DocsUse Cases

Website Copy

Financial services websites tend to be written in the language of the product team, not the customer. Dialogue lets you reverse that — pulling the exact words real people use to describe their problems, fears, and decisions, then using that language as the foundation for copy that actually converts.

This guide walks through building a copy brief for a life insurance landing page.

Step 1: Understand the emotional landscape

Before writing anything, get a picture of how people emotionally engage with the product area. Use the stats endpoint to see which emotional registers dominate.

$GET /api/v1/stats

Look at the emotional_register and product_named breakdowns. For life insurance, you’ll typically see grief_loss and anxiety_fear dominating — meaning your audience comes to this product through fear and bereavement, not rational planning. That alone should change how you open a landing page.

Step 2: Pull the highest-engagement verbatim quotes

Likes are a signal of resonance. These are the comments that made thousands of people think “that’s exactly it.” They’re the closest thing you have to tested copy.

$GET /api/v1/quotes?product=life_insurance&min_likes=50&limit=10
1{
2  "quotes": [
3    {
4      "text": "My husband died without life insurance. I had to sell our house within six months. Please, everyone, just get it sorted. I'm begging you.",
5      "like_count": 847,
6      "emotional_register": "grief_loss",
7      "life_stage": "post_bereavement"
8    },
9    {
10      "text": "I kept putting it off for ten years. Then my mate died at 41 with three kids. Sorted mine that afternoon.",
11      "like_count": 612,
12      "emotional_register": "anxiety_fear",
13      "life_stage": "mid_life_accumulator"
14    }
15  ]
16}

These aren’t just quotes — they’re the emotional territory your landing page needs to inhabit. The trigger is almost always a near-miss or bereavement, not a financial planning moment.

Step 3: Search for specific copy angles

Use semantic search to find comments that match particular messaging angles you want to explore — objections, hesitations, turning points.

What made people finally act:

$POST /api/v1/search
${
>  "query": "what finally made me get life insurance",
>  "limit": 10,
>  "threshold": 0.4
>}

The procrastination story:

$POST /api/v1/search
${
>  "query": "kept putting off life insurance and now regret it",
>  "limit": 10,
>  "threshold": 0.4
>}

Cost objections:

$POST /api/v1/search
${
>  "query": "life insurance is too expensive not worth it",
>  "limit": 10,
>  "threshold": 0.35
>}

Each search gives you 10 real comments from people who have been through that specific thought process. Use them to write objection-handling copy, FAQ answers, and supporting body text.

Step 4: Generate a copy brief with the AI analyst

Once you’ve gathered the raw material, use the chat endpoint to synthesise it into a structured copy brief. The AI analyst will pull from the corpus automatically.

$POST /api/chat
${
>  "messages": [
>    {
>      "role": "user",
>      "content": "Write a copy brief for a life insurance landing page targeting mid-life accumulators aged 35-55. Include: the core emotional territory, the primary trigger moment, three headline directions, objection-handling language for cost and complexity, and five verbatim quotes to use as social proof. Ground everything in real consumer language from the corpus."
>    }
>  ]
>}

The response will include a structured brief with evidence, ready to hand to a copywriter or feed directly into a content generation workflow.

Using the MCP server

If you’re working in Claude Desktop, you can run this entire workflow conversationally without making a single API call:

“Look at the life insurance comments in the corpus. What’s the dominant emotional trigger? Find me the five most-liked quotes and use them to write three hero headline options for a landing page targeting people in their 40s.”

Claude will call the corpus tools, retrieve the data, and synthesise a response in one turn.

Tips

  • Segment by life stage — a landing page for first-home buyers talking about mortgage protection needs completely different language than one for post-bereavement customers reviewing their cover. Filter by life_stage before pulling quotes.
  • Filter to high engagement — set min_likes=25 or higher to ensure you’re working with language that resonated broadly, not outlier voices.
  • FAQ copy — run semantic searches for every objection you can think of. The comments you get back are the raw material for FAQ answers written in customer language.