompt against each live engine in a color-coded grid—You ahead, Rival ahead, Omitted, or No data. Toggle Only where a rival is ahead to build an instant hit list, and click any cell to inspect the full AI response.
5. AI Audits Feed: managing prompts, diffs & cross-engine comparison
AI Audits Feed is where you manage your monitored prompts day to day, trigger immediate checks, and read the exact wording AI engines generated.
Filter by keyword, Funnel Stage, Engine, Country, Active/Paused state, or quick status pills (All Prompts, Cited, Omitted, Errors, Pending). While a check is running, the engine chip pulses with a live progress indicator and refreshes automatically every 4 seconds.
- Row actions: Run an on-demand verification (subject to a 5-minute cooldown per prompt), edit prompt settings (text, category, funnel stage, country/city/language, brands, engines, Daily or Weekly frequency, Live Web Search), open the Inspection Drawer, launch the GEO Article generator pre-loaded with that prompt, pause or resume monitoring, or delete the prompt.
- Bulk actions & CSV export: Select multiple prompts (even across pages) to Run now, Pause, Resume, or Delete in one batch, or click Export CSV for a 19-column spreadsheet containing one row per prompt and engine.
- Inspection Drawer — Runs view: Filter by engine, click any past run to inspect the full response with inline clickable citation pills, check rank, sentiment, intent, toxicity flags, and competitors mentioned, and toggle Diff vs Prev Run for a line-by-line comparison showing what changed since the previous audit.
- Inspection Drawer — Compare engines view: Places the latest answers from ChatGPT, Gemini, Perplexity, and Claude side by side with Grounded vs Knowledge badges, model names, latency, and a Shared vs Unique Citation Domains breakdown.
- Inspection Drawer — History view: Displays a 90-day multi-line trend chart tracking overall and per-engine presence rates for that single prompt.
Pausing frees up your active prompt quota
Your plan's prompt limit counts active monitored prompts only. Pause seasonal or completed campaign prompts to make room for new ones while keeping all historical audit logs intact. Resuming a paused prompt re-checks your plan's active limit.
6. Citation Opportunities: turning AI sources into PR wins
AI engines do not invent every recommendation from thin air—when web search is active, they lean heavily on third-party articles, review portals, directories, and community discussions. Citation Opportunities analyzes the last 90 days of cited URLs across your audits and turns them into a prioritized PR outreach list.
At the top, four Share of Sources (SoS) cards show what percentage of your AI citations come from UGC & Forums (Reddit, Quora, GitHub, Medium), Review Sites (G2, Capterra, Trustpilot, Product Hunt), Digital PR & Media, and Owned Media, alongside an Export Citations CSV button.
- Matched Domains (left column): Lists your own Direct Domain plus Earned Media Drivers—third-party pages ElixBrand scraped and confirmed already mention your brand.
- Priority Source Targets & PR Gaps (right column): Lists third-party domains that AI engines cite without mentioning you yet. Each domain is classified by Impact (how often AI cites it) and Effort, with quick filters for Quick Wins (high/medium impact, low effort like community or self-serve sites), Strategic (high-authority media and analysts), Low Effort, and Mentions competitors.
- AI PR Pitch Studio: Click AI Pitch on any gap domain to draft an outreach message. Choose from 8 angles (Unlinked Mention, Competitor List, Outdated Info, Expert Quote, Free VIP Demo, Exclusive Research, Guest Article, Broken Link Fix), 4 tones (Short & Punchy, Friendly, Professional B2B, Value & Data First), and Email or LinkedIn InMail format. Save as a draft or click Copy Pitch & Mark Sent.
- Post-pitch tracking: Once a domain is marked Pitch Sent or Secured, ElixBrand watches future AI audits and displays automatic badges—such as Brand now mentioned or Still cited after pitch—so you can prove PR impact.
How Citation Opportunities connects to Relationships
When you mark a citation domain as Pitch Sent, ElixBrand automatically logs a pitch activity and updates the last-contact date on that publisher inside Creators, Publishers & Relationships—provided the outlet is already saved in your Relationships list (it does not create new relationship entries on its own). Conversely, on the Influencers → Publishers tab, toggle Cited by AI only to view outlets AI assistants cited in the last 30 days and see every cited page on each outlet's profile.
