The Machine Relations Index v2 reports source-segment citation rates — how often AI answer engines cite each source domain — published only once a segment clears the evidence floor of at least 10 observations across at least 7 distinct run dates, with each domain graded into confidence tiers A, B, C or collecting by how much evidence stands behind it.
Machine Relations definition: MRI Score is the public name for the source-segment citation rate reported by the Machine Relations Index. Despite the word "score," MRI v2 does not assign points, combine signals into a weighted composite, or issue one global judgment about a domain.
The Machine Relations Index v2 reports source-segment citation rates — how often AI answer engines cite each source domain — published only once a segment clears the evidence floor of at least 10 observations across at least 7 distinct run dates, with each domain graded into confidence tiers A, B, C or collecting by how much evidence stands behind it.
The unit is a source segment: a subject category paired with a buyer-question type. Within that segment, the numerator is the observed answer runs that cited the root domain and the denominator is all observed answer runs in the segment.
MRI Score records citation behavior inside a stated observation window. It is not a claim that the cited domain is accurate, trustworthy, recommended, popular, or likely to be cited again.
Every domain an answer engine cites is measured on equal terms. For each subject category paired with a buyer question type, the model counts how many observed answer runs cited a domain and divides by the total observed runs for that segment to produce a citation rate. Rates are published only for segments that have accumulated enough observations across enough separate days to be stable; thinner segments are shown as still collecting rather than scored. Domains are ranked within their source type and across the full measured universe, and each domain carries a confidence grade reflecting the volume of evidence behind its rate.
The generated MRI artifact is authoritative for every mutable part of the release, including:
The current artifact covers 16,356 root domains and 80,740 source events from 625 eligible prompts across 6 monitored engines. Its observation window is 2026-05-10 to 2026-07-26, and its declared update cadence is daily.
Those values are hydrated from the generated public artifact. They are not maintained as hand-typed glossary facts.
A source segment publishes a citation rate after the artifact records 10 observed runs across 7 distinct run dates. A thinner segment remains collecting and has no published citation rate.
"Collecting" is an evidence state, not a low score. It means the public artifact does not yet contain enough observations across enough dates to publish that segment's rate.
For each published source segment, the artifact can report:
| Field | Meaning |
|---|---|
| Citation rate | Cited answer runs divided by observed answer runs in the segment |
| Cited runs | Observed answer runs that cited the root domain |
| Observed runs | Answer runs included in the segment denominator |
| Engine breadth | The monitored engines that cited the domain during the window |
| Confidence grade | The artifact's A, B, C, or collecting evidence-volume label |
| Source role | The artifact's deterministic source-function classification |
| Peer context | Rank within the published domain set and within source role |
Engine breadth, source role, confidence, and peer rank are reported context. They do not change the citation-rate formula and should not be added together as points.
The taxonomy and confidence labels belong to the release that generated them. The artifact, not this page, is the canonical inventory of current labels and assignments.
Read a citation rate with its segment, numerator, denominator, evidence state, confidence grade, engine breadth, and observation window. That context defines the observation's scope.
Valid comparison is bounded to the same published segment and compatible observation window. A rate from one subject or buyer-question type should not be treated as directly interchangeable with a rate from another segment.
MRI Score does not establish why an engine cited a domain. It records the citation event; it does not infer a private retrieval rule, causal mechanism, or endorsement.
mri_score_v1.1_6_engine was superseded by mri_score_v2.0 on 2026-07-05. The v1 composite score — engine breadth, query diversity, vertical spread, position quality, and temporal consistency combined into a single weighted number — is retired. v1 scores are not comparable to v2 citation rates; the two measure different things. The published v1-to-v2 crosswalk is the one-time transition receipt.
The current methodology reports source-segment citation rates. A retired v1 composite value must not be compared with, converted into, or presented as an MRI v2 citation rate.
| Metric | Observed unit | What it records |
|---|---|---|
| MRI Score | Root domain within a source segment | Citation rate in observed answer runs |
| AI Visibility Score | Brand or company within a defined query set | Brand presence in observed AI answers |
| Share of Citation | Brand or source within a defined query set | Share of observed appearances or citations |
| Citation Velocity | Brand or source across comparable windows | Change in observed citation count or rate |
These metrics describe different units and must remain separate.
The public dataset reports citation rates, rankings, and evidence counts. It excludes internal query identifiers, raw cited URLs, and answer-engine provider payloads.
The public artifact exposes enough evidence to inspect a published rate without exposing the internal collection payloads excluded by that boundary.
MRI Score is not:
The Index has no domain-submission or paid-score pathway. Domains enter the dataset when they are observed as cited root domains in the monitored answer set.
Machine Relations introduced MRI Score as part of the Machine Relations Index. The methodology and its public release contract are defined and published by Machine Relations.
The Index does not define one universal "good" rate. Interpret a published rate against its own segment, source role, evidence state, confidence grade, and observation window.
For a source segment, divide the observed answer runs that cited the root domain by all observed answer runs in that segment. Publish the rate only after the artifact's evidence floor is met; otherwise mark the segment collecting.
No paid-score or domain-submission pathway exists. The Index reports root domains observed in the monitored citation data.
No. MRI Score reports how often the source was cited in the source segment. Accuracy and claim support require separate evaluation.
No. MRI Score is an observed AI citation rate for a source segment. Domain-authority estimates model search-ranking strength. The units and methods are different.
No. MRI Score measures root-domain citation behavior within source segments. AI Visibility Score measures brand presence within a defined answer set.
The generated public artifact declares the current cadence as daily and supplies the observation window for each release.
The current artifact declares the monitored engine set as chatgpt, claude, gemini, google_ai_mode, google_ai_overviews, perplexity. The artifact is authoritative when that set changes.
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.
Supporting research
Framework context