Lumear
Docs
← Back to lumear.aiSign in

Prompt sets

A prompt set is a reusable bundle of buyer questions you run on a schedule. Treat it like a saved query — write it once, run it weekly to track how your visibility moves over time.

Three ways to create prompts

1. AI generator

The fastest. Describe your buyer persona (we suggest starter chips like “Operating partner at a PE firm screening portfolio companies”), pick an intent (informational, comparison, purchase, recommendation, troubleshooting, location-based), and a funnel stage. We generate 10–25 prompts that sound like real questions a real buyer would type into ChatGPT.

2. CSV upload

Already have a list from a workshop or interview? Upload a CSV with columns text, category, and intent. Only text is required.

3. Paste manually

One prompt per line, paste into the textarea. Useful for ad-hoc tests and quick experiments.

Intent matters

Setting an intent on each prompt unlocks two things downstream. First, the entity matcher uses the intent to decide which pages on your site are eligible answers — a contact page won't win an “informational” prompt no matter how high its similarity score is. Second, the recommendation engine suggests different fixes by intent: questions get an FAQ block, comparisons get a comparison table, transactional prompts get a dedicated landing page.

Editing + bulk operations

  • Inline-edit any prompt directly in the table.
  • Toggle is_active to exclude a prompt from runs without deleting it.
  • Bulk-select and delete with the checkboxes.
  • Re-run the generator to add more prompts to an existing set (we preserve the existing ones).

Sizing your prompt set

Sweet spot is 20–40 active prompts per set. Fewer than 10 and your visibility score becomes noisy. More than 50 and you start spending more on LLM calls than you get value back — diminishing returns. Optimize for coverage of the questions your buyers actually ask, not exhaustive permutations.

Reusing prompts

A single prompt set is reused across many runs. Each run captures the responses at that point in time, so you build a longitudinal view of how AI answers shift as you update your site. The dashboard's trend chart compares your latest run to the average of the prior 4.

Prompt sets — Lumear Docs