7. Content Optimizer: scoring pages, technical checks, llms.txt & GEO Articles
Winning third-party citations is half the equation. The other half is making sure your own website is easy for AI engines to crawl, understand, and quote accurately.
The Content Optimizer tab combines a four-mode Sandbox Editor (Paste Copy, Live Copy URL, Technical AI Diagnostic, and llms.txt) with the GEO Content Studio for drafting publication-ready articles.
Paste Copy, Live URL scoring, Technical AI Diagnostic & llms.txt
Switch between the four tabs in the Sandbox Editor depending on whether you are editing a draft, auditing a live page, checking technical access, or deploying an llms.txt directory.
- Paste Copy & Live Copy URL: Paste at least 50 characters of draft copy or enter any live HTTPS URL (ElixBrand fetches the page and evaluates up to 15,000 characters). You receive a 0–100 AI Citation Readiness score broken down across Content Structure, Factual Density, Semantic Clarity, and Answer Completeness. Below the editor, Content Check History keeps your last 20 checks and displays score deltas (+ / -) whenever you re-score the same page after editing.
- Technical AI Diagnostic: Enter any page URL for a deterministic 0–100 AI Indexability Score—50 points for robots.txt access across 5 AI crawlers (GPTBot, PerplexityBot, ClaudeBot, CCBot, Bytespider), 30 points for structured Schema.org data (JSON-LD or Microdata), and 20 points for a page <title> and <meta name="description">—paired with specific fix recommendations.
- llms.txt Generator & Deployment Verifier: Crawls your Website URL and sitemap (3 to 40 pages, default 20) to generate a clean, specification-compliant llms.txt file—plus an optional llms-full.txt companion (up to 200,000 characters) for deep reference indexing. Copy or download the files, publish them to your site root, and click Verify deployment to confirm https://yourdomain.com/llms.txt returns live plain text instead of an HTML 404 page.
GEO Content Studio: drafting ready-to-publish articles
When audits show your brand is omitted from high-value questions, GEO Content Studio helps your content team create structured pages designed for AI extraction.
- Pick a target query: Launch Content Studio from the Optimizer tab, from Priority Actions inside the Executive Briefing, or from any row in AI Audits Feed. Filter your monitored prompts to Omitted to prioritize genuine visibility gaps, or type a Custom Query.
- Choose a content architecture: Select—or let ElixBrand auto-detect—one of three proven formats: Head-to-Head Comparison (for vs and alternative queries), Buyer Decision Guide (for criteria and checklist queries), or Concept Explainer (for foundational questions). Pre-populated competitors and optional Brand Edge talking points keep the draft grounded in your real differentiators.
- Export clean Markdown & JSON-LD: Every draft opens in a three-tab viewer with an Inverted Pyramid Direct Answer hook at the top, full formatted Markdown (downloadable as .md with YAML frontmatter), and matching JSON-LD Schema ready for your CMS. Your project library keeps up to 50 recent drafts.
8. GEO Sandbox: live ad-hoc checks & side-by-side comparison
Sometimes you want to test a new positioning question, check a rumor, or compare how engines respond to a brand-new product name before committing it to your monitored prompt quota. GEO Sandbox queries live AI engines in real time without saving an audit log.
- Single-engine or Side-by-Side Compare: Query one engine at a time, or switch on Compare engines side by side to run every active engine in parallel (each engine run counts toward your daily AI usage budget).
- Customize context: Override Your Brand Name, add comma-separated Competitors, and pick the Funnel Stage you would use if you decide to save the prompt.
- Instant scorecard & 1-click promotion: Inspect whether your brand was recommended or omitted, mention position, sentiment, competitors mentioned, Live web search vs Model knowledge badge, latency, and cited links. When a test uncovers something worth tracking, click Save as Monitor Query to add it straight to AI Audits Feed.
Sandbox vs scheduled audit analysis
To return answers quickly during live testing, Sandbox uses lightweight rule-based matching for position and sentiment. Once you save a prompt to AI Audits Feed, scheduled and manual runs use the deeper AI judge for brand-level sentiment, framing summaries, and competitor discovery.
9. AI Crawlers: live bot analytics, IP verification & governance
Standard website analytics tools like Google Analytics run in a visitor's browser via JavaScript—which means they are blind to almost all AI crawlers. The AI Crawlers tab gives you server-side visibility into every AI bot hitting your website, whether it is fetching live answers or scraping your pages for model training.
