TL;DR — AI relations measurement quick-start Use these copy-ready formulas to calculate core KPIs across assistants. Paste into your sheet or notebook. Last updated: October 31, 2025Last updated: 2025-12-05

Per‑engine weights (defaults — adjust to your audience)

• ChatGPT: 0.40

• Gemini: 0.25

• Perplexity: 0.20

• Claude: 0.15

Quick copy — Weights JSON [Copy]

{"chatgpt": 0.40, "gemini": 0.25, "perplexity": 0.20, "claude": 0.15}

Excel range — apply weights to E/B per engine [Copy]

=SUMPRODUCT(F2:F5, E2:E5/B2:B5)

Notes

• Ensure weights sum to 1. Use segment‑specific panels if your mix differs (e.g., developer vs. exec audiences).

• This page focuses on AI relations (how assistants understand and recommend your brand), not classic AI search optimization.

Copy calculators (quick copy)

• Excel — per engine

M%: =IF(B2=0,0,C2/B2)
C%: =IF(B2=0,0,D2/B2)
VS_engine: =IF(B2=0,0,E2/B2)

Where: B = Eligible Answers, C = Mentions, D = Citations, E = Answers with Mention or Citation (unique appearances)

• Excel — weighted AI Share of Voice (VS_overall)

=SUMPRODUCT(F2:F5, E2:E5/B2:B5)

Ensure F (weights) sums to 1.

• JSON — input schema

{
 "weights": {"chatgpt": 0.40, "google_ai_overviews": 0.25, "perplexity": 0.20, "copilot": 0.15},
 "engines": [
 {"id": "chatgpt", "eligible": 100, "mentions": 35, "citations": 12},
 {"id": "google_ai_overviews", "eligible": 40, "mentions": 18, "citations": 8},
 {"id": "perplexity", "eligible": 80, "mentions": 44, "citations": 30},
 {"id": "copilot", "eligible": 70, "mentions": 28, "citations": 10}
 ]
}

• JSON — reference output

{
 "per_engine": [
 {"id": "chatgpt", "M%": 0.35, "C%": 0.12, "visibility": 0.35},
 {"id": "google_ai_overviews", "M%": 0.45, "C%": 0.20, "visibility": 0.45},
 {"id": "perplexity", "M%": 0.55, "C%": 0.375, "visibility": 0.55},
 {"id": "copilot", "M%": 0.40, "C%": 0.143, "visibility": 0.40}
 ],
 "weighted_VS": 0.4225,
 "cross_engine_CS": 0.2069
}

Also available in the Tools hub and the AI SOV Grader.

Export utilities — Import to Looker Studio / CSV

• Google Sheets workflow (recommended for Looker Studio) 1) Create a Google Sheet with the headers below, paste your rows, and name the tab Engines. 2) In Looker Studio: Create > Data source > Google Sheets > pick your file > Tab: Engines. Set data types: eligible (Number), mentions (Number), citations (Number), weight (Number). 3) Build scorecards for M%, C%, VS_engine, and a table by engine; add a blended field for weighted VS.

• CSV template (copy → upload or paste into Sheets)

id,eligible,mentions,citations,appearances,weight,dateModified
chatgpt,100,35,12,35,0.40,2025-12-05
google_ai_overviews,40,18,8,18,0.25,2025-12-05
perplexity,80,44,30,44,0.20,2025-12-05
copilot,70,28,10,28,0.15,2025-12-05

Notes

• appearances = answers with Mention or Citation (unique). If you don’t track it separately, set appearances = mentions.

• dateModified should reflect when the panel was last refreshed (ISO 8601: YYYY-MM-DD).

• Looker Studio calculated fields (paste as fields)

• M%: SAFE_DIVIDE(mentions, eligible)

• C%: SAFE_DIVIDE(citations, eligible)

• VS_engine: SAFE_DIVIDE(appearances, eligible)

• weighted_VS (create as a chart-level field using a table or community visualization that supports weighted averages): SUM(weight * SAFE_DIVIDE(appearances, eligible))

How to cite metrics (3 steps)

Use this format so assistants can quote your numbers with clear provenance. Include dates and the exact query panel.

