AEO for nonprofits: being the answer when people ask assistants for help

AEO for nonprofits is answer engine optimisation applied to the questions a charity is best placed to answer: who qualifies for help, how to get it, whether the organisation is legitimate, how to give and how to volunteer. It fits because a correct answer delivered without a click still moves someone towards the support the organisation exists to provide.

That is the reverse of the publisher's position. For a news or media site the content is the product, and an assistant that summarises the article removes the reason to visit, which is why our piece on AEO for media publishers spends much of its length on what to allow and what to withhold. A non-profit rarely faces that trade. If an assistant tells someone accurately that a helpline exists, who it serves and how to reach it, the organisation has done its job, whether or not its analytics ever record the visit.

Answer engine optimisation for charities is therefore less about capturing traffic and more about making sure the answer is right. The risk is not only invisibility; it is being summarised wrongly, with an eligibility rule dropped or an outdated contact route repeated, by an engine that never checked with you. Most of what follows is about reducing that risk with a small team's time.

Service-user questions are answered best by a page whose headings are the questions in the words people use, such as "can I get help with rent arrears if I am working?", each followed by a standalone answer that states who qualifies, how to make contact and what happens next, before any background about the organisation.

The sector page on this site makes the case for putting service users first; the AEO point is narrower and concerns shape. Assistants lift a passage, not a page, so the passage under each question has to carry the whole answer: the eligibility condition, the route in, and the realistic next step. A page that opens with the charity's history and reaches the eligibility rule in the fourth paragraph gives an engine nothing to quote, or worse, something partial to quote.

Collect the questions from the people who already answer them. Helpline staff, caseworkers and whoever reads the general inbox know the exact phrasing, including the questions people are embarrassed to ask. Write each answer as a caseworker would say it aloud, name the service in the answer rather than writing "this service", and state what happens after contact, because not knowing what happens next is what stops many people making the call.

Safety and eligibility caveats on sensitive topics, such as health, crisis support, immigration status, legal problems or debt, belong inside the same passage as the answer they qualify. An assistant that lifts the answer without a caveat stated three paragraphs later can send someone down a route that does not apply to them, or delay urgent help.

The rule is simple to state and uncomfortable to follow, because it makes every answer longer and less tidy. "You can apply for emergency support online" is quotable; "You can apply for emergency support online, but if you are in immediate danger, call the emergency services first" is both quotable and safe. The extra clause costs a line and removes the worst failure an extracted answer can cause. Where the rules differ by country or region, say which jurisdiction the answer covers inside the passage itself.

Dates matter here more than anywhere else on the site. Eligibility rules, opening hours and referral routes change, and an assistant may repeat an old version long after the page is corrected. Show a real "last reviewed" date on sensitive answers, review them on a schedule someone owns, and remove superseded pages rather than leaving them live alongside the new ones.

An assistant identifies a non-profit by matching its name, registration details, purpose and location across the organisation's own site, charity-regulator listings and other reference sources. Where those agree, the organisation resolves as one entity an assistant can name; where they differ, the assistant hedges, merges it with a similarly named body, or leaves it out.

Non-profits are unusually prone to fragmentation. The legal name on the register is often not the name people use, local branches publish their own sites, and a merger or rename can leave the old name alive on directories for years. Choose the canonical form of the name, state the legal name and the registration details beside it on the site exactly as they appear on the register, and correct the listings that still describe the old organisation.

Structured data should repeat those same facts, not improve on them. An Organization node, or its NGO subtype where that describes you, carries the legal name, address, logo and sameAs links to the register entry and official profiles, and it must match the visible page. Wikidata is worth an entry only where notability genuinely exists, meaning independent, serious coverage; an entry created without it tends to be deleted, and our Wikidata service page leads with when not to attempt one.

Assistants answer legitimacy questions about a charity mainly from third-party sources: the relevant charity regulator's register, independent evaluator listings, press coverage and, sometimes, published complaints or warnings. A non-profit cannot answer the question about itself convincingly, so the work is making sure those outside sources are accurate, current and consistent with each other.

Start with the register. If the filed accounts are late, the listed purpose is out of date or the contact details point to a closed office, that record is what an assistant may repeat when someone asks whether the organisation can be trusted. Filing on time is a governance duty anyway; for AEO it also keeps accurate one of the records an assistant can consult. Check the evaluator and directory listings next, and correct anything factually wrong through each listing's own process rather than by publishing a rebuttal.

Link to the register entry from the site, so a reader and a model can verify the registration without searching for it, and describe the organisation in the same words the register uses. Do not publish badges, ratings or endorsements you cannot link to at source. A legitimacy signal that fails verification does more damage than its absence, because it turns a neutral answer into a suspicious one.

Donation and volunteering questions are answered best by short standalone passages that say exactly how to give, including the accepted methods and whether regular gifts are possible, and exactly how to volunteer, including the roles, the commitment and the first step. Those passages should make no fundraising or impact claim the organisation cannot source.

Donors ask assistants what a gift achieves, and the temptation is to answer with a round number. Resist it unless the number is measured. Any impact figure the organisation publishes should be traceable to its accounts, an evaluation report or a named dataset, linked from the passage that states it, and dated. An assistant quoting an unsourced figure repeats it without context; a sourced figure travels with the means to check it. Where the honest answer is that outcomes are hard to measure, say so, and describe the method instead.

Volunteering answers can be brief, particularly where recruitment runs through a third-party platform. A passage that names the kinds of role, where they are based, any checks required and the link to apply is enough for an assistant to route someone correctly. What it must not do is describe roles that have closed, which is an easy way for a volunteering page to mislead the people most willing to help.

A small non-profit team should do AEO in order of harm and reach: fix the answers to service-user questions first, then the safety caveats, then the organisation's facts on its site and the regulator register, then the donation and volunteering passages. Schema refinements, Wikidata and prompt tracking come later, or not at all.

The list below is sequenced for an organisation with a few hours a week, one person who can edit the site and no developer on call. Each step is useful on its own, so stopping partway still leaves the site in a better state, and none of the early steps depends on a tool or a subscription. The skip list matters as much as the do list, because effort spent on the wrong things is how a small programme stalls.

A non-profit measures AEO with a small fixed panel of prompts, perhaps twenty to forty questions a service user, donor or volunteer would ask, run cold on each engine at a regular interval, recording whether the organisation is named, whether the answer is accurate, and whether any safety caveat survived.

Accuracy is the measure that matters most here, more than presence. An assistant that names the charity but gives the wrong eligibility rule is a failure worth fixing before any gain in mentions. Keep the panel fixed so results are comparable, use fresh sessions with no account history, and read the answers rather than only counting names. Our article How to Measure Who Gets Cited by AI sets out the method in full, including why a single screenshot proves nothing.

Two cautions are specific to the sector. Never paste a real service user's question, with its personal details, into an assistant to test it; write a representative version instead. And do not report the panel as impact: it describes how assistants answer, not how many people were helped, and treating one as the other would be exactly the unsourced claim this guide warns against.