- Connect in minutes (12 stacks + coding agent prompt): Click Setup Snippet (requires your Website URL in Project Settings) to copy lightweight, non-blocking middleware for Next.js, Vercel Edge, Cloudflare Workers, WordPress, Express/Node.js, SvelteKit, Nuxt, Astro, Django, Laravel, Python FastAPI, or cURL—or copy an all-in-one prompt for your coding agent. Project owners and admins can create or rotate tracking keys and click Send Test GPTBot Ping to verify end-to-end ingestion.
- 41 AI bots across 15 families: Traffic is classified automatically into Search & RAG (live retrieval bots such as ChatGPT-User, OAI-SearchBot, PerplexityBot, and Claude-SearchBot that power citations) versus Model Training (corpus scrapers such as GPTBot, ClaudeBot, Google-Extended, Bytespider, and CCBot).
- IP Verification & Spoofed Bot Detection: For six major providers—OpenAI, Anthropic, Perplexity, Google, Meta, and ByteDance—ElixBrand checks every bot's IP against published official IP ranges and reverse DNS. Real bots display a Verified badge; third-party scrapers pretending to be GPTBot or ClaudeBot are flagged as Spoofed.
- Top Pages, Path Drill-Down & Payload Inspector: Click any crawled URL to view its daily hit chart and visiting bots. In the live Log Stream (which refreshes every 15 seconds with filters for agent, status code, intent, and compliance), click Inspect on any visit to run an on-demand live HTML fetch (up to 150 KB), estimate LLM token count, and check server-side rendering, Schema.org markup, and heading hierarchy.
- AI Bandwidth & Server Overhead: Estimates total bandwidth, compute time, egress cost, and crawl velocity by provider family. Whenever your middleware does not send raw response size or timing, ElixBrand explicitly marks the card Estimated using transparent defaults (48 KB/hit, 120 ms, and $0.08/GB).
- Governance, WAF & CSV Export: Open the 3-dots menu to generate a tailored robots.txt file (Search Only, Allow All, or Block All presets across 9 bots), run a Live Site Validator parsing your live robots.txt across 19 AI bots, copy 1-click WAF blocking rules (Cloudflare, Vercel Edge, Nginx, or AWS WAF), or export up to 10,000 crawler log rows as a 13-column CSV.
Automatic robots.txt violation tracking
Once connected, ElixBrand checks your site's live robots.txt every 6 hours and flags every AI bot visit that ignores a Disallow rule as a Robots.txt Violation—so you can spot non-compliant scrapers and generate a firewall rule in one click.
10. Smart Alerts, weekly digests & Report Builder
AI Visibility plugs directly into ElixBrand's alerting, email digest, and executive reporting workflows so your team catches drops early and shares progress without manual screenshots.
- 6 GEO Smart Alert rules: In Smart Alerts, set project-wide alerts for AI Visibility Drop (default: 20-point drop over 7 days), Prompt Omitted (dropped to 0% after at least 50% presence), Competitor Overtake (rival leads by 10+ points over 7 days), Negative AI Sentiment, New Citation Domain (first seen in 30 days), and AI Audit Failure. Delivery supports email, Slack, Discord, Telegram, and custom webhooks.
- 2 automatic inline safety alerts: Even without custom rules, ElixBrand logs internal alert events whenever an AI judge flags negative or toxic language about your brand, or when a prompt with 50%+ prior visibility drops to 0%.
- 2 AI Crawler Alert rules: Trigger alerts on Robots.txt Violations (default: 10 disallowed hits in 24h) or AI Crawler Volume Spikes (default: 500 hits in 24h).
- Weekly AI Crawler & Visibility Digest: Every Monday at 08:00 UTC, projects with recorded AI bot hits receive a 7-day summary covering search vs training share, crawl health score, and recent hits. Toggle it anytime under Project Settings → Project Notifications.
- Report Builder (12 live blocks + 6-page template): Drag any of 8 GEO AI blocks (AI Visibility Executive Briefing, AI Share of Voice Trend, GEO AI Visibility, Funnel Stage AI, GEO AI Playbook, AI Competitor Dossier, Citation Outreach, Technical Diagnostic) and 4 AI Crawler blocks (AI Crawler Activity, Bot Family Distribution, Most Crawled Pages, Server Overhead & Bandwidth) into a branded PDF—or apply the ready-made 6-page Generative Engine Optimization (GEO) Audit template in one click.