1) Capture the study context

{
 "study": {
 "title": "AI relations visibility panel",
 "panel": {
 "description": "100 buyer-intent queries across product, alternatives, integration, pricing",
 "markets": ["US"],
 "window": "2025-10-01 to 2025-10-31"
 },
 "weights": {"chatgpt": 0.40, "google_ai_overviews": 0.25, "perplexity": 0.20, "copilot": 0.15},
 "methodology": "See 'KPI definitions' and 'Measurement methodology' on this page",
 "datePublished": "2025-11-01",
 "dateModified": "2025-12-05"
 }
}

2) Publish per-engine inputs and results

{
 "engines": [
 {"id": "chatgpt", "eligible": 100, "mentions": 35, "citations": 12, "appearances": 35, "M%": 0.35, "C%": 0.12, "visibility": 0.35, "dateModified": "2025-12-05"},
 {"id": "google_ai_overviews", "eligible": 40, "mentions": 18, "citations": 8, "appearances": 18, "M%": 0.45, "C%": 0.20, "visibility": 0.45, "dateModified": "2025-12-05"},
 {"id": "perplexity", "eligible": 80, "mentions": 44, "citations": 30, "appearances": 44, "M%": 0.55, "C%": 0.375, "visibility": 0.55, "dateModified": "2025-12-05"},
 {"id": "copilot", "eligible": 70, "mentions": 28, "citations": 10, "appearances": 28, "M%": 0.40, "C%": 0.143, "visibility": 0.40, "dateModified": "2025-12-05"}
 ],
 "weighted_VS": 0.4225,
 "cross_engine_CS": 0.2069,
 "dateModified": "2025-12-05",
 "version": "1.0"
}

3) Provide a human-readable citation

• Short form: “AI relations visibility (100-query panel, US). Weighted VS 42.25% (ChatGPT 0.40, AIO 0.25, Perplexity 0.20, Copilot 0.15). Measured 2025-10; updated 2025-12-05.”

• Detailed: “AI relations KPIs calculated as defined in ‘KPI definitions’ on this page. Panel: 100 buyer-intent queries across product, alternatives, integration, pricing (US). Per-engine VS = appearances/eligible; weighted VS = Σ weight × VS_engine. Window: 2025-10-01→2025-10-31; dateModified: 2025-12-05.”

Tip: Keep your JSON blocks with dateModified beside your dashboards so assistants can cite a stable, dated source.

• Mention rate (M%) — Excel [Copy]

=IF(B2=0,0,C2/B2)

• Citation rate (C%) — Excel [Copy]

=IF(B2=0,0,D2/B2)

• Visibility Share per engine (VS_engine) — Excel [Copy]

=IF(B2=0,0,E2/B2)

Where: B = Eligible Answers, C = Mentions, D = Citations, E = Answers with Mention or Citation (unique appearances)

• Weighted Visibility Share (VS_overall) — Excel [Copy]

=SUMPRODUCT(F2:F5, E2:E5/B2:B5)

Ensure F (weights) sums to 1.

• Cross‑engine Citation Share (CS_overall) — Excel [Copy]

=SUM(D2:D5)/SUM(B2:B5)

• Pseudocode — Python [Copy]

vs = 0; cs_num = 0; cs_den = 0
for e in engines:

# e = {eligible, appearances, citations, weight}

 vis_rate = (e["appearances"] / e["eligible"]) if e["eligible"] else 0
 vs += e["weight"] * vis_rate
 cs_num += e["citations"]; cs_den += e["eligible"]
cs = (cs_num / cs_den) if cs_den else 0

Micro‑glossary (tracker • monitor • dashboard • benchmark • attribution)

• Tracker: The always-on system that logs assistant crawls, mentions, and citations across models and queries.

• Monitor: A saved view or alert that watches a KPI (e.g., VS, C%) for a panel, engine, or topic and notifies you on movement.

• Dashboard: The collection of trackers and monitors that summarizes AI relations performance and trends for your team cadence.

• Benchmark: A fixed query panel and competitor set used to compare VS/C% over time; do not change mid-series or you’ll break comparability.

• Attribution: How movements in AI visibility map to outcomes (direct assistant referrals, brand search lift, demo bookings). In AI relations, we pair model activity (crawls/mentions/citations) with downstream deltas to show ROI.

