How do nonprofits show up in ChatGPT and Google AI answers? The 2026 AI visibility guide
A nonprofit shows up in ChatGPT, Google AI Overviews, AI Mode, Claude, Perplexity and Copilot when five things work together: retrieval crawlers can reach its site, the organization is identifiable as a single entity, its claims are dated and verifiable, its content can be extracted passage by passage, and independent sources such as charity registers confirm the same facts. A page-one ranking in Google does not guarantee any of the five.
2026 edition. Covers the United States, the United Kingdom, German-speaking Europe, the EU and smaller markets.
A nonprofit can rank on page one of Google and still be invisible inside ChatGPT, Google AI Overviews, AI Mode, Claude, Perplexity and Copilot. The two outcomes come from two different systems asking two different questions.
| System | Question it asks |
|---|---|
| Google's classic index | Which page best matches this query? |
| AI answer engine | Which organizations are real, relevant and verifiable enough to name safely? |
Most nonprofit websites are built to pass only the first test.
The scale of the shift is no longer speculative. The reach figures below are the vendors' own.
| AI surface | Reported reach | Reported |
|---|---|---|
| Google AI Overviews | More than 2.5 billion monthly active users | Google, June 2026 |
| Google AI Mode | More than 1 billion monthly users | Google, June 2026 |
| ChatGPT | Roughly 900 million weekly active users | February 2026 |
A share of the people who once found a nonprofit through a list of blue links now meet it, or fail to meet it, inside a paragraph of generated text.
Key takeaways
- AI visibility does not replace SEO. It sits on top of it.
- Five things have to work together: crawler access → entity clarity → verifiable evidence → extractable content → external corroboration.
- A site can rank well and still be hard for a machine to interpret, attribute or trust.
- The target is not more traffic. It is becoming a trusted entity and a usable source across AI discovery systems.
- An inaccurate mention is worse than no mention, and almost nobody is monitoring for it.
Why are clicks from search falling for nonprofits?
Clicks are falling because AI search answers the question on the results page. Pew Research Center analysed March 2025 browsing data from 900 US adults: users clicked a search result on 8 percent of visits with an AI summary, against 15 percent without one, and clicked a link inside the summary about 1 percent of the time. Semrush found that 92 to 94 percent of AI Mode sessions ended without a click to an external site.
The Semrush figure comes from US clickstream data collected between 1 May and 5 July 2025. Pew published its analysis on 22 July 2025.
Nonprofit search strategy ran on the same five-step model for twenty years. AI search collapses that sequence.
| Model | Sequence |
|---|---|
| Classic search | Search → ranking → click → website → action |
| AI search | Question → synthesis → three to six organizations named → one or two citations → action |
In the classic model, a prospective donor searched "best homelessness charities in Boston," opened four tabs, compared them, and gave to one. In the AI model, the user can learn a nonprofit's mission, location, target population and rough size without ever loading its homepage.
Google's public position is that AI features still drive substantial traffic to the web. Both findings can be partly true: total volume can grow while an individual site's click-through rate falls. The composition of a nonprofit's visibility has changed either way.
A nonprofit website now does two jobs at the same time.
| Role | Audience | Optimized for |
|---|---|---|
| Destination | Humans | Persuasion, trust, conversion |
| Evidence repository | Machines | Extraction, verification, attribution |
An AI system may build one answer about an organization out of a sentence from its impact report, a line from a regulator's register, a funder's grant listing and its programme page. The organization influenced the answer. It did not get the session.
Nonprofits that evaluate digital performance purely through sessions, pageviews and rankings now measure a shrinking part of their visibility.
The organization that wins AI discovery is often not the one with the biggest homepage. It is the one whose identity, programmes, geography and results are easiest to verify.
What is the difference between page visibility and entity visibility?
Page visibility is about ranking a URL. Entity visibility is about whether a machine can recognize the organization itself, independent of any one page, and confirm that its website, register entry, profiles and reports describe the same body. Traditional SEO optimizes a URL. AI systems often need to resolve an entity, and when they cannot, they skip the organization or blend it with another one.
