AI Visibility Overview

The page measures evaluated AI answers. It does not measure actual buyer traffic, the percentage of all AI searches where your brand appears, or revenue generated by AI.

What each section shows

• Visibility Index: Your company’s overall AI visibility score for the latest completed week. Use it to see how you’re doing and compare with earlier results.
• Score by Platform: The average score for your company and tracked competitors on each AI platform. These bars show results for the whole comparison group, not just your company.
• Visibility Trend: How your score compares with the average of your tracked competitors over the last 90 days. Each competitor has equal weight in the average.
• Competitors: Your company and tracked competitors ranked by visibility score, with percentage changes from the previous week.
• Competitive Product Areas: Your score compared with the competitor average in each product area. The percentage beside your score shows your own change from the previous week.
• Top Gaps to Close: Your three highest-priority recommendations. Open each one to understand the issue, review the evidence, and see the suggested action.
• Insights Summary: When available, a short AI-generated summary of key findings and suggested next steps. Review the related recommendations for more detail.

Review the page

1. Open AI Visibility and select Overview.
2. Check Last Updated to understand how recent the completed analysis is.
3. Review your Visibility Index and the Competitors preview.
4. Use Competitive Product Areas to identify a category to investigate.
5. Check Visibility Trend to see whether the observation is new or persistent.
6. Open a Top Gaps to Close title or All recommendations. Review Diagnosis, Action, Why this matters, and Related Sources.
7. Implement an appropriate change in your content or website system, then review later completed analyses.

Use the Prompts tab to inspect measured questions and answers. Administrators use Settings to manage product areas and competitors.

Why numbers can differ between sections

The sections use different populations and grouping rules. The Visibility Index is company-specific, while Score by Platform currently includes multiple companies. Product-area comparisons include companies with scores for that area. Historical points use the companies and settings associated with their completed analysis.

Scores are calculated with decimals but are often displayed as whole numbers. Do not reconstruct exact averages or changes from rounded labels.

Before interpreting a change

Keep the competitor set, product areas, and measurement coverage consistent where possible. A change in configuration or available answers can change scores without a corresponding change to your website.

A higher score is stronger performance under the measurement methodology. It is not proof of more demand, better product quality, or incremental revenue.

How the AI Visibility Index is calculated

The Visibility Index is your company's composite AI visibility score on a 0 to 100 scale. Channel99 first calculates scores for each company, product area, and AI platform. It then averages your company's available composite scores for the latest completed scoring week.

The default component weights are 40% unbranded discovery, 20% sentiment, and 40% head-to-head comparisons. Each analysis uses its saved configured weights, so the default formula should not be assumed to override a different account configuration.

Step 1 Calculate the three components

For a particular company, product area, and AI platform:

• Discovery score: the average ranked-discovery score from scored unbranded questions for that company, product area, and platform.
• Sentiment score: the average scored sentiment for that company and platform. Sentiment is assessed at company level and reused across that company's selected product areas on the same platform.
• Comparison score: the average available head-to-head scores for that company, product area, and platform.

Channel99 evaluates AI answers and then calculates these averages. The numeric aggregation is separate from the AI judgment used to interpret an answer.

Step 2 Apply the saved weights

With default weights:

**Composite score = 0.40 × D + 0.20 × S + 0.40 × H**

The score is calculated for each selected company/product-area/platform combination with at least one scored component.

**Hypothetical example:** D = 60, S = 70, and H = 50.

Composite score = 0.40 × 60 + 0.20 × 70 + 0.40 × 50 = **58**.

This is a score, not a 58% probability that an AI assistant will recommend the company.

Step 3 Average your company's composite scores

**Visibility Index = sum of your company's available composite scores / number of those scores**

The calculation averages the available product-area/platform rows for your company. It does not weight them by actual AI search volume, website visits, opportunity value, or the number of raw answer observations.

When all product areas have results on all platforms, each product-area/platform combination has equal weight. If coverage differs, the calculation averages the rows that are available; do not assume all platforms or product areas still have identical effective weight.

The Overview rounds the displayed index to a whole number. The ring represents the index out of 100; its amber color is a fixed presentation color, not a documented performance threshold.

What happens when data are missing

What happens when data are missing

A missing result and a score of zero mean different things.