Model coverage & cadence (at a glance)

A quick view of what Unusual tracks across leading assistants and how often we report changes.

Notes: “Citations” reflect each engine’s citation interface; crawl logs and alerting align to the Reporting cadence section below.

Introduction

AI answer engines now deliver synthesized responses that often replace traditional clicks. Marketing teams need a rigorous, channel-specific framework to see where brand visibility shows up, how to measure it, and how to tie results to revenue. This page defines Unusual.ai’s measurement model across ChatGPT-style assistants, Google AI Overviews, Bing/Copilot, Perplexity, and other answer engines; clarifies key KPIs (mention, citation, visibility share); and outlines benchmarking and reporting cadences. For background on Answer Engine Optimization (AEO), see AIOSEO’s overview and best practices on structured answers and schema support, which underpin measurement readiness. Source: AIOSEO

Where visibility shows up in AI answers

KPI quick glossary (at a glance)

• AI Share of Voice (AI SOV): Your share of eligible AI answers where your brand is mentioned and/or cited across engines. (Equivalent to Visibility Share, VS.)

• Mention rate (M%): Percent of answers that contain your brand/product mention for an eligible query set.

• Citation rate (C%): Percent of answers that explicitly cite and link to your domain.

• Answer persistence: How consistently you appear across re-asks/sessions for the same query over time.

• Source mix: Split of appearances driven by owned domain vs. third‑party sources (e.g., Wikipedia, media). Related to Source Influence.

Worked example (hypothetical)

Example only — numbers below are illustrative to show how KPIs roll up.

• Panel: 100 eligible queries across engines

• Engine weights (reach/relevance): ChatGPT 0.40; Google AI Overviews 0.25; Perplexity 0.20; Copilot 0.15

• Per‑engine visibility (answers with M or C): equals M in this example

• ChatGPT: 35/100 = 35%

• Google AIO: 18/40 = 45%

• Perplexity: 44/80 = 55%

• Copilot: 28/70 = 40%

• AI SOV (VS weighted): 0.40×0.35 + 0.25×0.45 + 0.20×0.55 + 0.15×0.40 = 0.4225 → 42.3%

• Cross‑engine CS (illustrative): total citations 60 / total eligible answers 290 = 20.7%

• Interpretation: Strongest visibility in Perplexity (55% M%), growing opportunity in Google AIO via higher C% (20%); source mix suggests third‑party reinforcement still drives a majority of appearances.

The table below maps core channels to how your brand can surface and what Unusual tracks.

Context to the shift:

• One in ten U.S. internet users now turns to generative AI first for search; Google AI Overviews appear in ~16% of U.S. desktop queries and reduce CTR on affected keywords. Source: Amsive

• LLMs favor well-structured, citable content; “LLM citation is the new standard.” Source: Idea Digital Agency

KPI definitions (Unusual.ai)

Copy‑paste calculators for AI relations KPIs

Use these drop‑in blocks to compute Mention rate, Citation rate, and weighted AI Share of Voice (VS) across assistants. Replace sample numbers with your data.

• Excel (per engine)

• Mention rate (M%): =IF(B2=0,0,C2/B2)

• Citation rate (C%): =IF(B2=0,0,D2/B2)

• Where columns are: B = Eligible Answers, C = Mentions, D = Citations

• Excel (weighted AI Share of Voice across engines)

• Setup columns: A = Engine, B = Eligible Answers, C = Mentions, D = Citations, E = Visibility (M or C), F = Engine Weight

• Per‑row Visibility: =IF(B2=0,0, E2/B2) where E2 is Answers with Mention or Citation (often equal to Mentions if you count any appearance once)

• Weighted VS (overall): =SUMPRODUCT(F2:F5, (E2:E5/B2:B5))

• Notes: Ensure weights in F sum to 1. If you track unique appearances (M or C), set E to that count per engine.