Refresh your Executive Briefing before exporting a GEO report
The Report Builder's AI Visibility Executive Briefing block reads your latest cached briefing without waiting on a fresh AI pass. If your briefing is more than 24 hours old, click Refresh on the Overview tab before exporting your client PDF.
11. The numbers, plans and limits, explained
Every score, tier, and limit across AI Visibility follows a fixed formula or plan rule. Here is what each number means so you can defend it in any marketing, SEO, or executive meeting.
- Visibility Score (AI SOV)0–100%
Percentage of successful audit runs in the selected timeframe where the AI assistant mentioned your brand or aliases (mentioned runs divided by total successful runs).
Pro tip: Always compare Visibility Score alongside Funnel Stage and Engine splits—a 50% average often hides 85% leadership in one engine and 15% in another.
- Response Stability0–100%
Measures how consistent AI answers are per prompt. A prompt mentioned 100% of the time—or omitted 100% of the time—scores 100% stability; a 50/50 coin-flip scores 0%.
Pro tip: Low stability on a high-intent prompt means AI answers are volatile—strengthen third-party citations and structured pages to lock in consistent recommendations.
- Brand Sentiment0–100% Positive
Percentage of all successful audit runs in the timeframe where the AI assistant characterized your brand positively (omitted runs count as neutral, not positive).
Pro tip: Read Brand Sentiment alongside Visibility Score: low positive percentage often means omission rather than hostility—check the Tone donut to see actual Critical runs.
- Executive Briefing StatusMarket Leader · Strong Contender · Needs Attention · Omission Risk
Automatically assigned from your latest per-prompt visibility rate: 70%+ is Market Leader, 40–69% is Strong Contender, 1–39% is Needs Attention, and 0% is Omission Risk. Cached for 24 hours.
Pro tip: If your visibility is 0% or you hold no #1 ranked spots yet, ElixBrand intentionally keeps the Where you win list empty rather than inventing praise.
- Competitor Threat Score & Displacement0–100 · % head-to-head wins
Displacement (win rate against you) counts every run mentioning a rival where your brand was omitted or ranked lower. Threat Score weights that win rate by sample size, reaching full weight at 5 or more mentions.
Pro tip: Ignore high displacement percentages marked with a low-sample asterisk (<3 mentions) until a rival appears repeatedly across prompts.
- AI Citation Readiness0–100
Content Optimizer score evaluating a draft or live URL (50 to 15,000 characters analyzed) across Content Structure, Factual Density, Semantic Clarity, and Answer Completeness.
Pro tip: Re-score the same live URL after publishing edits—Content Check History tracks your exact point gain run over run.
- AI Indexability Score0–100
Deterministic Technical AI Diagnostic score: 50 points for robots.txt permission across 5 core AI bots (10 points each), 30 points for JSON-LD or Microdata schema, and 20 points for <title> and meta description.
Pro tip: Pair this page check with AI Crawlers → Robots.txt Governance (19-bot live validator) and llms.txt deployment verification for complete site readiness.
- IP Verification StatusVerified · Spoofed · Unverified
Matches crawler IP addresses against official CIDR ranges and Forward-Confirmed Reverse DNS for OpenAI, Anthropic, Perplexity, Google, Meta, and ByteDance.
Pro tip: Filter the live log stream by Spoofed Bot IP before writing firewall rules—you can block impersonators without hurting genuine AI search bots.
- Monitored Prompt Limits & SamplingPro: 25 · Business: 100 · Enterprise: 500
Maximum active monitored prompts per project (AI Visibility is unavailable on Individual plans). Paused prompts do not consume quota. Standard plans run 1 pass per check; Enterprise supports up to 3 runs per pass for deeper consistency sampling.
Pro tip: Use Daily scans on your highest-value commercial prompts and Weekly scans on broader educational queries.