• A score of zero is a measured result. It stays in the average.
• An answer that cannot be scored is left out of that component’s average.
• If one component is missing, but others are available, the missing component contributes zero. Its weight is not shared between the remaining components.
• If all three components are missing, no combined score is calculated for that company, product area, and platform.
• If your company has no overall score, the Overview shows a dash.

Example

Suppose your company has:

• Discovery score: 60
• Sentiment score: 70
• Comparison score: unavailable

Using the default weights, the calculation is:

(60 × 40%) + (70 × 20%) + (0 × 40%) = 38

The missing comparison score contributes zero. The discovery and sentiment weights remain unchanged.

Check for missing results before deciding why a score increased or decreased.

What contributes to the Visibility Index

The index combines discovery, sentiment, and comparison scores.

Citation counts, website findings, recommendation counts, confidence ratings, traffic, pipeline, and revenue do not directly add points to the index. Website and citation findings help support recommendations.

How individual AI answers become visibility scores

Channel99 evaluates three types of answers:

1. Discovery: Does your company appear when a buyer asks which vendors to consider?
2. Sentiment: How does the AI platform describe your company?
3. Comparisons: Does the AI platform prefer your company or a competitor?

By default, the combined score uses:

40% discovery + 20% sentiment + 40% comparisons

An AI evaluator interprets each answer. Channel99 then uses the scoring rules below to calculate the results.

Discovery scores

Discovery questions ask about a product area without naming your company. For example: “Which vendors should we consider for marketing analytics?”

Channel99 asks two versions of each discovery question across the configured AI platforms.

The evaluator creates an ordered list of companies from each answer. Companies higher in the list receive higher scores.

Discovery score = 100 × (number of companies − position + 1) ÷ number of companies

For an answer containing four companies:

The number of companies in the answer matters. The last company receives 100 divided by the number of companies, rather than zero.

Channel99 averages the scored results for each company, product area, and platform. When both question versions can be scored, they have equal weight.

If an answer cannot be evaluated, it is left out. If an answer has a usable list but your tracked company is missing from it, your company receives zero.

Sentiment scores

Sentiment measures how an AI answer describes a company.

The evaluator assigns a whole-number score from 0 to 100, with higher scores representing a more positive assessment.

There are two important exceptions:

• If the answer explicitly says it does not recognize the company, it receives 0. This indicates a recognition problem, not necessarily a negative opinion.
• If the answer does not provide enough information to assess sentiment, the result is unavailable.

Sentiment score = total of the scored sentiment results ÷ number of scored results

Sentiment is measured for each company on each platform. That company’s sentiment score is then used across its selected product areas on that platform.

The evaluator judges the meaning of the answer. Individual words or phrases do not have fixed point values.

Comparison scores

Comparison questions ask how your company compares with a competitor in a shared product area. Channel99 asks the question in both name orders to reduce dependence on which company is named first.

The evaluator scores the preference expressed in the answer:

These values always refer to your company, even when it is named second.

If the answer explicitly does not recognize your company in the comparison, it is scored as strongly favoring the competitor. An answer that cannot support a comparison is left out rather than treated as a tie.

Channel99 averages the scored answers for each comparison. The competitor receives the opposite score:

Competitor score = 100 − your company’s comparison score

The available comparison results are then averaged to calculate each company’s comparison component.

Comparisons between competitors

Channel99 can also estimate how two competitors compare using their results against your company. These are calculated comparisons, rather than direct answers to a question comparing those two competitors.

The calculation works as follows:

1. Find the difference between the competitors’ scores against your company.
2. Divide the size of that difference by 25 and round up to the next whole number.
3. Multiply that number by 12.5.
4. Add it to 50 for the stronger competitor and subtract it from 50 for the weaker competitor.

If their starting scores are equal, both receive 50.

Hypothetical example: Competitor A scores 75 against your company, while Competitor B scores 25.

The difference is 50. Dividing by 25 gives 2, and multiplying by 12.5 gives 25.

• Competitor A’s calculated score against B: 50 + 25 = 75
• Competitor B’s calculated score against A: 50 − 25 = 25

Confidence and supporting evidence

The evaluator also records supporting answer excerpts and its confidence in the interpretation.

