— Core principle
Generative AI does not touch donor-facing copy. Full stop.
Every AI use case below is either (a) operational — making the team faster — or (b) operating on already-human-approved copy. The moment an AI writes a sentence that a donor reads without a human signing off on it, NDOT carries reputational risk that no conversion lift justifies.
— USE CASE 01
Copy-variation factory (with human approval)
AI generates 5–8 hero headline variants per campaign. Brand + comms team approves the shortlist. Site shows variants A/B. Winner ships. AI is a copywriter assistant, never the publisher.
↗ Faster A/B cycle · 2–4× variant volume
— USE CASE 02
Segment-aware donation prompts
Lapsed donors, repeat donors, and first-time visitors see different (human-written) pre-set amounts and asks. AI does the segmentation in real time. Humans wrote every word the donor reads.
↗ Conversion lift target +3–7%
— USE CASE 03
AI-search readiness
Structure your campaign pages so ChatGPT, Perplexity, Claude, and Gemini cite Cancer Council when a donor asks "when is Daffodil Day 2026". This is the biggest organic shift since SEO began.
↗ Organic donor acquisition · new channel
— The AI-search opportunity, concretely
This is what's already happening — and where Cancer Council needs to be cited.
Most charity-sector boards still think SEO is the game. SEO is the rearview mirror. The donor of 2026 is asking ChatGPT and Perplexity questions like:
DONOR ASKS: "when is daffodil day 2026"
→ AI gives a date. Which charity does it CITE for the answer? That's where the donor clicks. That's where the donation comes from.
DONOR ASKS: "best cancer charity to donate to in NSW for breast cancer research"
→ AI synthesises an answer. If Cancer Council NSW's site is structured for LLM ingestion, you're in the top 3 cited sources. If not, you're not in the answer at all.
DONOR ASKS: "I want to host a morning tea — how do I do that"
→ AI walks the donor through it. Are the instructions cited from biggestmorningtea.com.au, or from a third-party blog that scraped your content three years ago?
What I'd ship in Q1: LLMs.txt + llms-full.txt manifests on every campaign site, FAQPage + Event schema on every page, plain-text answers to every donor question above the fold (LLMs scrape text not images), monitored daily against the top 4 AI search engines.
These are honest numbers. Not the 30% lifts the AI-snake-oil consultancies promise. Real, defensible, and ship-able inside the contract window.
AI in a cancer-charity fundraising context is a force multiplier for the team — never a substitute for the human voice the donor is owed.