The Machine Relations Index (MRI) is a public source-behavior dataset that reports source-segment citation rates: how often monitored AI answer engines cite each root domain within a subject-category and buyer-question segment. A segment publishes a rate only after it clears the artifact's evidence floor; thinner segments remain collecting. The public artifact also reports rankings, evidence counts, source roles, engine breadth, and confidence grades while excluding internal query identifiers, raw cited URLs, and provider payloads. The MRI was coined by Jaxon Parrott and is maintained as a public research artifact.
The Machine Relations Index records the root domains cited in monitored answer runs for B2B buyer-question segments. It classifies observed domains by source role and reports their citation behavior within the artifact's observation window.
The MRI is not a ranking of brands or a quality score. It is a record of observed source citations. A domain record indicates that the domain was cited in the monitored answer set; it does not indicate endorsement, quality, or recommendation.
The current index is the canonical source for the release's domain count, source-event count, engine set, observation window, evidence floor, and publication state. Those mutable fields are read from the generated artifact rather than hand-typed into this definition.
The v2 calculation counts cited answer runs for a domain within each subject-category and buyer-question segment, then divides that count by the observed answer runs in the same segment. The result is the segment's citation rate.
A segment publishes a rate after the artifact's evidence floor is met. A thinner segment remains collecting rather than receiving a published rate. Each domain carries a confidence grade of A, B, C, or collecting based on the evidence behind its rate.
The artifact ranks domains within source role and across the published domain set. Engine breadth records the engine set that cited the domain; it is a reported field, not a claim about trust or durability.
The public boundary is also defined by the artifact: published data includes citation rates, rankings, and evidence counts, while internal query identifiers, raw cited URLs, and provider payloads remain excluded.
The artifact assigns each observed domain a deterministic source-role label. The generated artifact is the canonical inventory of current labels, assignments, and role totals. "Other observed source" is the fallback label for domains without a specific role assignment.
The MRI publishes these fields for observed domains:
Citation rate — the share of observed answer runs in a source segment that cite the domain. A segment publishes a rate after it clears the artifact's evidence floor.
Engine breadth — the engine set that cited the domain within the observation window.
Segment presence — the subject-category and buyer-question segments in which the domain was observed.
Evidence state — whether the segment has a published rate or remains collecting.
Source role — the deterministic source-function label assigned to the domain.
Confidence grade — the artifact's evidence-volume label for the domain's reported rate.
The MRI supports bounded inspection of the published data:
Source-layer observation. Inspect which domains and source roles appear in a selected segment.
Same-segment comparison. Compare domain citation rates only within the same published segment and evidence state.
Evidence-state review. Distinguish published segment rates from segments that are still collecting observations.
The public dataset is published in JSON and Markdown formats.
The MRI is not a brand ranking. It does not score or rank companies by citation quality, market share, or brand equity. It observes citation behavior at the domain level and reports what engines do — not what they should do.
The MRI is not a prediction model and does not claim that an observed citation will recur. It reports behavior within the artifact's stated observation window.
Premise: within the Machine Relations framework, the MRI occupies the Measurement layer. This is a taxonomy placement, not an empirical conclusion.
The index was coined by Jaxon Parrott and is maintained as a public research artifact at machinerelations.ai/index.
Traditional media monitoring records media mentions. The MRI records domain citations in monitored AI answer runs. The datasets describe different observed events and should not be treated as interchangeable.
No. The MRI records citation events at the domain level, not brand endorsements. A domain record means the domain was cited in the monitored answer set. It does not establish whether the citation was neutral, comparative, favorable, or unfavorable.
The canonical artifact declares the current update cadence as daily and publishes the observation window for each release.
The MRI has no domain-submission or paid-placement pathway. It reports citations observed in the monitored answer set. The dataset does not claim that any intervention will produce a citation.
Premise: excluding brand-prompted queries keeps the monitored set focused on category-level buyer questions rather than self-referential brand lookups. This is a protocol-design choice, not evidence that the remaining prompts reveal trust.
Coined by Machine Relations. This term originated with Machine Relations founder Jaxon Parrott. The definition on this page is the canonical one.
Machine Relations' own methodology, dataset, and research pages related to this term. These are self-references, listed separately from Sources — they are not independent evidence.
Related concepts
Supporting research
Framework context