- Cooldowns, Exports & Roles5-min run cooldown · 10k CSV rows · Editor+ / Admin+
Manual prompt runs have a 5-minute per-query cooldown; AI tools share a daily 150,000-token per-user budget. Crawler exports cap at 10,000 rows (up to 365 days). Owners, admins, and editors manage prompts and AI tools; analysts, viewers, and clients are read-only; only owners and admins can create or rotate crawler tracking keys.
Pro tip: Test new phrasing in GEO Sandbox or use bulk Pause/Resume in AI Audits Feed rather than repeatedly re-running the same prompt.
12. Scenario: auditing and winning a competitive category prompt
Sales tells you prospects keep asking ChatGPT and Perplexity for the top platforms in your category—and your biggest rival is showing up first while you are missing.
Launch a focused prompt pack
Open the Custom Prompt Wizard, enter your core product category, click Suggest Competitors, enable ChatGPT, Gemini, and Perplexity with Live Web Search on, and save your MOFU and BOFU comparison prompts.
Pinpoint where rivals outrank you
On the Overview tab, toggle Only where a rival is ahead in the Rival Ranking Grid, then open the top rival's Competitor Dossier to see which engines they win and which third-party domains AI cites in their favor.
Publish a citation-ready comparison page
In AI Audits Feed, click Draft GEO Article on your highest-value omitted prompt. Choose Head-to-Head Comparison, add your real product differentiators, export the Markdown + JSON-LD Schema, and publish it on your site—then score the live URL in Content Optimizer.
Track recovery across engines
Over the next 30 to 90 days, open the Inspection Drawer's History and Compare engines tabs—and toggle Diff vs Prev Run—to verify when your brand enters the answer and climbs in rank.
13. Scenario: turning AI citation gaps into earned PR
Your website already explains your product well, yet AI engines keep citing third-party roundups, trade magazines, and software directories that leave your brand out.
Isolate high-impact citation gaps
Open Citation Opportunities and filter Priority Source Targets by Quick Wins and Mentions competitors. Review the scraped page titles and excerpts to see the exact articles shaping AI recommendations.
Add priority publications to Relationships
Before pitching high-value outlets, open Influencers → Publishers, switch on Cited by AI only, and add those outlets to your Relationships workspace so every outreach touch is tracked alongside press history.
Generate and send tailored pitches
Back in Citation Opportunities, click AI Pitch on each target, select Competitor List or Unlinked Mention, choose your tone and Email or LinkedIn format, customize the draft, and click Copy Pitch & Mark Sent.
Prove earned-media lift inside AI answers
Watch the post-pitch status badges in Citation Opportunities—when an updated article gets re-cited by AI with your brand included, move the status to Secured and share the win in your monthly report.
14. Scenario: preparing your website for AI search without leaking training data
Your marketing team wants ChatGPT Search, Perplexity, and Claude to cite your documentation and blog—while legal and engineering want to block unauthorized model-training scrapers and spoofed bots.
Deploy crawler telemetry and verify ingestion
Confirm your Website URL in Project Settings, open AI Crawlers → Setup Snippet, deploy the middleware for your web stack (such as Next.js, Cloudflare Workers, or WordPress), and click Send Test GPTBot Ping until all verification rows turn green.
Audit what AI crawlers actually receive
Run Technical AI Diagnostic on your key landing pages, then open Payload Inspector on top crawled paths in AI Crawlers to confirm server-rendered HTML, Schema.org markup, and clean headings are visible without JavaScript.
Deploy llms.txt and a Search Only robots.txt policy
Generate and publish llms.txt from Content Optimizer (and click Verify deployment). Then open AI Crawlers → Robots.txt Governance, pick the Search Only preset to allow live retrieval bots while disallowing training crawlers, publish it, and run the Live Site Validator.
Block spoofers and set a violation tripwire
Filter the crawler log stream by Spoofed Bot IP and Robots Disallowed. Open Export WAF Rule to copy a Cloudflare, Vercel Edge, Nginx, or AWS WAF block rule for non-compliant bots, and create an AI Crawler Violation Smart Alert.
15. Scenario: delivering a monthly AI Visibility briefing to leadership
Your CMO asks for a clear monthly update on how the brand performs in AI search alongside traditional press and social coverage—without logging into dashboards.
Refresh the Executive Briefing and check 30-day trends
On the Overview tab, set the timeframe to Last 30 days and click Refresh on the Executive Briefing card so the latest leadership synthesis is ready.