Confidence does not directly increase, reduce, or multiply the visibility score.

The scores describe the answers Channel99 measured. They do not independently verify every product claim in those answers or guarantee what an AI platform will say next time.

How Score by Platform is calculated

Score by Platform shows an average for each AI platform in the latest completed scoring week.

The current calculation includes your company and tracked competitors with available scores. These bars therefore do not show only your company’s performance.

Calculation steps

1. Calculate each company’s combined score for every measured product area on a platform.
2. Average those product-area scores to get one score per company on that platform.
3. Average the company scores to get the platform score.

Platform score = total of the available company averages ÷ number of scored companies

Every company has equal weight in the final average.

Example

All numbers below are hypothetical.

Platform score = (40 + 60 + 80) ÷ 3 = 60

A company with more product areas does not get more weight in the final company average.

Because competitors are included, a platform’s score can increase even when your own company’s score stays the same. This also explains why the platform bars may not average to your Visibility Index.

How to use this section

Use the bars to understand results across your company and tracked competitors. To investigate your own company’s performance, review its scores and the underlying answers in Prompts.

A platform score is not the percentage of its users who know your brand or the chance that it will recommend your company.

How Visibility Trend is calculated

Visibility Trend compares your company’s weekly score with the average of your tracked competitors over approximately the last 90 days.

Each point represents a weekly result.

How competitor rankings and weekly changes are calculated

The Competitors panel ranks your company and tracked competitors by their current Visibility Index.

The percentage beside each company shows how its score changed from the immediately previous scoring week.

How companies are ranked

Each company’s score is the average of its available combined scores across its measured product areas and platforms.

Companies are sorted from highest to lowest using their scores before rounding. Missing scores appear last.

Companies may be measured across different product areas, so their scores do not always come from an identical set of questions.

How the percentage is calculated

Weekly change = (current score − previous week’s score) ÷ previous week’s score × 100

The percentage is calculated before the scores are rounded.

Hypothetical example: your score increases from 40 to 41.

• Score increase: 1 point
• Percentage increase: 2.5%
• Rounded display: 3%

The displayed 3% means relative growth, not a three-point increase.

How to use the ranking

Read rank and score together. Your score can improve while your rank falls if competitors improve more.

Use Prompts and product-area results to investigate the differences. The ranking describes measured AI visibility within your tracked group, not market share or overall product quality.

How Competitive Product Areas scores are calculated

This table compares your company with competitors in each measured product area.

The final percentage shows your company’s change from the previous week. It does not show your gap against competitors.

How the score columns are calculated

First, Channel99 averages each company’s available platform scores for that product area.

Company product-area score = total of its available platform scores ÷ number of scored platforms

Your company’s result appears in its own column.

For the Competitors column, Channel99 averages the available competitor company scores:

Competitor average = total of the competitors’ product-area scores ÷ number of scored competitors

Your company is excluded from this average. Each scored competitor has equal weight. A competitor without a score for that area is left out rather than automatically given zero.

The underlying combined scores use the saved scoring weights, which default to 40% discovery, 20% sentiment, and 40% comparisons.

Example

All numbers below are hypothetical.

Your platform scores for one product area are 30, 40, 50, and 60:

Your product-area score = (30 + 40 + 50 + 60) ÷ 4 = 45

Two competitors score 40 and 60:

Competitor average = (40 + 60) ÷ 2 = 50

Your company trails the competitor average by 5 score points.

How to use this table

Start with product areas that matter most to your business. Where you trail, review the measured answers and related recommendations.

Keep product-area definitions and competitor participation consistent when comparing results over time. A low score alone does not explain the cause or identify a specific website change.

How Top Gaps to Close and Insights Summary are created

Top Gaps to Close shows up to three prioritized recommendations from the latest completed and published recommendation analysis.

Insights Summary, when available, provides a short explanation of key findings and suggested next steps.

Neither section directly adds points to the Visibility Index.

What recommendations are based on

Channel99 uses evidence from:

• Measured AI answers.
• Sources cited in those answers.
• Relevant pages on your website.
• Website access and structure checks.

A recommendation needs evidence for the suggested action. A low score alone does not mean a page needs rewriting.