Four terms circulate for this work. The terms are less different than the people selling them suggest.
| Term | Full name | What it optimizes for | Unit of competition |
|---|---|---|---|
| SEO | Search Engine Optimization | Ranking a page in results | A URL |
| AEO | Answer Engine Optimization | Making information easy to extract as a direct answer | A passage |
| GEO | Generative Engine Optimization | Raising the odds of being referenced in a generated response | A claim |
| AI visibility | (umbrella term) | Being understood, retrieved, named and cited | The organization |
Google has publicly treated AEO and GEO as vocabulary layered on top of ordinary SEO, not as a separate discipline. That framing is broadly right, with one important exception for nonprofits: the unit being optimized.
An entity is the organization itself, independent of any one page. A fictional example, Hope Children Foundation, exists across:
- its own website
- a Google Business Profile
- a national charity register
- a donor-rating or seal-of-approval body
- funder and grant databases
- news coverage
- partner organizations' websites
- annual reports and audited accounts
A machine has to decide whether all of those references describe one organization or several.
The real optimization target is not a URL such as example.org/youth-programme. The target is a fact that holds up everywhere:
Hope Children Foundation is a registered nonprofit based in Chicago providing after-school education for children aged 6 to 14. In 2025 it served 1,247 children across 14 schools.
The more consistently and verifiably that sentence appears across sources, the easier the organization is to name.
How do AI systems decide which nonprofits to name?
AI systems name the nonprofits they can retrieve evidence for and verify. Generative search does not rank ten results. A reasonable simplified model of the pipeline is: query, query expansion, retrieval, evidence selection, synthesis, citation. Google has described its AI Search systems as using query fan-out, which decomposes one complex question into multiple related searches before an answer is assembled.
Suppose someone asks:
"Which nonprofits provide job training for refugees in New York?"
The system may effectively run several searches at once.
| Implicit subquery | What satisfies it |
|---|---|
| Refugee-serving nonprofits in New York | Programme pages, directories |
| Employment and vocational training programmes | Service descriptions |
| Eligibility and referral criteria | Explicit eligibility sections |
| Registration and legitimacy | Statutory register entries |
| Programme outcomes | Dated impact statistics |
| Service geography | Address and areaServed data |
One well-optimized landing page cannot answer all six subqueries. That is the structural reason single-page SEO underperforms in AI search.
Why do third-party registers carry more weight than a nonprofit's own claims?
Structured, externally maintained data is disproportionately useful to a retrieval system because it is attributable and hard to fake. A regulator's register entry is machine-readable, dated and independently maintained, so it resolves an entity. A slogan on the organization's own website resolves nothing. The nonprofit's own site is still the anchor, and it is much stronger when independent sources agree with it.
| Signal | Example | What a machine can do with it |
|---|---|---|
| Own website | "We transform thousands of lives every year." | Nothing. No entity, quantity or date to resolve |
| Regulator's register | Organization: Hope Children Foundation. Registration number: 12-3456789. Activity: Youth development. Income: 2.3M. Location: Chicago, Illinois | Resolve the entity. The entry is machine-readable, dated and independently maintained |
The register table in the entity consistency section maps those independent sources across jurisdictions.
How do you run a 90-minute AI visibility audit?
The 90-minute AI visibility audit scores six areas at five points each, for a total of thirty points: crawler access, entity clarity, external verification, citeable evidence, structured data and measurement. Run it before writing another article. The score shows whether machines can reach the site, identify the organization and verify its claims, and whether anyone is measuring the result.
Area 1: crawler access (5 points)
Area 1 checks whether machines can get in.
| Check | How |
|---|---|
| Read your robots.txt | curl -s https://yoursite.org/robots.txt |
| Test for a firewall block | curl -A "OAI-SearchBot" -I https://yoursite.org/ and expect 200, not 403 |
| Repeat for other retrieval bots | Same command with PerplexityBot, Claude-SearchBot |
| Confirm server-rendered content | View page source (not the browser inspector) on a programme page |
| Confirm the sitemap | https://yoursite.org/sitemap.xml returns and is current |
Area 2: entity clarity (5 points)
Open the homepage and the About page. A stranger should find all eight items inside 30 seconds:
- legal name
- working name
- registration number
- country and city
- mission in one sentence
- who you serve
- where you serve them
- how to contact you
If any of those items requires two clicks or opening a PDF, dock a point.
Area 3: external verification (5 points)
Search the organization's name alongside the national register, the seal or rating body, LinkedIn and "annual report." Open each result. Note every fact that disagrees.
Area 4: citeable evidence (5 points)
Search the site for the phrases thousands of, countless, many, transformed, impacted. Each hit is a sentence a machine cannot use.