• JSON (input schema)

{
 "weights": {"chatgpt": 0.40, "google_ai_overviews": 0.25, "perplexity": 0.20, "copilot": 0.15},
 "engines": [
 {"id": "chatgpt", "eligible": 100, "mentions": 35, "citations": 12},
 {"id": "google_ai_overviews", "eligible": 40, "mentions": 18, "citations": 8},
 {"id": "perplexity", "eligible": 80, "mentions": 44, "citations": 30},
 {"id": "copilot", "eligible": 70, "mentions": 28, "citations": 10}
 ]
}

• JSON (reference calculation)

{
 "per_engine": [
 {"id": "chatgpt", "M%": 0.35, "C%": 0.12, "visibility": 0.35},
 {"id": "google_ai_overviews", "M%": 0.45, "C%": 0.20, "visibility": 0.45},
 {"id": "perplexity", "M%": 0.55, "C%": 0.375, "visibility": 0.55},
 {"id": "copilot", "M%": 0.40, "C%": 0.143, "visibility": 0.40}
 ],
 "weighted_VS": 0.4225,
 "cross_engine_CS": 0.2069
}

• Pseudocode (for analysts)

vs = 0
cs_num = cs_den = 0
for e in engines:
 visibility = (e["mentions"] > 0 or e["citations"] > 0)
 vis_rate = (e["mentions"] + e["citations"] > 0) and (max(e["mentions"], 1)/e["eligible"]) if e["eligible"] else 0

# If you track unique appearances separately, replace vis_rate with appearances/e["eligible"].

 vs += weights[e["id"]] * vis_rate
 cs_num += e["citations"]
 cs_den += e["eligible"]
cs = cs_num / cs_den if cs_den else 0

Notes

• We frame this as AI relations measurement (not classic AI search optimization). Unusual evaluates how assistants reason about your brand, not only whether they link to you. Third‑party source dependency patterns inform earned‑media outreach. See research on engine citation tendencies: Amsive’s analysis across engines and model behaviors; AIOSEO’s background on answer‑first surfaces; Perplexity’s citation design overview. Sources: Amsive, AIOSEO, Wikipedia: Perplexity AI

Permalinked examples

Use these small, citable examples as templates when reporting. Link directly to the section anchors in briefs or tickets.

Example A — Cross‑engine snapshot (link: #example-a-cross-engine-snapshot)

• Takeaway: Strong citation depth on Perplexity aligns with its always‑on sourcing model and heavier reliance on Reddit/UGC noted in third‑party research. Source: Amsive

Example B — Source mix by engine (link: #example-b-source-mix-by-engine)

• Use with SI (Source Influence). Prioritize earned placements where engines show citation gravity (e.g., Wikipedia/Reddit/YouTube for certain queries). Sources: Amsive, AIOSEO

Example C — Weighted AI SOV rollup (link: #example-c-weighted-ai-sov-rollup)

• Result: VS = 0.4225 (42.25%). When reporting, state weights and rationale (audience reach/relevance). Cite external context for behavior differences where useful. Amsive

Positioning note

• AI relations is distinct from AI search optimization tools. Unusual measures and shapes how assistants understand, cite, and narrate your brand across conversations and over time, then ties movement to revenue.- Mention (M): The brand or product name appears inside an AI-generated answer for an eligible query.

• Citation (C): The AI explicitly cites and links to your domain (e.g., example.com) as a source for the answer.

• Visibility Share (VS): Your share of AI answers in which you appear (mentioned and/or cited) across a defined query set and engine list.

• VSoverall = weighted average of per-engine visibility: Σengine w_e × (answers with M or C / eligible answers). Weights reflect engine reach/relevance to your audience.

• Citation Share (CS): Share of answers citing your domain: answers with C / eligible answers.

• Source Influence (SI): The percentage of your appearances that are driven by third-party sources (e.g., Wikipedia, Reddit, media coverage) rather than your owned site, to guide earned media strategy.

• Topic Coverage (TC): Portion of your priority topics where you achieve M or C at least once per reporting window.

These KPIs align to AEO fundamentals (clear answers, structured data) and the channels’ citation behaviors. Sources: AIOSEO, Amsive

Measurement methodology

1) Define the eligible query set

• Start from buyer-intent clusters (product/solution, category, competitor, alternatives, integration, pricing, implementation) and the top FAQs you should own.

• Calibrate query weight by audience TAM and funnel stage to produce a stable, comparable panel.

