Last updated July 2026
Every brand asking “how visible am I in AI?” needs an answer they can act on. A vague sense that ChatGPT “mentions us sometimes” is not enough. Neither is a raw percentage from a single tool.
What you need is a rubric: a defined method that assigns points to the factors that actually drive AI visibility, weights them by importance, and produces a number you can track over time, compare to competitors, and use to prioritize fixes.
This post defines that rubric. It covers five factors, explicit point weights totaling 100, and a worked example you can apply to your own brand today.
Why a rubric beats a single share-of-voice percentage
Most AI visibility platforms report a share-of-voice number: your brand appeared in X% of prompts. That number is useful but incomplete.
Two brands can have the same mention rate and wildly different AI presence quality. Brand A gets mentioned once, briefly, in the fifth sentence of a long response. Brand B gets mentioned first, described positively, cited as a source, across four AI engines. The raw percentage treats them identically.
The rubric below fixes that by rewarding quality and breadth, not just presence.
The 5-factor rubric (0-100 total)
| Factor | Weight | What to measure | How to measure it |
|---|---|---|---|
| Mention Frequency | 30 pts | % of prompt runs in which your brand is named | Run each prompt 3-10x; score = (mentions / runs) × 30 |
| Position Prominence | 25 pts | Where in the response your brand first appears | First mention (25 pts), middle third (15 pts), final third (8 pts), not mentioned (0 pts) |
| Multi-Engine Coverage | 20 pts | How many engines mention your brand on a given prompt | 4+ engines (20 pts), 3 engines (15 pts), 2 engines (10 pts), 1 engine (5 pts), none (0 pts) |
| Sentiment | 15 pts | Whether AI engines describe your brand positively, neutrally, or negatively | Positive (15 pts), neutral (10 pts), mixed (5 pts), negative (0 pts) |
| Citation Linkage | 10 pts | Whether AI engines link to your pages as a source within the answer | Cited on 3+ engines (10 pts), cited on 1-2 engines (6 pts), mentioned but not cited (2 pts), neither (0 pts) |
The weights reflect the relative impact each factor has on actual buyer behavior. Mention frequency is the largest because being named at all is the baseline. Position prominence is second because early mentions carry more weight with readers and models. Multi-engine coverage is third because research shows the two largest platforms draw from almost entirely separate source pools: according to Profound’s analysis of 100,000 prompts, only 11% of domains are cited by both ChatGPT and Perplexity. Sentiment and citation linkage round out the rubric as quality signals.
Factor 1: Mention Frequency (30 points)
Mention Frequency is the percentage of prompt runs in which your brand is named anywhere in the response.
To score it: run your target prompt at least three times (more for contested queries). Count the runs where your brand appears. Divide by total runs. Multiply by 30.
A brand mentioned in 8 of 10 runs scores 24 out of 30. A brand mentioned in 3 of 10 runs scores 9 out of 30.
Do this across a prompt cluster of 10-20 representative queries, not a single prompt. Average the scores across the cluster to get your Mention Frequency score.
Factor 2: Position Prominence (25 points)
Position Prominence measures where in the response your brand first appears. The first sentence is not the same as the last paragraph.
Score each run based on the location of the first mention:
- First third of the response: 25 points
- Middle third: 15 points
- Final third: 8 points
- Not mentioned: 0 points
Average across runs and across your prompt cluster. A brand that leads the answer earns the full 25 points. A brand buried at the end of a long comparison paragraph earns 8.
Position matters because of how readers and models process text. According to Kevin Indig’s 2026 analysis of 18,012 verified ChatGPT citations, 44.2% of citations were drawn from the first 30% of a page’s content. The same front-loaded pattern applies inside AI responses: brands named early are more likely to be retained by readers scanning a generated answer.
Factor 3: Multi-Engine Coverage (20 points)
Multi-Engine Coverage measures how many AI engines mention your brand on a given prompt.
Score by the number of distinct engines returning a mention:
- 4 or more engines: 20 points
- 3 engines: 15 points
- 2 engines: 10 points
- 1 engine: 5 points
- 0 engines: 0 points
Track across at minimum four engines: ChatGPT, Perplexity, Google AI Overviews, and Gemini. Add Microsoft Copilot, Google AI Mode, Grok, and Meta AI for broader coverage.
The weight on this factor reflects the fragility of single-engine visibility. Because ChatGPT and Perplexity draw from largely non-overlapping source pools, a brand present only in one platform is one algorithm update away from near-invisibility. Multi-engine coverage is a resilience metric.
Factor 4: Sentiment (15 points)
Sentiment measures how AI engines describe your brand when they mention it.
Score the overall tone of descriptions across all runs:
- Positive (described favorably, “best for X,” “trusted by Y”): 15 points
- Neutral (mentioned without evaluation): 10 points
- Mixed (some positive, some negative elements): 5 points
- Negative (described critically, limitations emphasized): 0 points
Sentiment is not a binary. An AI response might call your product “powerful but complex” (mixed). Another might call it “the easiest option for small teams” (positive). Read each response and classify the net tone.