Then count sentences of this shape:
"In 2025, our after-school programme served 1,247 children across 14 schools in Cook County."
The ratio of dated, counted sentences to vague ones matters more than volume.
Area 5: structured data (5 points)
Run the homepage through Google's Rich Results Test and the Schema Markup Validator. Check for Organization plus NGO, legal name, URL, logo, address, contact point, sameAs, and nonprofitStatus where applicable.
Area 6: measurement (5 points)
The organization should be able to answer three questions today:
- How many impressions did the site get in Google's generative AI features last month?
- Does any traffic arrive from
chatgpt.comorperplexity.ai? - For which prompts does an AI name the organization, and which organizations does it name instead?
Scoring
| Score | Band | What it means |
|---|---|---|
| 0 to 10 | Weak foundation | Machines may struggle to find, identify or verify you |
| 11 to 20 | Partial | Discoverable, but evidence or consistency is patchy |
| 21 to 25 | Strong | Entity is clear and most key information is machine-readable |
| 26 to 30 | Advanced | Access, consistency, evidence and measurement all in place |
Expect a low score on a first pass. A result in the weak or partial band is common for small and mid-size nonprofits, and the gap usually sits in areas 1, 3 and 4.
How do you make sure AI crawlers can reach your site?
Check four things before optimizing any content: robots.txt allows the search and user-triggered crawlers, the firewall or CDN does not return a 403 to them, the Search generative AI control in Google Search Console includes the site, and key content is served as crawlable HTML. Content on a site that blocks retrieval bots cannot be retrieved, however well it is written.
Which AI crawler does what?
Each major AI vendor runs a small fleet of crawlers split by job: a training bot that collects content for future models, a search bot that indexes pages for AI answers, and a user bot that fetches a page the moment someone asks about it. Treating all AI crawlers as one category is the most expensive mistake in this field.
| Vendor | Training crawler | AI search indexing | User-triggered fetch |
|---|---|---|---|
| OpenAI | GPTBot |
OAI-SearchBot |
ChatGPT-User |
| Anthropic | ClaudeBot |
Claude-SearchBot |
Claude-User |
| Perplexity | not declared separately | PerplexityBot |
Perplexity-User |
Google-Extended |
Googlebot |
not applicable | |
| Apple | Applebot-Extended |
Applebot |
not applicable |
| Common Crawl | CCBot |
not applicable | not applicable |
Two consequences follow from the split.
First, blocking GPTBot does nothing for ChatGPT search visibility, and blocking OAI-SearchBot does everything. OpenAI's documentation tells publishers that sites blocking OAI-SearchBot will not appear in ChatGPT search answers, though navigational links may still appear.
Second, Google-Extended is not a Google Search control. Google states that it governs whether crawled content can be used for certain Gemini training and grounding purposes, that it is not a Search ranking signal, and that it does not affect inclusion in Google Search.
A nonprofit can therefore legitimately say: do not train on us, but do cite us.
What should a nonprofit's robots.txt look like?
A nonprofit that wants AI visibility should explicitly allow the search and user-triggered crawlers: Googlebot, OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot and Perplexity-User. Blocking the training crawlers (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended and CCBot) is a separate, optional decision for the board and does not affect AI search visibility. The two blocks below are a starting point.
For a nonprofit that wants maximum discoverability and has no objection to training use:
# Search and retrieval: allow these if you want AI visibility
User-agent: Googlebot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Perplexity-User
Allow: /
Sitemap: https://www.example.org/sitemap.xmlIf the board decides against training use, add the block below and leave everything above untouched:
# Model training: optional opt-out, does not affect AI search visibility
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: Google-Extended
Disallow: /
User-agent: Applebot-Extended
Disallow: /
User-agent: CCBot
Disallow: /Two caveats apply. Compliance with robots.txt is opt-in: a directive only works if the bot reads and honours it. Some crawlers have historically ignored robots.txt, and a user-agent string can be spoofed.
When a person asks an assistant to summarize a specific URL, providers often treat the request as user-directed access, not crawling, so crawl rules may not apply the way a site owner expects. robots.txt is a norm, not a lock.
Why can a firewall block AI crawlers when robots.txt allows them?
robots.txt and the firewall are separate controls, and the firewall is probably the real problem. A site's robots.txt can say Allow while Cloudflare, Sucuri, Akamai or the host's bot protection returns a 403 to the same crawler. Small nonprofits are especially exposed, because bot protection is usually switched on by a volunteer or agency during an incident and never revisited.
