AI monitoring

For a long time, search engine optimisation was exclusively about Google. More and more people now ask ChatGPT, Gemini and Perplexity their questions instead of googling. As AI tools become primary sources of information, companies need to monitor their presence there, not just in traditional search results.

AI monitoring means systematically tracking how and when the company is mentioned in AI-generated answers, analysing the tone of those mentions and comparing its visibility with that of its competitors.

It is particularly important to measure what share of the mentions in Google’s AI Overviews goes to the company compared with its competitors, known as share of voice.

Without monitoring, problems are discovered too late: ChatGPT recommends competitors but not the company. Perplexity describes the company incorrectly. In a hypothetical example, Google AI Overviews mentions the company in the answers to 5 per cent of relevant queries, while a competitor is mentioned in 40 per cent.

Search patterns are changing fundamentally

Traditionally

People google → Click on a result → Visit websites

New behaviour

People ask ChatGPT/Gemini/Perplexity → Get an answer directly → Visit no websites (or only those the AI recommends)

If the company is not mentioned in AI answers, it loses attention and traffic.

AI answers shape perceptions

Anyone reading an AI answer forms an impression of the company without visiting its website.

Problem

If the AI describes the company incorrectly, that impression can reach thousands or millions of users, and if the AI does not mention the company at all, they form no impression of it.

Example

ChatGPT says “The best SEO agencies in Stockholm are A, B and C” (but not the company).

Result

The company loses potential customers who never find out that it exists.

Competitors can dominate AI answers

If ChatGPT recommends three companies and yours is not one of them, your competitors get 100 per cent of the attention.

Measure your share of mentions relative to other companies.

Insight: Competitor A dominates AI visibility (8 times as many mentions).

Why it pays to start monitoring AI answers early

Today’s measurements become the baseline for future comparisons.

Those who start monitoring and optimising now gain a competitive advantage.

Later: When everyone is optimising, it becomes harder to stand out.

1. Mentions: how often is the company mentioned in AI answers?

What is monitored: The number of mentions for relevant queries.

Example

Results (100 queries tested; one answer can mention several companies)

Insight

Your company has low visibility (11 per cent) compared with its competitors.

Action

Optimise content and sources so that AI cites the company more often.

2. Where in the AI answers does the company rank?

What is monitored: When the company is mentioned, is it the first recommendation or the last?

The company mentioned first stands out most. The one mentioned last stands out least.

Example

Ranking

3 of 3

Goal

Reach position 1 or 2.

3. Is the sentiment of the mentions positive or negative?

What is monitored: How is the company described when it is mentioned?

Measurement: The share of positive, neutral and negative mentions.

Example

Positive (70 per cent)

“Company X is an experienced SEO agency.”

Neutral (25 per cent)

“Company X is a digital marketing agency in Stockholm.”

Negative (5 per cent)

“Company X has a limited presence on social media.”

Goal

Over 80 per cent positive, under 5 per cent negative.

4. In what context is the company mentioned?

What is monitored: Whether the mention concerns the right industry, the right area and the right company.

Example

Relevant

“For SEO services in Stockholm, consider Company X.”

Irrelevant

“Company X (the clothing brand) has new collections.” (the wrong company)

Problem

AI can confuse companies with similar names. This must be monitored and corrected.

5. What share of the mentions in AI Overviews goes to the company?

What is monitored

The company’s share of mentions in Google AI Overviews relative to other companies.

Why it is critical

AI Overviews appear at the top of Google Search results for the searches where Google generates an overview. The share of mentions here is directly linked to visibility and potential customers.

Measurement

Example

Search phrase tested: “digital marketing Stockholm” (200 searches over the month)

Insight

Your company has a 12 per cent share of voice in AI Overviews. Company A dominates with 44 per cent (almost four times as many mentions).

Action

Optimise content and sources to increase the share of mentions in AI Overviews.

Reporting

The share of mentions is a key metric in monthly reports. Visualise it in a graph or table to show the difference from competitors clearly.

6. How do the mentions change over time?

What is monitored: Trends in mentions, sentiment and ranking.

Example

Trend: Mentions and share of voice are increasing, and sentiment is improving.

ChatGPT

Why it matters: The most widely used AI tool, with hundreds of millions of users.

Monitoring: Ask the test queries via the OpenAI API, log the answers and analyse the mentions. API answers may differ from the answers in ChatGPT.

