Expanded Sentiment Tracking in Brand Perception
Brand Perception shows the themes AI models use when they talk about you. Each theme now expands to explain what the models actually said, in their words, with sentiment attached.
the themes now explain themselves.
Brand Perception has always shown you the themes AI models use when they describe your brand, each with a frequency and a sentiment reading. A theme name is short by design, two or three words, which makes the table quick to scan and easy to misread. While knowing models use words like "Premium Quality" and "Durability Concerns" are valuable to know, our users care about their businesses and want to know more about what is driving that.
We are thrilled to release newly expanded sentiment tracking. Every theme in the table now expands, providing a summary of how AI models described you. Now, you can better understand the sentiment behind how AI describes you across hundreds of prompts, so you can better influence it.
Brand Perception
| Theme | Frequency |
|---|---|
Exceptionally Soft Fabrics | |
What the models said Models repeatedly single out the hand feel, describing the knit as brushed and buttery against the skin. Several responses say softness is the first thing reviewers mention. | |
Versatility (Gym-to-Street) | |
What the models said Most responses frame the pieces as working for a workout and the rest of the day without a change of clothes. Models mention gym sessions, errands, and travel together. | |
Modern Minimalist Design | |
What the models said Models describe clean lines, muted colorways, and minimal branding. Several note the look reads as understated rather than athletic, which is why it gets recommended for everyday wear. | |
Technical Performance Fabrics | |
What the models said Moisture-wicking and four-way stretch come up in most answers as standard across the range. Models position the fabric as genuinely technical rather than athleisure styling alone. | |
Comfort-Focused Design | |
What the models said Responses emphasize flatlock seams, tagless collars, and waistbands that models say hold without digging. Comfort is described as the organizing principle rather than one feature among many. | |
Sustainability & Eco-Conscious | |
What the models said Several models mention recycled fabrics and a stated commitment to lower-impact manufacturing. A few note the claims are more visible in marketing than in published detail. | |
Rapid Growth & Rising Popularity | |
What the models said Models describe the brand as newer and growing quickly, often citing its expansion from a single product line. Responses treat the trajectory as context rather than praise. | |
Men's & Unisex Focus | |
What the models said Models note the range began in menswear and still leans that way, with unisex pieces filling much of the catalog. Several mention the women’s selection is narrower. | |
Durability & Quality | |
What the models said Responses describe construction that holds its shape through repeated washes, with stitching called out specifically. A few models temper this by noting the range is too new for long-term verdicts. | |
Travel-Friendly | |
What the models said Models recommend the joggers and pants for flights, citing wrinkle resistance and hidden pockets. Several say the pieces look presentable on arrival rather than slept in. | |
Popular Core Products (Joggers/Shorts) | |
What the models said Joggers and lined shorts dominate the specific product mentions across answers. Models describe them as the entry point most reviewers recommend buying first. | |
Premium Price Point | |
What the models said Most models place the range above mass-market activewear and below luxury labels. In category answers the pricing is reported as a positioning fact rather than a complaint. | |
Here it is live with sample data. Click on any theme to expand it and learn more with a summary of the findings.
their words, not ours.
When a model names your brand in an answer, Centium captures the language around that mention, the actual phrasing the model used to describe you. That happens across every answer we collect, from ChatGPT, Claude, Gemini, Perplexity, Google AI Mode, and Grok. Hundreds of those fragments are then grouped into the themes you see, each weighted by how often it came up and analyzed for sentiment.
The description under a theme is now more context from that same source material. It carries no outside knowledge about your company, and no speculation about why the models say what they say. That is intentional as we report purely what the AI models said, for better or worse. As always, our Optimize section will continue to recommend how you can improve this perception in AI models.
A description can be wrong about you. If the models have your pricing out of date or credit you with a product you discontinued, that is what the page will say, because that is what a prospect asking the same question would be told. Finding it is the point and an important part of your AI Visibility strategy.
applied automatically.
There is nothing to enable. Descriptions are written as part of your normal update, with no change to your prompts, your methodology, or your cost.
We have also gone back and applied it to dashboards that already exist. If your most recent update was from July onward, the descriptions are on your dashboard now and you do not need to wait for your next update to read them. Snapshots keep the themes they were captured with, so it is the current dashboard that gains the detail.
Perception has its own tab in the dashboard sidebar. On mobile it's available once you tap the "more" button. Themes that have a description show an arrow next to the name, and selecting the row opens it. The toggle in the card header switches between category answers and direct questions about your brand, and both sides expand.
Your export now has the same detail. The Brand Perception tab includes each theme with its frequency, its sentiment, and the full description, so the language can be pulled into a slide or report without anyone needing to log in.
Open the demo dashboard to read it with sample data, or check the Perception tab on your own brand.
questions, answered.
They come from the answers themselves. When a model names your brand in a response, Centium captures the words around that mention, the phrasing the model used to describe you. Across hundreds of answers those phrasings are grouped into themes, each with a frequency and a sentiment. Nothing is written about your brand from outside knowledge.
It reports what the models said about that theme. A theme name like "Durability Concerns" is only a headline. The description underneath names the specifics, which complaints recurred, which models raised them, and where the answers disagreed with each other.
No. Every description reports what AI models said, and is written to be read that way. It is evidence of how the models characterize you, not an assessment of your brand. Where the models are wrong about you, that is the finding.
Because they did. Models often split on the same attribute, praising a product in one answer and flagging a problem with it in another. Those splits are reported rather than smoothed over, and they usually carry a mixed sentiment. Disagreement is often the most useful thing on the page.
The scrolling ticker is the wide view, a broad range of what was said about you. The table below it is the curated read, the most significant themes sorted by frequency. Only the table expands, because that is where the detail is worth reading.
Both, and they are kept separate. Category answers are where a prospect first meets you, and the tone skews positive because a mention there is already a recommendation. Direct questions about your brand draw more critical, balanced assessments. Use the toggle in the card header to switch between them.
No. It is built from the AI answers you already collect, with no new prompts and no added cost. It refreshes on your normal schedule.
Yes. Your export includes them. The Brand Perception tab carries the theme, its frequency, its sentiment, and the full description, so you can drop the language straight into a report or share it with a team that does not log in.
More guides
Which Writers AI Cites in Your Category
Two new features are live. Influential Voices names the people behind the pages AI cites, and Content Freshness shows how old that content is.
Content Amplification: Own Your Wins
Media companies are blocking AI crawlers, and the coverage you worked hardest to earn can go invisible overnight. Content amplification keeps those wins where the models can still find them.
Now Tracking Google AI Mode
Centium now tracks Google AI Mode as a sixth model, included on every plan automatically. Six models across 120 prompts is 720 AI responses behind every report.