KnitKnot
You are reading the agent-optimized layer of this page: the literal markdown we serve to AI crawlers and assistants, shipped in the page source of every visit. Making sure AI reads the right facts about a company is literally what KnitKnot does.

# Subjects — brands and products

How KnitKnot models your brand and its products separately, and why a multi-product company needs both.

Everything KnitKnot benchmarks is anchored to a **subject** — either your brand or one of its products. A single-product company has one subject and can skip this page. A multi-product company gets one brand subject plus a subject per product, and each gets its own prompt library, competitor set, and report.

## Why products are separate subjects

Buyers don't evaluate "your company" — they evaluate the thing that solves their problem. If you sell three products into three different categories, the AI conversations that matter are different for each: different questions, different competitors, different sources. Modeling them as one blob would average away exactly the signal you need.

Per product, KnitKnot keeps separate:

  • - **Prompts** — generated from that product's category and keywords, so a benchmark asks what that product's buyers actually ask.
  • - **Competitors** — the vendors AI compares *that product* against, which usually isn't your brand-level competitor list.
  • - **Reports and scores** — each product's AI Presence Score trends independently, so you can see one product winning its category while another lags.

## The brand level

Brand-level prompts cover questions about the company as a whole — reputation, comparisons against brand-level rivals, "who makes X" questions. The brand report rolls up alongside the per-product views, and the public report lets readers switch between company-wide and per-product results.

## Family-aware scoring

The product model isn't just organization — it changes scoring. When an AI answering a category question recommends one of your products over another of your products, that's not a loss for the runner-up: both belong to you, and the buyer landed on your brand either way. KnitKnot recognizes the family relationship and scores it as a brand win.

This matters more than it sounds. Multi-product companies routinely look worse than they are in naive AI monitoring because sibling products get counted as competitors. It also surfaces the opposite problem: if AI treats your products as unrelated brands and never mentions the parent company, that's a brand-coherence gap worth fixing — and it will show up in your report.

Raw mirror of this content: https://knitknot.ai/docs/subjects.md. Site-wide summary: /llms.txt · full content: /llms-full.txt

Docs navigation
Docs › Core concepts

Subjects — brands and products

How KnitKnot models your brand and its products separately, and why a multi-product company needs both.

Updated

Everything KnitKnot benchmarks is anchored to a subject — either your brand or one of its products. A single-product company has one subject and can skip this page. A multi-product company gets one brand subject plus a subject per product, and each gets its own prompt library, competitor set, and report.

Why products are separate subjects

Buyers don’t evaluate “your company” — they evaluate the thing that solves their problem. If you sell three products into three different categories, the AI conversations that matter are different for each: different questions, different competitors, different sources. Modeling them as one blob would average away exactly the signal you need.

Per product, KnitKnot keeps separate:

  • Prompts — generated from that product’s category and keywords, so a benchmark asks what that product’s buyers actually ask.
  • Competitors — the vendors AI compares that product against, which usually isn’t your brand-level competitor list.
  • Reports and scores — each product’s AI Presence Score trends independently, so you can see one product winning its category while another lags.

The brand level

Brand-level prompts cover questions about the company as a whole — reputation, comparisons against brand-level rivals, “who makes X” questions. The brand report rolls up alongside the per-product views, and the public report lets readers switch between company-wide and per-product results.

Family-aware scoring

The product model isn’t just organization — it changes scoring. When an AI answering a category question recommends one of your products over another of your products, that’s not a loss for the runner-up: both belong to you, and the buyer landed on your brand either way. KnitKnot recognizes the family relationship and scores it as a brand win.

This matters more than it sounds. Multi-product companies routinely look worse than they are in naive AI monitoring because sibling products get counted as competitors. It also surfaces the opposite problem: if AI treats your products as unrelated brands and never mentions the parent company, that’s a brand-coherence gap worth fixing — and it will show up in your report.