Measure every answer engine
Your real buyer questions, run through ChatGPT, Google AI Overviews & AI Mode, Perplexity, Gemini, Grok (beta) and Claude — each grounded on the index it actually retrieves from.
See the engines →Measured across every major AI answer engine
Seven answer engines across five platforms. One calibrated score.
OpenAI
ChatGPT
Live
Anthropic
Claude
Live
AI Overviews · AI Mode · Gemini
Live
Perplexity
Perplexity
Live
xAI
Grok
Beta
Trademarks are the property of their respective owners. Lokrix AI measures these engines and is not affiliated with or endorsed by them.
Why AI visibility is measurable
AI answers now sit above the ten blue links — and each engine cites a different, measurable slice of the web. Optimizing for them is a discipline, and it starts with the numbers.
- 0%
- of ChatGPT citations match a Bing top-10 result
- 0%
- URL overlap between Google AI Overviews and AI Mode
- 0%
- of Perplexity’s top citations come from Reddit
- +0%
- citation lift from adding expert quotations
Sources: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4; Princeton “GEO: Generative Engine Optimization” study (Aggarwal et al., KDD 2024).
01The problem
Seven engines. Seven retrieval behaviours. One score that respects that.
ChatGPT leans on Bing. Google AI Overviews and AI Mode both pull from Google — yet cite the same URL only ~14% of the time. Perplexity runs live search and over-weights Reddit. A credible GEO score has to model each engine as it really behaves, not average them into mush.
- ChatGPT
- Bing
- Bing top-10 rank + freshness
- Google AI Overviews
- Google top-20; separate from AI Mode
- Google AI Mode
- Shares a cited URL with AIO only ~14% of the time
- Perplexity
- Google + Bing (live)
- Over-weights Reddit (~47% of top citations) + freshness
- Gemini
- Google Search grounding; schema-sensitive
- Grok (xAI)
- X + live web
- Grounds on X posts + live web; freshness-weighted (beta — live via a web-search fallback; draws not yet counted in headline scores)
- Claude
- Parametric + ClaudeBot
- Factual density and entity clarity are the primary levers; parametric knowledge with shallow ClaudeBot live search
Where each engine’s citations come from
Share matching that index, where published
Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.
02How it works
From a URL to a probability you can defend.
Build the prompt universe
Lokrix AI detects your brand and industry and auto-generates the real questions buyers ask AI engines — the prompts that actually decide whether you’re surfaced.
Sample every engine, K times
Each prompt runs through every engine with browsing/grounding on, K times across temperature, persona and locale, capturing the answer and its citation cards.
Report a calibrated score + CI
Mentions and citations are detected (URL + brand NER + semantic match) and turned into a Monte-Carlo presence probability with a 95% confidence interval.
Citation lift by content lever
Relative, vs. an unoptimised page
Source: Princeton “GEO: Generative Engine Optimization” study (Aggarwal et al., KDD 2024).
03What actually moves the needle
Proven levers — and the honesty to skip the ones that don’t work.
Peer-reviewed research is clear about what raises your odds of being cited: quotations +41%, statistics +40%, authoritative citations +30%. Lokrix AI turns these into prioritized, simulated recommendations for your pages — and will tell you plainly that llms.txt has no measured effect, so you don’t waste a sprint on it.
See how uplift is simulated →04The report
Every number carries its confidence interval.
Lokrix AI Score over time
With its 95% CI band
This is the shape of a Lokrix AI report: a visibility trend where the shaded band is the honest 95% confidence interval — because a single AI answer is a sample, not a fact.
Regressions are flagged. Provenance is badged on every point — live, cached, or predicted. The instrument never dresses a sample as a live measurement.
05The platform
Measure, explain, improve, and automate AI visibility.
A calibrated probability, not a guess
Every prompt is sampled K times across temperature, persona and locale. You get a Monte-Carlo presence probability with a 95% confidence interval — non-determinism handled honestly.
How scoring works →A simulated path to #1
The optimization engine ranks concrete moves by predicted visibility uplift — add quotations, statistics, cited sources, earned-media corroboration — and simulates the lift before you spend a word.
See the optimization engine →Track it over time — and via API
History, trends, regression alerts, and competitor share-of-voice. The whole engine is exposed through a REST API and an integrated MCP server for your own agents.
Explore the platform →Data honesty as design
We show the confidence interval. That’s the whole brand.
Anyone can print a big number. Lokrix AI samples, shows the interval, discloses the locale and persona, and badges whether each figure is live, cached, or modeled. You get a measurement you can take to a board — and defend.
06Pricing
Four plans. One credit currency.
Start free, then step up as you need more engines, prompts, and live credits. Every paid plan bundles a monthly balance of live answer-engine credits — predictive what-if simulations are always free.
