How AI Actually Forms an Opinion About Your Brand

The 8 factors behind every AI perception score, and what you can do about each one

TL;DR: Kime scores every AI response about your brand across 8 dimensions: Language Tone, Competitive Position, Brand Focus, Endorsements, Source Credibility, Information Recency, Risk Factors, and Confidence Level. Each one shifts your score differently. Knowing which factor is pulling you down is what separates a targeted fix from a blind guess.

Most brand tracking tools will tell you that your AI sentiment changed. What they won't tell you is why.

You get a number. It goes up, it goes down. And you're left trying to figure out the cause. Was it a new competitor entering the conversation? An old controversy resurfacing? AI pulling from a three-year-old source that no longer reflects your product?

AI Perception is Kime's answer to that problem. Instead of a single sentiment label, it gives you a breakdown of what AI responses about your brand actually contain, scored with a consistent, rule-based algorithm and traceable back to the real AI text behind every data point.

Why isn't positive, neutral, or negative enough?

A three-bucket sentiment label tells you the outcome, not the cause. It's a bit like a doctor saying your bloodwork looks mostly fine without showing you the actual values. You know something might be off, but you don't know where to look.

AI perception is a composite signal made up of several distinct factors, and they don't always move together. A brand can score well on Language Tone while scoring poorly on Competitive Position, because every recommendation list places them third or fourth. The overall sentiment reads as neutral. The underlying reality is specific, and fixable.

Consider two brands in the same category. Brand A gets consistently positive language but is always mentioned after two competitors, almost as an afterthought. Brand B gets slightly uncertain language but is the first recommendation in almost every response. Standard sentiment says Brand A is doing well. Kime shows that Brand A has a Competitive Position problem, and that no amount of good press coverage will fix it on its own.

To take action, you need to know which specific factor is moving and in which direction. That's exactly what AI Perception measures.

What are the 8 factors that drive your AI Perception score?

When Kime analyzes an AI response about your brand, it applies a consistent, rule-based scoring algorithm to the raw text. No black box, no ambiguous sentiment labels. The algorithm evaluates 8 distinct dimensions of every response. Together, these dimensions don't just describe what AI is saying about you. They explain why you scored the way you did, and point to where you should focus.

1. Language Tone

What it measures: the specific words and phrasing AI uses when describing your brand.

Why it matters: word choice is the most direct indicator of how AI has been trained to talk about you, and it shapes how readers perceive you before they've looked at anything else.

2. Competitive Position

What it measures: where AI places your brand in ranked lists, comparisons, and recommendations.

Why it matters: a large share of AI-driven discovery comes through comparative queries. Being listed third rather than first can have a real impact on which brands users follow up on.

3. Brand Focus

What it measures: how much of the AI response is actually focused on your brand.

Why it matters: a brief mention in a longer answer contributes far less than being the central subject of the response.

4. Endorsements

What it measures: whether AI explicitly recommends your brand, stays neutral, or issues a caution.

Why it matters: there's a meaningful difference between being listed as an option and being the brand AI actively tells people to use.

5. Source Credibility

What it measures: the authority and reputation of the sources AI draws on when discussing your brand.

Why it matters: a single reference from a well-regarded publication carries significantly more influence than many mentions from low-authority sites.

6. Information Recency

What it measures: how up to date the information behind the AI answer actually is.

Why it matters: AI can describe an outdated version of your brand with complete confidence, and most users reading the response won't know the difference.

7. Risk Factors

What it measures: whether AI mentions controversies, legal issues, recalls, breaches, or other concerns related to your brand.

Why it matters: even minor or historical risk mentions can introduce doubt into an otherwise positive response, particularly for users who are still evaluating their options.

8. Confidence Level

What it measures: how much hedging or uncertain language AI uses when talking about your brand.

Why it matters: when AI expresses uncertainty, that uncertainty is passed on to the reader, which can undermine trust even when the underlying facts are positive.

How do you read your AI Perception score breakdown?

Your AI Perception score reflects all 8 of these factors, but they don't all carry the same weight, and they don't all move in the same direction at the same time. The Kime dashboard presents them as a ranked breakdown, so you can immediately see which dimensions are working in your favour and which are dragging you down.

The key insight is that you don't need to fix everything at once. The right approach is to identify the one or two factors with the largest negative impact on your current score, understand what's driving them, and address those first.

This matters because the fix depends entirely on the problem. A brand with strong Language Tone and Endorsements but weak Source Credibility needs a different strategy than a brand that scores well on credibility but consistently ranks low on Competitive Position. The breakdown is designed to make that distinction clear, so you're not wasting effort on factors that aren't actually holding you back.

Why does the same brand score differently across AI models?