Export sales-ready Competitor Battlecards
Open the Competitor Dossier for your top two rivals and download their Markdown battlecards to share directly with product marketing and sales enablement.
Build the 6-page GEO Audit PDF
Open Report Builder, apply the Generative Engine Optimization (GEO) Audit template (which pulls Executive Briefing, SOV Trend, Engine & Funnel Breakdown, AI Playbook, Competitor Dossier, AI Crawler Impact, and Technical Diagnostic automatically), add your cover branding, and export the PDF.
Keep stakeholders informed between monthly reviews
Leave the Weekly AI Crawler & Visibility Digest enabled in Project Settings → Project Notifications and set an AI Visibility Drop alert in Smart Alerts so sudden shifts surface immediately.
16. Common questions
Why is AI Engine Visibility locked on my project? Generative Engine Optimization (GEO) requires a Pro, Business, or Enterprise plan. Project owners on an Individual plan can upgrade from Billing; collaborators will see a prompt asking the project owner to upgrade.
What is the difference between AI Engine Visibility and GEO Intelligence in the sidebar? AI Engine Visibility (Radar icon) tracks how AI assistants like ChatGPT, Gemini, Perplexity, and Claude recommend and cite your brand. GEO Intelligence (Globe icon) maps where your traditional media and social mentions come from geographically by country.
Which AI engines are live today, and why do some show a Beta or Soon badge? ChatGPT (OpenAI) and Google Gemini are core engines enabled by default. Perplexity AI and Anthropic Claude are live in Beta and can be enabled on any prompt or in the Sandbox. Google AI Overviews is marked Coming Soon and remains disabled by default until activated.
How many prompts can I monitor, and do paused prompts count? Active monitored prompt limits are 25 on Pro, 100 on Business, and 500 on Enterprise. Paused prompts do not count against your active limit, though re-activating a paused prompt checks whether you have room.
Why did my Visibility Score change even though we did not edit our website? AI answers change dynamically when Live Web Search pulls new articles, reviews, or forum threads—or when an AI provider updates its model. Open the Inspection Drawer and use Diff vs Prev Run to see the exact wording and sources that shifted.
Why does Response Stability show 100% when my Visibility Score is 0%? Response Stability measures consistency, not visibility. If a prompt omits your brand on every single run, the engine is 100% consistent in omitting you—giving you a clear signal to create targeted content and earn citations for that question.
Why do some AI Crawler bandwidth and egress cost numbers say Estimated? Not every web framework or middleware reports raw response byte sizes and timings on every hit. Whenever unmeasured hits are present, ElixBrand labels the overhead card Estimated using transparent defaults (48 KB per hit, 120 ms response time, and $0.08 per GB).
Why didn't Google Analytics show the AI bots I see in AI Crawlers? Almost all AI search and training crawlers fetch raw HTML without executing browser JavaScript, so client-side analytics scripts never fire. ElixBrand's server and edge middleware captures those visits directly before or as your server responds.
If I mark a citation domain as Pitch Sent, does it automatically create a publisher in Relationships? No. Marking a domain Pitch Sent automatically logs a pitch activity and updates the last-contact date on that outlet if it already exists in Creators, Publishers & Relationships, but it never creates brand-new relationship records on its own.
Why can't I run a prompt again right after clicking Run? On-demand verification has a 5-minute cooldown per prompt (and cannot run on paused prompts) to prevent duplicate spend while the previous pass completes.
Can AI Visibility detect factual hallucinations about my brand? Today, audits grade brand mention, ranked position, recommendation type, brand-level sentiment, a concise framing summary, toxicity, and whether live web search was used—not automated fact-checking against an external knowledge base. Read the full response and framing summary in the Inspection Drawer to spot inaccuracies directly.
Who on my team can edit prompts, run Sandbox checks, or rotate AI Crawler keys? Owners, admins, and editors can create, edit, pause, delete, and run prompts, run Sandbox and Optimizer checks, manage citation pitches, and draft GEO articles. Analysts, viewers, and clients have read-only access. Creating or rotating AI Crawler tracking keys and sending test pings require owner or admin access.
See how AI assistants describe and recommend your brand
Track brand visibility across ChatGPT, Gemini, Perplexity and Claude, uncover the sources shaping AI answers, optimize your content and monitor AI crawlers.