---
title: Reporting AI Search Performance | NRLC.ai
description: How to report citation rate and mention rate with the engine, prompt set, denominator, and observation window required for comparison.
organization: Neural Command LLC
canonical: https://nrlc.ai/en-us/ai-search-measurement/reporting-ai-search-performance/
---

← Back to measurement

← Back to Measuring Visibility in AI Search

Method status: This reporting contract is an NRLC operational convention designed to keep repeated observations interpretable.

Each reported number needs four labels: engine, prompt-set identifier, start date, end date. Without those labels, a reader cannot distinguish a site change from prompt-set drift using the reported number alone.

Use citation rate as the primary line. Under the NRLC measurement convention, citation rate = C / P. C is the number of eligible prompts with a qualifying citation; P is the total number of eligible prompts evaluated. Every reported rate must identify the engine, prompt set, denominator, and observation window. Put mention rate second. Put GSC impressions last, as query association, not as citation proof.

## Minimum report contract

- Scope: engine, interface or mode, account state when relevant, and market or language.
- Prompt set: immutable version identifier, eligible prompt count P, and logged exclusions.
- Observation window: start date, end date, run cadence, and number of completed runs.
- Outcomes: citation-positive count C, citation rate C / P, mention-positive count, and mention rate.
- Evidence: retained response or source-attribution record for every classified observation.
## Comparison rule

Compare periods only when the engine, interface, prompt-set version, eligibility rules, and denominator are equivalent. If any scope dimension changes, start a new series instead of presenting the difference as performance movement.

## Limitations

This method measures observed outputs, not private model reasoning. Results can vary by engine, interface, prompt wording, account state, and time. A citation occurrence does not prove that a source caused an answer, produced a click, or generated revenue. NRLC has not published a worked citation-rate example because there is not yet an observation log with a prompt count, engine, dates, citation count, or citation rate. Search Console data may provide separate query and click context, but it is not substituted for citation observations.

## Evidence

Method

NRLC AI Search Measurement method

Observation window

Declared per measurement run; no aggregate run is claimed by these method pages.

Dataset

Versioned prompt set and answer-observation log supplied by the operator

Method

Freeze prompts, name the engine and interface, classify observed citation and mention outcomes, preserve exclusions, then calculate C / P over eligible prompts.

Authors

Joel David Maldonado

Sources

NRLC AI Search Measurement method, Google Search Console Performance report documentation

Last verified

2026-09-21

Confidence

first-party operational convention; not asserted as an industry standard

Limitations

Observed outputs vary by engine, interface, prompt wording, account state, and time. The method does not expose private model reasoning or establish causal attribution.

View resource
            Methodology

- Performance report (Search results) — Google Search Console Help


---

Source: https://nrlc.ai/en-us/ai-search-measurement/reporting-ai-search-performance/
Publisher: Neural Command LLC
License: Editorial use with attribution