Negative sentiment embedded in AI responses can persist for weeks or months without active correction, because the model continues drawing from the same retrieval sources that originally produced the negative framing.
Factor 5: Citation Linkage (10 points)
Citation Linkage measures whether AI engines link to your specific pages as sources within the generated answer, not just mention your brand name.
Score by the number of engines citing your domain:
- Cited on 3 or more engines: 10 points
- Cited on 1 or 2 engines: 6 points
- Brand mentioned but no citation link: 2 points
- Neither mentioned nor cited: 0 points
A citation link is a qualitatively different signal from a mention. It tells you the model has incorporated your content into its retrieval layer, and it drives direct referral traffic. It is also harder to earn: you need content that AI engines pull as a source, which means structured, factual, well-sourced pages that directly answer the prompts your audience uses.
Worked example: Acme Analytics
To make the rubric concrete, here is a worked example for a hypothetical brand called Acme Analytics.
Setup: 15 prompts in the “business intelligence software” category. Each prompt run five times. Engines tracked: ChatGPT, Perplexity, Google AI Overviews, Gemini.
| Factor | Raw result | Score |
|---|---|---|
| Mention Frequency | Mentioned in 54 of 75 runs (72%) | 0.72 × 30 = 21.6 pts |
| Position Prominence | Average position: first third in 40% of runs, middle third in 45%, final third in 15% | (0.40 × 25) + (0.45 × 15) + (0.15 × 8) = 10.0 + 6.75 + 1.2 = 17.95 pts |
| Multi-Engine Coverage | Mentioned on 3 of 4 engines consistently | 15 pts |
| Sentiment | Positive across 70% of descriptions, neutral in 30% | (0.70 × 15) + (0.30 × 10) = 10.5 + 3.0 = 13.5 pts |
| Citation Linkage | Pages cited on 2 engines (ChatGPT and Perplexity) | 6 pts |
| Total | 74.05 / 100 |
A score of 74 puts Acme Analytics in the strong-presence tier. The gaps to address: position prominence (brand named too late in responses) and citation linkage (missing from Google AI Overviews and Gemini as a source). Those two factors combined represent 19 points of unrealized potential.
Score interpretation guide
| Score range | What it means | Priority actions |
|---|---|---|
| 0-29 | Minimal AI presence. Competitors dominate most responses. | Build mention frequency first. Publish content that directly answers your target prompts. Earn third-party coverage. |
| 30-59 | Emerging presence. Mentioned sometimes, inconsistently. | Improve position (lead with brand-defining answers). Expand from one engine to three or four. |
| 60-79 | Strong, consistent presence. Category contender. | Deepen citation linkage. Correct sentiment gaps. Audit which engines are still under-serving you. |
| 80-100 | Category authority. Dominant across most prompts and engines. | Defend frequency and sentiment. Monitor for drift. Expand prompt coverage to adjacent categories. |
Scoring cadence and drift
A score is a snapshot. What matters over time is the trend line.
Re-score your brand at minimum once per month for stable markets and once per week for fast-moving categories. AI engines update their retrieval sources continuously. A brand that earns a 75 in Q1 can slide to a 55 by Q3 without any change to its own content, simply because competitors published stronger sources and earned more citations.
The rubric’s value is not the score itself. It is having a consistent method that lets you isolate which factor moved, in which direction, and why.
Tools that supply the rubric inputs
You do not need to run every prompt manually. Dedicated platforms pull the data and surface the inputs you need to populate each factor.
Temso (from $89/mo) is the all-in-one AI SEO platform that tracks mention frequency, multi-engine coverage, sentiment, and citation rates automatically across up to 8 engines: ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Grok, Microsoft Copilot, and Meta AI. Its built-in workflow converts the monitoring data into a prioritized fix queue and executes content and citation actions inside the same subscription. For a brand running this rubric for the first time, Temso reduces the data-collection step from hours to minutes.
Profound provides deeper citation intelligence and prompt-volume data at $399/mo for full engine coverage. Its citation maps are the strongest in the category for teams needing enterprise-grade reporting on citation linkage specifically.
Otterly.AI (from $29/mo) covers six platforms and provides structured GEO audit guidance alongside prompt-level citation tracking. It is well-suited to teams running the rubric manually who need a low-cost data layer to validate their inputs.
Semrush’s AI toolkit provides keyword-level visibility data that can complement the rubric’s prompt-selection step, particularly for identifying which queries in a category drive the most buyer intent.
The full ranked list of AI visibility platforms, with scoring on engine coverage and feature depth, is at /rankings/ai-visibility-tools. Methodology for how tools are evaluated is at /methodology. Definitions for all terms used in this rubric are at /glossary.
Start scoring
Pick 10 prompts a buyer in your category would type into ChatGPT today. Run each five times. Collect the data. Apply the five factors. Add up the points.
The number you get is your baseline. Everything you do to improve AI visibility, publishing cited content, earning third-party mentions, correcting negative framing, expanding to underserved engines, moves one or more of these factors up.
Temso automates the collection and tracking so the rubric becomes a live dashboard rather than a manual exercise. Start a free trial from $89/mo, no credit card required.