Google AI Overviews (and AI Mode)

Why it matters: Integrated into Google Search. Appears for some searches.

Important: This is the most critical place to measure the share of mentions relative to other companies because it directly affects visibility in Google Search.

Gemini

Why it matters: Google’s AI, integrated into the Google ecosystem (Gmail, Drive and others).

Monitoring: Ask Gemini the test queries and log the answers.

Perplexity

Why it matters: Research-oriented users. Citation-driven (always shows sources).

Monitoring: Ask Perplexity the test queries and analyse the sources.

Step 1: Define test queries

Identify which questions are relevant to the company.

Categories

Example (SEO agency)

In total: 20 to 50 test queries.

Step 2: Set up automated monitoring

Tools

Step 3: Analyse the mentions

For each test query

Was the company mentioned? Yes/No

Calculate share of voice

Formula: Share of voice = (Own mentions ÷ Total number of mentions for the industry) × 100

Step 4: Focus on AI Overviews analysis

A separate process for AI Overviews

1. Sample size: Search for each important search phrase 20 to 50 times over the month (spread across different days and times). Vary the location and device, and search without being logged in, so that the result is not based on a single search setting.
2. Log the data:

• Did an AI Overview appear? (Yes/No)
• Which companies were mentioned?
• Position per company (1st, 2nd, etc.)

1. Calculate the visibility share: For each company: (Number of AI Overviews mentioning the company ÷ Total number of AI Overviews shown) × 100. The total can exceed 100 per cent, because one AI Overview can mention several companies.
2. Compare: Visualise in a table or graph for a clear comparison.
3. Identify patterns:

• For which search phrases does the company have a high share?
• Which search phrases do competitors dominate?
• Where are the biggest gaps to close?

Step 5: Sentiment analysis

Categorise the sentiment of each mention

Positive

Contains words such as “leading”, “best”, “expertise”, “experienced”, “recommended”.

Neutral

A factual description without judgement.

Negative

Contains criticism or problems.

Automate (if possible)

Use a language model via an API (for example OpenAI). Google’s Natural Language API does not support Swedish.

Manually

Read through the mentions and categorise them.

Step 6: Trend analysis

Track the metrics over time

Graph: Visualise the trend (are mentions increasing? Is share of voice improving?).

Example

Trend: Positive. Mentions and share of voice are increasing.

Step 7: Alerts and actions

Set up alerts

Example alert

“Competitor A’s share of voice in AI Overviews rose from 40 per cent to 55 per cent (plus 15 percentage points) over the past month.”

Action

Analyse what competitor A is doing (new content? New backlinks? Media citations?) and adjust your own strategy.

Optimise for greater AI visibility

Monitoring identifies problems. Optimisation solves them.

1. Create content worth citing

2. Get cited by credible sources

AI weighs sources: If the company is cited by credible third-party sources (media, trade publications), the chance that AI mentions the company increases.

3. Optimise for AI Overviews specifically

4. Monitor and correct incorrect information

If AI describes the company incorrectly

Problem: ChatGPT gives the wrong founding year for the company.

Monthly report

Content

Competitor comparison: Particularly for AI Overviews: Share of mentions relative to other companies

Example table

Share of voice in AI Overviews (March 2026)

Focus on relevant questions

Do not monitor random questions.

Always compare with competitors

Absolute figures say little

“We were mentioned 10 times this month.” Good or bad?

Relative figures give context

“We were mentioned 10 times, competitor A 40 times.” Now we know we are behind.

Focus on share of voice, particularly in AI Overviews.

Track trends, not individual months

One bad month is not a disaster.

Act on negative mentions

How to go from monitoring to greater AI visibility

AI monitoring is the first step.

Process

1. Monitor → Understand the current situation (mentions, share of voice, particularly in AI Overviews)
2. Analyse → What are competitors doing better? Why are they mentioned more often?
3. Optimise → Create content that AI cites, become a source, optimise for AI Overviews
4. Measure again → Is the share of mentions improving?
5. Iterate → Continuous improvement

Goal: To appear in AI answers for the most important questions in the industry.

Contact us about AI monitoring

Carnaby helps companies monitor their visibility in AI tools, from tracking mentions and sentiment analysis to measuring the share of mentions in AI Overviews relative to other companies, so that they can optimise for the new search behaviours.

Contact us →