Free
See whether AI answers already cite you — two engines, 50 one-time trial credits, no credit card.
Free
No credit card required
50 trial credits · 1 seat · upgrade to add credits
Start free- 1 project (site/brand)
- 2 answer engines tracked
- 10 tracked prompts
- Manual scans only
- 50 one-time trial live credits
- 1 competitor benchmarked
- 7-day history retention
- Predictive what-if / path-to-#1 previews (unmetered, 0 credits)
- Community support
- No API/MCP, no SSO, no overage purchase (must upgrade)
Starter
One brand, tracked properly: three engines, weekly scans, and read-only REST API / MCP access.
$49/mo
Billed monthly
700 credits/mo · 2 seats · overage $9/100
Start with Starter- 1 project
- 3 answer engines tracked
- 100 tracked prompts
- Weekly scan cadence
- 700 included live credits/mo
- 3 competitors benchmarked
- 3-month history retention
- Email alerts
- Read-only REST API / MCP access
- Manual optimization apply (no CMS publish)
- Unlimited predictive what-if previews (0 credits)
- SOC 2 report / DPA on request
- Email support
- Overage credit packs at $9.00/100
Growth
Most popularEverything, on autopilot — all seven engines, daily scans, agentic fixes, full REST API / MCP and BYOK.
$199/mo
Billed monthly
3,000 credits/mo · 5 seats · overage $8/100
Start with Growth- 3 projects
- All 7 answer engines tracked (ChatGPT, AI Overviews, AI Mode, Perplexity, Gemini, Grok, Claude)
- 500 tracked prompts
- Daily scan cadence
- 3000 included live credits/mo
- 10 competitors benchmarked
- 13-month history retention
- Full analytics + segments
- Email + Slack alerts
- Full REST API / MCP access
- Agentic optimization fixes + CMS publish
- BYOK: bring-your-own provider keys to route some engines direct-to-provider
- Confidence intervals + change-point detection
- 99.9% SLA, SOC 2 / DPA
- Optional SSO/SAML add-on
- Priority email support
- Overage credit packs at $8.00/100
Enterprise
Custom volume with SSO / SAML / SCIM, region pinning or on-prem, and a dedicated success manager.
From $2,500/mo
Custom, negotiated per contract
30,000 credits/mo · custom seats · overage $6/100
Contact sales- Custom / unlimited seats and projects
- All 7 engines + custom engine connectors
- Custom tracked-prompt volume
- Sub-daily / custom scan cadence
- 30000 committed live credits/mo
- Custom competitor volume
- Custom history retention (per DPA)
- Full analytics + revenue attribution
- Alerts + custom routing / SIEM
- Agentic fixes + white-label reports & client portals
- BYOK-or-Pay hybrid: tenant keys per engine, platform fee only on BYOK calls
- SSO / SAML / SCIM included
- Higher API/MCP limits + SLA, region pinning / on-prem option
- SOC 2 report + security review, custom DPA
- Dedicated CSM + support tiers
- Volume overage at $6.00/100
All prices in USD. Live-credit consumption is metered per answer-engine response — see how credits work below.
07FAQ
Straight answers.
- What is Generative Engine Optimization (GEO)?
- GEO is the practice of making a brand more likely to be cited, linked, recommended, or named inside AI-generated answers (ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini, Grok (beta), Claude). It is to answer engines what SEO is to the ten blue links.
- How does Lokrix AI handle the fact that AI answers are non-deterministic?
- Lokrix AI never reports a single run as a probability. It samples each prompt K times across temperature, persona, and locale, then reports a Monte-Carlo presence probability with a 95% confidence interval. Non-determinism is treated as a first-class measurement, not a bug.
- Which answer engines does Lokrix AI measure?
- Seven, each grounded on the index it actually retrieves from: ChatGPT, Google AI Overviews and AI Mode (treated separately), Perplexity, Gemini, Grok (beta), and Claude. Grok is live — it returns answers via a web-search fallback — but kept conservatively in beta, so its draws are not yet counted in headline scores.
- Is llms.txt a ranking factor for AI visibility?
- No. There is no measured effect of llms.txt on AI citation rates in current research, so Lokrix AI never sells it as a lever. The proven levers are adding quotations, statistics, and cited sources, plus earned-media corroboration.
- Does Lokrix AI show real live data or estimates?
- Both, always labelled. Every score badges its provenance — live engine run, cached, or predicted by the calibrated model — and every probability shows its confidence interval. Sampled or mock data is marked explicitly and never dressed up as a live measurement.
Find out if AI answers name you.
Run a free report on your site, then unlock live multi-engine scans, history, and the optimization engine.
Start free · 50 trial credits · no credit card