Why nonprofit communicators need an LLM strategy — and what that actually means for your website.
By now, you know from your declining page visits and uptick in LLM traffic that most audiences are getting their content from Claude or ChatGPT. Search isn’t always where the audience starts anymore. This is part of the current AI wave that most nonprofit comms teams haven’t caught up to yet.
People are asking ChatGPT, Claude, Gemini, or Perplexity a question — and those systems are going out into the web, grabbing and reading your content, and deciding whether your org’s content is their answer.
You’re not just writing for your audience…. you’re writing to see if AI will decide to quote you, paraphrase you, or skip you entirely. As the nonprofit digital strategy firm Elevation puts it, the digital world is no longer just about search rankings — it’s about being the source of reliable answers.
The good news: the skills that make for great content strategy — clarity, structure, authority, consistent publishing — are exactly what LLMs want to see. The bad news: many nonprofit websites are still built on legacy assumptions about how the web works. Buried PDFs. Walls of unstructured text. Content that was last updated in 2021. These sites exist on the web but are functionally invisible to AI-mediated discovery.
LLMs don’t scroll. They read structure. Your H1 is a handshake. Your section headers are a table of contents. Your update date is a credibility signal.
Here’s 10 ways nonprofit comms folks want their website to be readable — and citable — by AI systems like the ones becoming central to how audiences find information:
- Use clear, descriptive headings. H1, H2, H3 tags aren’t just visual — they’re how LLMs understand your page’s structure. I try to write headings that would make sense even if they were read out of context, because odds are, an AI will do just that.
- Answer questions directly, up top. LLMs favor content that frontloads the answer. Don’t bury your lede. If your page is about food security programs in West Africa, say so in the first sentence — not the fourth paragraph.
- Publish consistently. Freshness is a trust signal. A page updated six months ago is more credible to an AI than one last touched in 2019. Aim to review and update key pages quarterly at a minimum. As one nonprofit SEO guide puts it, nonprofit SEO isn’t a one-time project — it’s a practice, like tending a garden.
- Write in plain, precise language. Jargon confuses LLMs the same way it confuses readers. Short sentences, active voice, concrete nouns. If a 10th grader couldn’t follow it, rewrite it.
- Use schema markup.Structured data — Article, Organization, FAQ, Event — tells AI systems exactly what your content is. For nonprofits, there’s even a specific NGO schema type. A developer can add this in an afternoon using Schema.org‘s vocabulary, and it pays dividends for years. Whole Whale has a solid rundown of the top schema types for nonprofits.
- Make your “About” content authoritative. LLMs place high weight on who-you-are signals. Your About page, leadership bios, and mission statement should be crisp, factual, and regularly updated. This is how AI decides whether you’re a credible source worth citing.
- Link internally with intention. Contextual internal links help LLMs map the relationships between your content. Don’t just link to “click here” — link from meaningful anchor text that signals what the destination page is about.
- Create a sitemap and submit it. XML sitemaps remain the most reliable way to tell crawlers — including the ones feeding AI training data — what exists on your site. Keep it current. If you’re on WordPress, Yoast SEOgenerates and maintains one automatically.
- Write substantive long-form content, not just news blurbs. LLMs tend to cite content that goes deep. A 1,200-word explainer on your program model is more likely to be surfaced than a 150-word announcement. Depth signals expertise.
- Don’t lock your best content in PDFs. PDFs are poorly indexed by most AI systems. If your flagship report or impact data lives only in a PDF, publish a web-native version alongside it — even a summary page with key findings as plain HTML text.
A few tech notes about content and LLMs
Beyond just the content, a few low-lift tech choices matter.
- Make sure your site loads fast — page speed affects crawl depth.
- Use HTTPS.
- Recently at my work, we discovered the robots.txt file may have ben blocking AI crawlers (a surprisingly common mistake — check it directly at yourdomain.com/robots.txt).
- Use Google Search Console to monitor indexing health and catch crawl errors before they become invisible-page problems.
You may also have heard about llms.txt — a proposed new file format, analogous to robots.txt, that would tell AI systems which pages on your site matter most. The idea was proposed in 2024 by Jeremy Howard of Answer.AI, and it’s gaining attention in technical SEO circles. But be measured about it: as of mid-2025, none of the major AI providers — OpenAI, Google, or Anthropic — have confirmed they actively use it, and server log analysis by several publishers found that AI crawlers weren’t even requesting the file. It’s worth understanding and may matter more as standards evolve, but robust content structure and a clean sitemap will do more for your AI visibility right now than an llms.txt file.
The underlying principle is simple: make it easy for smart systems to understand what you do, why it matters, and why you’re a trustworthy organization. That’s good communications strategy. It always has been. LLMs just make it more measurable.