Skip to main content
Language models don’t just name brands — they characterize them. When ChatGPT mentions your brand in a response, it typically frames it in terms of specific qualities: “easy to use,” “enterprise-grade,” “affordable,” “reliable,” “technical.” These attributes aren’t random. They reflect patterns in the training data and live web content that the model draws on, and they repeat consistently across different prompts and response runs. Brand Perception extracts and quantifies those patterns so you can see how AI actually talks about your brand at scale.

​
What perception attributes are

A perception attribute is any recurring descriptive quality that ChatGPT associates with your brand name in captured responses. GenRank identifies these by analyzing the language used in close proximity to your brand mentions across all captured responses and surfacing the qualities that appear with statistically meaningful frequency. Examples of perception attributes include:
  • Functional qualities: “easy to set up,” “integrates with existing tools,” “API-first,” “no-code”
  • Value positioning: “affordable,” “enterprise pricing,” “cost-effective for small teams”
  • Trust signals: “well-documented,” “reliable uptime,” “strong support”
  • Audience framing: “popular with developers,” “used by large enterprises,” “suited for beginners”
  • Comparative positioning: “more flexible than alternatives,” “simpler than [competitor]”
These are the terms that shape how a user’s first impression of your brand forms when they encounter it in a ChatGPT answer — before they’ve visited your site or read any of your own marketing.

​
How perception data is surfaced

The Brand Perception view shows your most frequently occurring attributes ranked by appearance frequency across your full prompt set. Each attribute entry includes:
  • The attribute phrase as extracted from responses
  • How many captured responses it appears in
  • Which prompts it’s most associated with
  • How the frequency has trended over recent weeks
You can also compare your attribute profile against competitors. This side-by-side view shows which qualities appear for your brand, which appear for specific competitors, and which are contested — appearing for both.

​
Using perception data

Understanding your current perception profile is the first step. The more valuable question is whether that profile matches what you want it to be.
Map your perceived attributes against your intended positioning. If your target positioning emphasizes enterprise reliability but your most frequent ChatGPT attributes are “affordable” and “suited for small teams,” there’s a gap between how you present yourself and how the broader information environment — which the model reflects — frames you. That gap points directly at content and PR priorities: you need more authoritative coverage that frames you in enterprise contexts, from the kinds of sources ChatGPT treats as credible.

​
Aligning content with desired attributes

Perception attributes in ChatGPT responses are downstream of the information environment your brand exists in — your own content, third-party reviews, editorial coverage, community discussions, and citations across the web. Shifting your perception profile means changing the inputs the model uses.
1

Identify the gap

Compare your current top attributes in GenRank against the attributes you want to be associated with. Note which desired qualities are absent or underrepresented, and which undesired qualities appear more than you’d like.
2

Audit what's shaping current perception

Use the Sources view to identify which domains are being cited in responses about your brand. Those sites are contributing most directly to how the model characterizes you. Read the content on those pages to understand what language they use about you.
3

Publish content that signals desired attributes

Create content — on your own site and through earned coverage — that explicitly frames your brand in the attributes you want to own. Case studies, technical documentation, comparison content, and third-party reviews that use the right language consistently are the inputs the model will draw on.
4

Track the change

Monitor your perception attribute frequencies in GenRank over the following weeks. Attribute shifts are typically slower than mention rate changes, but they are measurable. A rising frequency for a target attribute confirms that the information environment is starting to reflect your updated positioning.

​
Identifying misalignments

Perception misalignments — attributes that don’t match your intended brand position — can surface in a few different ways: Outdated positioning: If your brand went through a repositioning, a pricing change, or a product pivot, the model may still reflect the older framing because the underlying content landscape hasn’t fully updated yet. Category assumption: The model may apply category-level attributes to your brand because of who you’re typically compared to, even if those attributes don’t accurately describe you. Competitive contamination: If a competitor is closely associated with a particular attribute, some of that association may bleed into how the model frames the broader category — and by extension, you. Identifying these misalignments gives you a concrete starting point for content corrections.

​
Comparing perception across competitors

The competitor perception view shows how your attribute profile compares to each tracked competitor side by side. This reveals:
  • Attributes you own exclusively (a potential differentiation signal)
  • Attributes your competitor owns that you don’t appear in at all (a positioning gap or a deliberate differentiation)
  • Attributes both brands share (contested ground, where differentiation is less clear in AI responses)
Perception data is derived from the same captured responses as all other Response Tracking metrics. Attributes are extracted from the full response text associated with your tracked prompts — they reflect what ChatGPT says about your brand in the context of the questions your target audience is actually asking.