7 enginesAnswer-engine intelligence

Every answer engine, measured on its own terms

Each engine retrieves from a different index, so a credible score models each one as it really behaves — not averaged into a single meaningless number.

ChatGPTGoogle AI OverviewsGoogle AI ModePerplexityGeminiGrok (beta)Claude
Start measuring freeHow Lokrix AI scores engines →

Seven engines · seven behaviours

Different indexes. Different biases. Different scores.

Lokrix AI runs your prompt universe through each engine separately, grounded on the index it actually queries — so you know exactly where you appear (or don’t), and why.

ChatGPT

Bing

Grounds primarily on the Bing index — Bing top-10 rank and page freshness drive citation selection.

87%

of citations match this index (published figure)

Lokrix AI watches

  • 87% of ChatGPT citations match a Bing top-10 result — Bing rank is the primary proxy signal
  • Keyword and entity coverage in Bing-indexed content

Google AI Overviews

Google

Grounds on Google’s own index; ~54% of citations land in the Google top-20.

54%

of citations match this index (published figure)

Lokrix AI watches

  • Google index depth and passage-level retrieval — 54% of citations sit in top-20
  • Schema markup and entity authority in Google’s Knowledge Graph

Google AI Mode

Google

Also grounds on Google, yet shares a cited URL with AI Overviews only ~14% of the time.

Lokrix AI watches

  • Scored as a distinct engine from AI Overviews — ~14% URL overlap demands separate measurement
  • Multi-step reasoning and query-reformulation patterns unique to AI Mode

Perplexity

Google + Bing (live)

Issues live cross-index queries on every request and over-weights Reddit (~47% of top citations) plus freshness.

47%

of citations match this index (published figure)

Lokrix AI watches

  • Reddit and earned-media corroboration — ~47% of top Perplexity citations trace to Reddit
  • Cross-index freshness: live queries mean stale content is penalised immediately

Gemini

Google

Grounds on Google Search and is particularly sensitive to structured data and entity signals.

Lokrix AI watches

  • Structured data (schema.org) depth and entity coverage in Google’s index
  • Semantic precision and authoritative source signals

Grok (beta)

Web + X

xAI model with real-time web search and X (Twitter) data grounding. Live via a web-search fallback, kept conservatively in beta — its draws are not yet counted in headline scores.

Lokrix AI watches

  • Real-time web freshness and crawl accessibility for xAI
  • X (Twitter) presence and corroboration signals

Claude

ClaudeBot

Anthropic’s ClaudeBot crawler builds the training corpus; tool-use and search integrations expand retrieval context.

Lokrix AI watches

  • ClaudeBot crawl accessibility, robots.txt policy, and content structure
  • Citation quality, factual precision, and authoritative sourcing patterns

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.

Published figures

How many citations actually trace back to each engine’s index.

Where a peer-reviewed or published citation-match figure exists, Lokrix AI uses it as a calibration anchor — shown below. Engines without a published headline % are still measured, just without this external benchmark.

Citations traced to each engine’s retrieval index

Published figures only — null engines omitted

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.

Quick-reference

Engine · retrieval index · strongest measurable lever

The table below maps each engine to the index it grounds on and the single most impactful signal Lokrix AI has observed for that surface. Use it as a diagnostic when a specific engine’s score diverges from the others.

ChatGPT
Bing
Bing rank · freshness · keyword coverage
Google AI Overviews
Google
Google rank · schema markup · entity authority
Google AI Mode
Google
Scored separately — only ~14% URL overlap with AI Overviews
Perplexity
Google + Bing (live)
Reddit corroboration · freshness · cross-index citations
Gemini
Google
Schema.org depth · Google entity signals
Grok (beta)
Web + X
Real-time web search · X (Twitter) corroboration · live via web-search fallback (beta — not yet in headline scores)
Claude
ClaudeBot
Crawl access · factual precision · authoritative sourcing

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.

Data honesty

AI Overviews and AI Mode are not the same engine.

Both surface from Google’s index, but they cite the same URL only approximately 14% of the time. A score that folds them into one engine discards a real, measurable signal. Lokrix AI runs and scores each separately — you see exactly where each one surfaces you, not a blended average that obscures both.

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.

Google AI Overviews

Google

Google top-20; separate from AI Mode

54% citation index match

Google AI Mode

Google

Shares a cited URL with AIO only ~14% of the time

Start measuring your visibility across seven engines — Grok in beta.

Run a free report on your site — every engine scored on the index it actually uses, with a 95% confidence interval on every probability.

Start measuring freeHow Lokrix AI scores engines →