Brand perception

Visibility tells you whether you appear, sentiment in what tone. The Brand perception page (in the menu, under Brand) shows what exactly AI answers say about you: "fast", "certified", "expensive", "hard to reach" — grouped into fixed categories, so you can compare yourself with your competitors.

What we read from an answer#

For each answer, an AI model reads what the answer says about your brand and the competitors named in it. Every statement gets:

• a category from a fixed list (below) — the same for every brand, so the numbers can be compared;
• a polarity: Strength (a reason to choose the brand), Weakness (a drawback or a reason for caution) or Mentioned (a plain fact, with no tone: "has a showroom on Main Street", "does technical translations");
• the quote from the answer, in the answer's language, and a short detail.

Some rules the model follows:

• Only statements about a business named in the answer count. Products, places and institutions ("marble", "the city hall") are not brands and are left out.
• The drawbacks of a type of product are not given to the business that sells it: "hexagonal tiles take longer to lay" is not a weakness of the tile shop.
• A statement goes to the brand the sentence is about, not to every brand in the paragraph.

Since when#

Only answers analyzed after the feature was launched have categories, so the page counts answers from the first one analyzed this way ("Answers analysed since …" next to the tabs). Until then the page shows Collecting data.

How we count#

The unit is the answer, not the statement. An answer that calls you fast three times counts once for Speed.

rate = answers that say it about the brand ÷ answers that name the brand

"Answers that name the brand" are those with a mention of it, with the period and model filters applied. Next to each rate you see the raw numbers ("14/33").

Example#

33 answers name your brand in the period. 14 of them present it as fast (Speed, Strength); 2 say delivery takes long (Delivery, Weakness).

Speed — strength:     14 ÷ 33 = 42.4%
Delivery — weakness:   2 ÷ 33 = 6.1%

A category can be both a strength and a weakness — some answers praise your prices, others find them high. It then appears in both lists.

Overview#

Strengths and weaknesses#

Your brand's categories with at least one Strength (or Weakness), most frequent first. When no answer names a weakness, the list says so: AI rarely brings up drawbacks on its own. To find out what it would say, add a prompt such as "what are the downsides of [your brand]?".

Distinctive#

A strength is marked Distinctive when it is said of you much more often than of your competitors:

• your rate is at least 2× the highest rate among the top 5 direct competitors named in the period;
• at least 5 answers name it as your strength (if no competitor has it, 5 answers are enough);
• at least one competitor was named in the period, so there is something to compare with.

Little data#

Below 30 answers that name your brand in the period, the page warns that results may still change, and nothing is marked Distinctive.

You vs. your top rival#

Side-by-side bars for the strengths of your brand and of one competitor, each as the share of the answers that name that brand. The rival is, by default, the direct competitor named most often in the period; you can pick another from the list (the 10 named most often).

You vs. the market#

A table of your brand and the 5 direct competitors named most often, across the 8 categories brought up most often. Each cell is the share of that brand's answers that say anything about the category — strength, weakness or plain mention. The darker the cell, the higher the share. A red dot means at least one answer names it as a weakness.

Only direct competitors appear on this page. Mark competitors as direct on the Competitors page.

By prompt#

For each prompt with answers that name your brand: how many answers, and the top 3 categories they praise you on and fault you on, with the number of answers.

The answers behind a number#

Click a category — in the lists, a bar, a cell in the table or a badge in By prompt — to see what is behind it:

• Cited pages: the pages those answers cite, by how many of them cite each one. Pages of your site are marked Your site.
• Answers: the latest 50, with the model, the prompt, the date and, for each statement, the polarity, the detail and the quote. Click an answer to read it in full.

The numbers in the window are the same as in the cell you clicked: same period, same model, same prompt.

Turn a weakness into an action#

When a weakness of your brand comes from answers that cite a page of your site, that page has an Add to Actions button. It creates a manual recommendation on the Recommendations page, "Weakness in AI answers: <category>", with the number of answers and up to 3 quotes from the answers that cite the page — what to fix or clarify on it. You handle it there like any other recommendation.

Each page and category can be added once. The recommendation does not close on its own when the weakness goes away.

Categories#

How to read it#

• A strength that is Distinctive is your argument in front of customers — and the one to keep visible on your site, so answers keep finding it.
• A strength your rival has and you don't is often information missing from your site, not a real gap: check the cited pages behind the rival's bar.
• A weakness behind a page of your site is the easiest to fix: the page says it, or leaves it unclear.

Where it appears#

• Brand perception — everything above.
• Competitors — the attributes window of a competitor still shows the free-text attributes, including those from before this page.