03 — Score my next campaign

Question: Before I commit budget on this list of postcodes, rank them against my existing donor profile.

Outputs: answer.md (ranked list, cut-line recommendation, channel suggestions) + answer.html (MapLibre choropleth of the target postcodes only, coloured 0-100 by similarity score with the recommended cut-line ring highlighted in gold, CARTO light_all basemap).

Files in this folder

File	What it is
`prompt.md`	Instruction set.
`sample_donors.csv`	500 synthetic donors.
`sample_campaign_targets.csv`	31 postcodes mixing already-donors, demographic look-alikes, mid-match outers, and low-match outers/regional/interstate. Designed to surface a meaningful cut-line.
`CLAUDE.md`	Auto-load for Claude Code.
`expected_output.md`	Shape of a good markdown answer.

Nothing else is bundled. ABS boundaries + demographics come from the public endpoints listed in the prompt.

How to test

cd starter/examples/03_score_next_campaign
claude
> "Run CLAUDE.md."

What the target list is designed to do

The 31 targets are deliberately mixed:

4 already-donors (should score high but offer no new lift).
~10 demographic look-alikes that have never been asked (the real opportunities — should also rank near the top).
~6 mid-match outer suburbs (the cut-line should fall here).
~10 low-match outer/regional/interstate (should rank at the bottom and be excluded).

A passing answer recommends a cut-line around score 65 and renders it visually with a gold ring on the map.