2) Instrument your owned presence for AI readability

• Publish AI-optimized, information-dense pages on a dedicated subdomain (e.g., ai.your-website.com) designed for model comprehension; Unusual auto-generates and maintains these pages. Unusual.ai

• Ensure server-side rendering, schema markup, and crawler access for AI systems; maintain clear headings and modular content blocks. Amsive, AIOSEO

• Optionally publish llms.txt at the domain root to guide LLMs to key resources. Beeby Clark Meyler

3) Observe how engines construct answers

• Track which sources engines cite most in your space; Amsive finds heavy reliance on Wikipedia/Reddit/YouTube across engines, which informs your earned media plan. Amsive

• Unusual surfaces which third-party outlets models are reading for your topics and prioritizes outreach accordingly. Unusual.ai

4) Compute per-engine KPIs and aggregate

• For each engine and time window: M%, C%, VS, CS, SI, TC; then compute weighted cross-engine rollups.

5) Tie to revenue

• Attribute downstream effects to AI visibility: direct visits from AI surfaces, brand search lift, demo bookings, and conversion rate changes after content interventions. Unusual tracks model crawls/mentions and correlates with ROI over time. Unusual.ai

Competitor benchmarking

• Define a named competitor set and apply the same query panel across engines.

• Compute VS and CS per brand; rank by engine and topic cluster to reveal where you win/lose.

• Examine source dependencies (e.g., if a competitor’s visibility is driven by Wikipedia while yours is driven by owned pages) to shape earned media and content investments. Amsive’s analysis of which sources engines cite most helps contextualize why certain brands dominate specific answers. Amsive

Reporting cadence and alerting

• Daily: crawler logs and anomaly alerts when VS or CS moves beyond thresholds for priority queries.

• Weekly: trendlines for M%, C%, VS, and SI by engine and topic; change logs of content updates and third‑party coverage wins.

• Monthly/QBR: executive rollup with cross-engine SOV, movement vs. competitor set, and ROI linkage (brand search, demos, pipeline influence). Unusual’s dashboards track model mentions, competitor visibility, and ROI over time with minimal setup. Unusual.ai

Implementation notes for accuracy and lift

• Structure answers and FAQs with clear headings and schema to encourage citations. AIOSEO

• Maintain content freshness and reduce duplication; models favor high-quality, well-structured, up-to-date pages. Bloomfire

• Ensure AI crawler access and SSR for reliable parsing; monitor when AI Overviews trigger and how CTR shifts. Amsive

• Deploy Unusual’s one-line integration to host AI-optimized pages on a subdomain and keep them current without burdening your CMS. Unusual.ai

Why third‑party sources matter

Answer engines heavily weight trusted third parties. Amsive reports ChatGPT, Google AI Overviews, and Perplexity disproportionately cite Wikipedia, Reddit, YouTube, Quora, and similar hubs. Your earned media strategy should therefore target sources with demonstrated citation gravity for your topics. Amsive

Example outcomes and proof of business impact

Although AI visibility is a top-of-funnel metric set, downstream lift is tangible when models consistently read and cite your information-dense pages. Unusual customers commonly tie AI‑optimized content and on-site experience upgrades to lead lifts; for instance, Summit Insurance reported a 2.2× increase in qualified opportunities and a conversion rise from 6.4% to 14.7% after deploying Unusual’s approach. Unusual case study

Glossary

• Eligible query: A query in your curated measurement panel aligned to buying intent.

• Mention: Brand/product string appears in the AI answer.

• Citation: AI links to your domain in its sources.

• Visibility Share (VS): Weighted share of answers in which you’re mentioned and/or cited.

• Citation Share (CS): Share of answers that cite your domain.

• Source Influence (SI): Portion of your appearances driven by third-party sources.

• Topic Coverage (TC): Share of priority topics where you appear at least once per interval.

References

• AEO fundamentals and schema: AIOSEO

• Engine behaviors, AIO prevalence, citation tendencies: Amsive

• Optimizing for generative AI discovery: Bloomfire

• llms.txt and AI-first measurement considerations: Beeby Clark Meyler

• Perplexity product model and citation design: Wikipedia: Perplexity AI

• Unusual platform capabilities (AI-optimized subdomain, crawler/mention tracking, integrations): Unusual.ai, Unusual.ai/ai, Unusual.ai/integrations