← 5-notebook set · Notebook 2 of 5 · 5-min read

Notebook 2 — Intent

Per-category breakdown of the 5 prompt categories · This notebook answers: "Which query types are bleeding the most citations, and which is the best place to start?"

The 5 prompt categories

We bucket every prompt into one of 5 categories. The category a brand wins on tells you which buyer journey the brand is visible in. The category a brand loses on tells you which buyer journey to invest in first.

Category A — Buying intent (7 prompts)

These are queries a buyer types when they are actively looking to buy. Example: "best GEO audit service for跨境品牌 in 2026".

Intent reading: At Day 1, no buyer searching for a GEO audit tool finds us. The Chinese-engine buying-intent answers skew to established CN vendors; the Western-engine answers skew to Profound/Otterly. The gap — "bilingual tool" — is the wedge that the v3 plan targets.

Category B — Competitive (7 prompts)

These are queries a buyer types when they are comparing vendors. Example: "Profound vs Otterly vs Peec.AI 对比".

Intent reading: This is the prompt category that produced our single data point of the day — Perplexity naming tryprofound.com on the prompt "alternatives to Profound for跨境品牌". This is the signal we needed: the prompt set is working, the LLMs are not broken, and the gap in the answer is precisely the bilingual / 跨境 positioning the v3 plan names.

Category C — Methodology (6 prompts)

These are queries a buyer types when they are educating themselves. Example: "what is GEO (generative engine optimization)".

Intent reading: This is the prompt category that LLM answer engines are most confident on — they all know what GEO is, and they all cite Princeton, llms.txt, and Baidu Zhanzhang. The Clarivy opportunity here is not to win on "what is GEO" — the LLM answers that already. The opportunity is to be cited as a source for a methodology datapoint (e.g. "per Clarivy's Day-1 self-audit, 0/270 is the expected baseline for a brand-new domain"). That requires us to publish 30+ unique datapoints, one per methodology question, with named authorship and CC-BY licensing.

Category D — Brand-specific (5 prompts)

These are queries a buyer types when they have heard the brand and want to evaluate it. Example: "clarivy.ai 是什么".

Intent reading: This is the most reassuring category. Every engine we tested refused to hallucinate a Day-1 brand. Perplexity even correctly identified that the domain is registered but lacks public content. This is the anti-hallucination safety check working as designed. The category will move first as we publish content; expect 1-2 of the 9 engines to begin surfacing clarivy.ai here within 30 days of consistent publishing.

Category E — Purchase intent (5 prompts)

These are queries at the very bottom of the funnel. Example: "GEO audit pricing 2026", "is GEO audit worth it".

Intent reading: ChatGPT, Perplexity, and Gemini are explicitly naming Profound and Otterly price points on purchase-intent queries. We are absent. This is the category that converts — losing it is the most expensive line in this audit. First action: make our $99/$299/$1,499 pricing page indexable and let the LLMs cite it. See Notebook 5 §1.

Summary by category (the 1-page version of this notebook)

Next: Notebook 3 — Content →   ← Notebook 1