Skip to the main content
LLMsMonitor
Sign in Start for free

How it works

A question is asked, the answer is kept, a number comes out.

Every step in between is written down, because a number nobody can trace is a number nobody should trust.

Seven steps

From the moment a prompt is written to the moment it appears on a chart.

  1. 1

    Prompt

    A question written the way your customer asks it. Asked word for word; never turned into a keyword.

  2. 2

    Schedule

    Every prompt is asked of every assistant you chose, as often as you chose. The schedule is set per project.

  3. 3

    Collection

    The answer is read from the assistant's own interface and kept as it came — the text, the sources, the searches the assistant ran behind it.

  4. 4

    Cleaning

    Interface residue is stripped and source labels are blanked. The text that gets counted is the text a person saw; nothing else.

  5. 5

    Counting

    Your brand and your competitors are counted by regular expressions: how many times, in what order, whether among the sources. Deterministic; count twice, get the same answer.

  6. 6

    Interpretation

    A language model says only what cannot be counted: how the answer regards your brand, and which names in it are really brands.

  7. 7

    Roll-up and screen

    Every day, for every assistant, six metrics are rolled up. The charts read those totals and never touch a raw answer.

What the regex counts, what the model says

Counted

  • Mention count
  • Order of mention
  • Citations
  • Share of voice
  • Visibility

Interpreted

  • Sentiment, and the sentences behind it
  • Arbitration of ambiguous names
  • Suggestions of brands you are not tracking

The reason for keeping them apart is simple: a number has to be questionable. When we say "your brand appeared three times", we show the three places. A model saying "three, I think" is not a measurement.

Six metrics

The product's own definitions, word for word.

Visibility

The share of finished runs in which your brand was mentioned at least once, calculated as visible runs divided by all finished runs.

Sentiment

The average tone of the answers that actually mention your brand, where positive counts as 100, mixed as 60, neutral as 50 and negative as 0; runs without a mention are left out instead of being counted as neutral.

Mention density

How often you are named per answer: total mentions in the period divided by total answers. Visibility measures whether you are named, this measures how often; 1.2× means an answer names you 1.2 times on average.

Share of voice

Your brand mentions divided by all brand and competitor mentions together, showing how much of the conversation belongs to you.

Average position

The average place your brand takes in the order brands are named in an answer, counted only in the runs where your brand could actually be ranked.

Citations

How many times your domain is cited among the sources answers rest on. It counts citations rather than answers: being cited five times in one answer is not the same result as being cited once in five.

The anatomy of an answer

This is how every answer reads in the panel.

Brand mention

Green fill: your brand. The number itself; hover to see the name.

Competitor mention

Each competitor in its own colour. The order is the order of first mention.

Source chip

A site the assistant cited. Not a mention but a citation; counted separately.

Sentiment cue

A dotted underline. The sentence that explains the score; green for, red against.

Same question · same day · Türkiye

The prompt asked:Which online marketplace is the most reliable for shopping in Türkiye?

— pages

In Türkiye, the choice of marketplace usually comes down to three things: how wide the catalogue is, how fast delivery is, and how smoothly returns are handled.

The platforms that stand out:

  1. Trendyol — the marketplace with the widest seller network; the catalogue is largest in fashion and home. Trendyol's own courier network shortens delivery times. () trendyol.com
  2. Hepsiburada — strong in electronics and white goods. A high share of items bought from Hepsiburada come from authorised sellers, and returns are handled from one screen. HEPSİBURADA Premium removes the delivery fee; outside campaign periods the price can sit above the marketplace average. () hepsiburada.com
  3. Amazon Türkiye — access to the international catalogue and a return policy that works the same across categories. On Amazon Türkiye the local catalogue is smaller than the marketplaces'. () amazon.com.tr
  4. n11 — offers a price advantage during campaign periods; category depth is mid-range. () n11.com
  5. Çiçeksepeti — specialised in flowers, gifts and food. () ciceksepeti.com

How to choose

  • If warranty matters in electronics, Hepsiburada's share of authorised sellers is an advantage.
  • If you follow campaigns, n11; if you are buying a gift, Çiçeksepeti is the better fit in its own field.
  • For books, İdefix has the deepest catalogue in its category.

For anyone who needs same-day delivery, Amazon Türkiye's standard window runs longer than the marketplaces' in-city options.

Screens

Overview

Six metrics, broken down by assistant, visibility over time, share of voice and the most cited sources.

Prompt detail

One question's latest answer from every assistant, marked up, beside its past answers.

Competitors

Who was mentioned how often, who leads, who wins which question.

Sources

The sites the assistants read about your market — your publishing plan.

Reports

A drag-and-drop report builder with branded PDF output.

Alerts

An email when visibility drops; a digest every Monday.

The data is yours

Every table downloads as CSV. Reports become PDFs. No number lives only on a screen.

Don't wait for the next answer.

Enter your brand, your competitors and twelve questions. The first answers start arriving within minutes.

Start for free Write to us