Skip to main content
Being mentioned is one thing. What the answer says about you is another. Brand Attributes collects the qualities AI answers ascribe to your brand — fast implementation, strong support, high cost — and sorts them into strengths and weaknesses, so you can see what the assistants tell people about you, and where you stand against the brands they compare you with.
Brand Attributes is in beta. You’ll find it under Brand in the sidebar.

How attributes are read

Every answer is read for the qualities it gives each brand it mentions — yours and everyone else’s.
  • Only framed qualities count. A quality is recorded when the answer presents it as an advantage (“stands out for”, “a real plus”) or a drawback (“the catch is”, “unfortunately”). A plain fact with no verdict is left out.
  • Only the answer’s own words count. A brand’s reputation anywhere else never adds an attribute.
  • Similar phrases are grouped. Quick setup and fast onboarding become one attribute, so the same quality isn’t split across several rows. Attributes are named in English, whatever language the answers are in.
  • Everything is counted in answers. An answer counts once for an attribute, however many times it repeats it.

Overview

Strengths and weaknesses

The qualities the answers frame in your favour, and those they frame against you, ranked by how many answers used them. The card shows the top ten; the expand button in its corner opens the full list. Each strength can carry a badge that says how it compares with the rest of the market: A strength with fewer than three answers gets no badge — there’s too little to go on either way.

You vs the top rival

Your top strengths side by side with a competitor’s: your answers on the right, theirs on the left. Pick any brand from the list, or leave it on Auto to compare each strength against whichever brand is described that way most.

Strengths vs weaknesses over time

How many strengths and weaknesses the answers gave you each day, week or month. A change in how the assistants talk about you shows up here with a date on it.

You vs the market

Every brand described with your most common strengths, and how often. Read down a column to see who owns a quality; read across your row to see what you’re known for. The card shows the ten most-described brands; expand it for all of them. This covers every brand the answers mention, not only the ones you’ve marked as competitors — so a brand you don’t track yet can still turn up here.

One attribute in detail

Click any attribute to open it in a side panel:
  • By model — how many answers from each assistant described you this way.
  • Pages cited — the sources the answers cited when they did, with your own site marked. These are the pages shaping how you’re described.
  • Prompts behind it — the questions whose answers produced this attribute.
  • Phrases AI used — the exact wording the answers used, before it was grouped.

By prompt

The By prompt tab turns it around: every prompt whose answers described your brand, with how many strengths and weaknesses each produced. Click a prompt to see its attributes and which assistants answered it. This is where to look when one question keeps producing the same weakness.

New projects and early data

  • Attributes appear once the answers have been grouped, shortly after each day’s answers are collected. Until then the page shows Grouping attributes, with the strengths-vs-weaknesses chart already filled in.
  • While there are fewer than 30 answers or fewer than seven days of data, the page says the data is early. Expect the lists to move as more answers arrive.
  • From 30 answers on, attributes mentioned in only one answer are left out of the lists. A note under each list says how many.

Filtering and export

Both tabs follow the date range, model and segment filters. Export downloads what the page shows as CSV — the strengths and weaknesses, the brand comparisons, and the day-by-day chart. The side panels have their own export for one attribute or one prompt. See Sharing and export. You can also ask about your attributes through the MCP server — for example, “What are our weaknesses in ChatGPT answers, and which prompts produce them?”