One thing that often surprises brands when they first start tracking AI perception is the variation across models. The same brand, measured in the same time period, can look meaningfully different depending on which AI you check.

This doesn't mean one score is right and another is wrong. It reflects the fact that different models draw on different training data and retrieval sources, so users asking the same question in ChatGPT, Perplexity, or Gemini may get genuinely different answers about your brand.

This is actually useful information. Cross-model variance shows you where your brand's representation is consistent, where it's fragile, and which models should be prioritised in your monitoring. Treating AI perception as a single number flattens that picture.

What should you do when a specific factor is pulling your score down?

AI Perception is built to be diagnostic. The score tells you where you are. The factor breakdown tells you why. But Kime goes one step further: it shows you the exact AI excerpts that generated each signal, along with the sources the model cited in that response. So when a factor is working against you, you're not left guessing. You can read the actual wording, see the ranking, identify the outdated claim, and trace it back to the source.

Here's how to act on each factor:

1. Low Language Tone: Review the content AI is drawing on when it talks about your brand. High-traffic negative reviews and critical editorial pieces tend to have an outsized effect. Focus on building authoritative coverage that uses clear, specific, positive language about what you do.

2. Low Competitive Position: Your brand isn't appearing early enough in AI comparison outputs. The most effective fix is editorial: invest in shortlist coverage and best-of articles from sources AI trusts, and work on improving your scores on the third-party review platforms AI regularly cites.

3. Low Brand Focus: AI doesn't have enough substantive content about you to draw on. Create deeper, topic-specific material and build coverage where your brand is the main subject rather than a passing reference.

4. Low Endorsements: AI is mentioning you but not recommending you. Develop case studies and get them published on credible third-party sites. Pitch editorial coverage that frames you as the right choice for a specific use case or audience.

5. Low Source Credibility: Most of the coverage mentioning your brand comes from low-authority sources. Prioritise earned media in tier-one publications and established trade outlets, and make sure your brand is present on the review platforms that AI draws from most.

6. Low Information Recency: AI is working from outdated information about your brand. Publish updated content regularly, and proactively pitch your latest milestones, product updates, and announcements to relevant outlets so the information AI retrieves stays current.

7. High Risk Factors: A past issue is still showing up in AI responses. The most reliable way to reduce its impact is to build enough authoritative, accurate content around your brand that the weight of the risk mention becomes proportionally smaller.

8. Low Confidence Level: The claims AI makes about your brand aren't well supported across sources. Getting the same positive, verifiable facts cited by multiple independent, high-authority sources is the most direct way to increase AI confidence in what it says about you.

The goal is to move from a situation where AI has an opinion about your brand and you can't see it, to one where you have full visibility into the text, the sources, and the specific factors driving that opinion. That's when you can actually do something about it.

Frequently Asked Questions

What is an AI perception score?

An AI perception score is a measure of what AI language models actually say about your brand when users ask about it. Kime calculates it by applying a consistent, rule-based algorithm to real AI responses, evaluating each one across 8 dimensions: Language Tone, Competitive Position, Brand Focus, Endorsements, Source Credibility, Information Recency, Risk Factors, and Confidence Level.

Why do AI perception scores vary across models?

Because different models draw on different training data and retrieval sources, a brand that scores well in one model may score quite differently in another. The answers users receive aren't uniform across ChatGPT, Perplexity, Gemini, or other AI tools, and your perception score reflects that.

How is AI perception different from traditional brand sentiment?

Traditional brand sentiment tracks how people describe your brand in reviews, social media posts, and press coverage. AI perception tracks how AI models describe your brand in the answers they generate. These are different signals, shaped by different inputs, and improving one doesn't automatically improve the other.

How often does an AI perception score change?

Scores can shift within days if new content, positive or negative, enters the sources that AI retrieves from. Kime tracks your score continuously, so you can connect changes in your score to specific content or coverage events as they happen.

What is the fastest way to improve a low AI perception score?

The most effective starting point is usually the single factor with the largest negative impact on your current score. Source Credibility and Information Recency tend to respond most quickly to targeted action, because changes in authoritative coverage and fresh content can influence AI outputs relatively fast.

How do you get started with AI Perception tracking?

AI Perception is available in Kime now. Here's how to run your first analysis:

1. Connect your brand. Add your brand name and the competitors you want to track. Kime starts pulling real AI responses immediately.

2. Set your tracked queries. Choose the prompts that users in your category are most likely to ask. Kime scores your brand across all of them.

3. Open your Perception dashboard. Your factor breakdown appears automatically, with each dimension ranked by impact and the AI excerpts behind every score available with one click.

From there, sort by the factor with the largest negative impact, read the underlying AI text, and you'll have a clear picture of what to address and why.