Brand perception
What AI says your brand is good and bad at, attribute by attribute, and how you compare with competitors
Sentiment gives you one number for how AI talks about your brand. Brand Perception breaks that number down into the things buyers judge a brand on: pricing, ease of use, support, reliability, and the capabilities that matter in your category. For each one it shows how often AI brings it up, whether it reads positive, neutral or negative, the words it uses, the answers behind them, and how you compare with your competitors.
It turns "our sentiment is 6.2" into "AI keeps saying we are hard to set up, and praises our integrations", which is something you can act on.
Sentiment vs brand perception. Sentiment is how warmly AI talks about each brand, scored from 0 to 10 and split by topic. Brand perception is why: the attributes AI describes the brand on, the quotes behind them and the sources it cited.
Brand Perception is included on Starter and Growth, on trials, and in agency pitch reports. On Free it shows an example and what upgrading adds.
Where to find it
- Overview, Sentiment tab: the What's driving this strip sits under the Sentiment Score, and the full Brand Perception section closes the tab.
- Command palette: press Cmd+K (Ctrl+K on Windows) and pick Brand Perception.
- Shared reports: a Brand perception section in the standard report. See In shared reports.
How it works
- AI's own words are collected. Whenever an answer names your brand or a competitor you track, GEOflux picks out the short phrases the answer uses to describe that brand, such as "steep learning curve" or "affordable plans", usually up to 8 per brand per answer. Each phrase gets a 0 to 10 sentiment score and keeps the quote it came from, in the answer's own language.
- The phrases are grouped into attributes of two kinds:
- Fundamentals: what every brand is judged on. Thirteen fixed attributes: Pricing & value, Ease of use, Complexity, Onboarding & setup, Customer support, Reliability, Performance & speed, Scalability, Flexibility & customization, Integrations & ecosystem, Security & compliance, Reputation & trust, and Innovation.
- Industry: what brands in your category are judged on, such as reporting or automation for software, or delivery and returns for retail. GEOflux works these out from the phrases AI uses about you and your competitors, usually 8 to 16 of them.
- The numbers are counted for your filters. An attribute's sentiment is the average score of its phrases in the current view. An attribute appears once it has at least 3 mentions in scope.
Attribute names are always in English; the phrases and quotes stay in the language of the answers. New phrases join their attribute in a daily pass, so a phrase seen for the first time today is counted from tomorrow. From time to time the Industry attributes are rebuilt from scratch, which can rename, merge or split them.
Not every phrase fits an attribute. The footer of the section says how many were grouped, for example "Grouped from 1,240 of 1,410 keywords (88%) across this brand and its competitors", with the date of the last rebuild.
Reading the profile
The toolbar above the attributes filters what you see:
- All, Fundamentals or Industry, each with a count.
- Positive, Neutral and Negative, by the attribute's average score.
- Search attributes, which matches attribute names and their most used phrases.
- Profile or Benchmark (see below).
Each attribute card shows:
- The attribute's name and kind, with new when it had no mentions in the previous period.
- Its sentiment face. Hover it for the score, for example "Negative sentiment, 3/10".
- A bar splitting its mentions into positive, neutral and negative.
- How many mentions and distinct phrases it has.
- The change in mentions against the previous period of the same length. This measures how much AI talks about the attribute, not whether the tone improved.
- Its most used phrases.
Cards are sorted by mentions, most talked-about first.
What's driving this
The strip on the Sentiment tab lists up to three attributes in each of two columns: Pulling sentiment down (attributes with more negative than positive mentions) and Holding sentiment up (the reverse), ranked by how many more. It is the quickest answer to "why is our sentiment where it is". See all attributes jumps to the full section.
Attribute detail
Click an attribute to open its detail:
- Mentions over time: daily mentions with the sentiment line.
- By AI engine: mentions and sentiment on each engine, so you can see whether one engine holds a view the others do not.
- What AI actually said: the phrases behind the attribute, most used first.
- Cited sources in these answers: the pages the engines cited when they said it. Each links to its source page. These are the pages to correct, answer or outrank.
- Answers behind this attribute: every answer that contributed, 10 per page, searchable by prompt, quote or phrase. Click one to read the full answer.
Benchmark against competitors
Benchmark puts your brand next to up to five of your tracked competitors, the ones AI talks about most, attribute by attribute.
- Each cell is the share of what AI says about that brand that concerns this attribute. A long bar means the attribute is a big part of how AI describes that brand; its color is the sentiment zone. A dash means no mentions.
- Edge is your sentiment on the attribute minus the best competitor's. Green means you lead by more than half a point, red means you trail by more than half a point.
Use it to find the attributes where a competitor is described better than you, and the ones where you have a lead worth defending. The benchmark needs answers that name both your brand and at least one tracked competitor.
Filters
Every Overview filter applies: period, engine, topic, tag, Buyer Journey, Branded vs Non-branded and persona. So you can ask "what does ChatGPT say about our pricing on Evaluate prompts" directly.
- The profile is about one brand at a time: your own, or a competitor you pick in the Brand filter. A competitor is scored on your project's attributes, so the two profiles are comparable. Under All brands the profile shows your own brand.
- The benchmark covers your brand and your tracked competitors. The brand in the Brand filter (your own by default) is pinned first, and Edge is measured for it.
- Sentiment is scored for every brand an answer names, tracked or not, but phrases are collected only for your brand and the competitors you track. Tracking a competitor adds it to the Brand filter, so you can open its profile, and starts collecting its phrases from the next answers on, grouped into your attributes in the daily pass. Answers collected before you tracked it are not read again, so a newly tracked competitor's profile starts empty and fills in as new answers arrive.
In shared reports
The standard report (The full report) and Against the competition include a Brand perception section after sentiment: the attributes pulling sentiment down and holding it up, and up to 8 attribute cards. It has no drill-down or benchmark. The section only appears when the project has Brand Perception. See Shared reports.
When there is nothing to show yet
- No perception profile yet: the project's answers have not been analysed yet. This is normal for a new project and clears on its own once enough answers are in.
- Building your perception profile: the attributes are being worked out. This takes a couple of minutes.
- No attributes in this period: too few mentions in the current view. Try a longer period or fewer filters.
- Perception profile could not be built: try again later, and get in touch if it keeps happening.
Good practice
- Start with the negative attributes that have the most mentions. A strongly negative attribute that comes up twice a month matters less than a mildly negative one in every other answer.
- Read the quotes before you react. Open the attribute and read what was actually said. AI often repeats an old review or a fixed problem.
- Follow the citations. The cited sources tell you where the description comes from, and therefore where to fix it.
- Use the Industry attributes as a content brief. They are what AI thinks buyers in your category care about. An important attribute where you have few mentions is a gap in what the web says about you.
- Check the Edge column monthly. Note the attributes where you lead or trail by the widest margin, and track them over the quarter.