Firewall blocks surprise nonprofit teams more often than any other access problem. Check logs and edge rules for the four symptoms below.
| Symptom | Where to look |
|---|---|
| 403 or 429 to AI user agents | Server access logs, CDN analytics |
| Managed challenge or CAPTCHA on non-browser traffic | WAF bot-management rules |
| Geo-blocking left over from an old spam incident | Firewall country rules |
| Aggressive rate limiting on key paths | Rules applied to /programmes/, /impact/ |
What is the Search generative AI control in Search Console?
The Search generative AI control is a Google Search Console setting that lets a site owner decide whether the site's content can be used in Google's AI search features. Google announced it on 3 June 2026 alongside the Search generative AI performance reports, and both were rolled out to all websites worldwide as of 31 August 2026. The control is currently all-or-nothing at property level.
The UK's Competition and Markets Authority has set a deadline of March 2027 for Google to introduce page-level controls for generative AI features.
The control is the fastest way for a nonprofit to erase itself from AI search by accident. A well-meaning trustee, IT volunteer or agency reads a headline about AI scraping charities, opens Search Console, finds a toggle that looks like an AI opt-out, and switches it off.
The organization then disappears from AI Overviews and AI Mode, and nobody notices for a quarter because no alert fires.
Action: open Search Console, confirm the Search generative AI control includes the site, and write down who has access to that setting. Add the check to the annual governance checklist.
Which content should not depend on JavaScript?
Mission, programmes, eligibility criteria, impact figures, locations, donation information and contact details should not depend on client-side JavaScript. Google can render JavaScript, but it still recommends server-side rendering or pre-rendering for important content, and not every retrieval crawler executes JavaScript reliably. The rule: if the information matters for discovery, serve it in crawlable HTML, at a stable URL, in one hop.
Real-time search crawlers also have lower tolerance for slow pages and redirect chains than training crawlers do. One extra hop can be enough for a page to be dropped from a generated answer.
How do you write content an AI system can quote?
Write so that the useful fact survives being removed from its paragraph. Four patterns do most of the work: lead each section with the answer, keep each section self-contained under a question heading, date every number and name its source, and use the audience's vocabulary alongside internal programme names. AI-friendly writing is not robotic writing.
Lead with the answer
| Version | Heading | Opening text |
|---|---|---|
| Weak | Who We Are | For more than twenty years we have passionately believed in the transformative power of community... |
| Strong | What does Hope Children Foundation do? | Hope Children Foundation is a registered nonprofit in Chicago providing after-school education, mentoring and family support for children aged 6 to 14. In 2025 it served 1,247 children through programmes in 14 public schools. |
The strong version delivers entity, location, service, population, quantity and date in two sentences. It can be quoted in full and still be accurate.
Write self-contained sections
A good section makes sense when extracted: one section, one concept, one question as the heading. Example headings:
- Who is eligible for this programme?
- Where is it available?
- How many people did we serve in 2025?
- How are donations used?
- Are we a registered charity?
- How do referrals work?
Donors, beneficiaries, referrers and journalists ask exactly these questions. The questions also match the shape of the subqueries a fan-out system generates.
Date every number and name its source
| Do not write | Write instead |
|---|---|
| We serve more than 10,000 people. | In the 2025 financial year we provided food assistance to 10,482 people. |
| Most of our funding goes to programmes. | According to our 2025 audited financial statements, 82 percent of expenditure supported programme delivery. |
| We work across the region. | We deliver services in Cook County, Illinois, from six fixed locations and two mobile units. |
An undated statistic ages into an unreliable one. A dated statistic stays citable, because it describes a period, not a claim about the present.
Where do you find impact numbers when monitoring and evaluation is weak?
The numbers almost always exist already, just not in publishable form. Donor and grant reports, logframe indicator tables, attendance records, annual accounts, distribution logs and partner reports all contain counted outputs. Pull three numbers, verify each with the person who owns the record, add the period and publish. Three verified, dated facts on an Impact page will do more for AI visibility than ten new blog posts.
Most guides skip this step, and it is where nonprofits get stuck. A team writes "thousands of lives" because nobody can tell it the number, not because it loves vague prose.
| Source | What you will find there |
|---|---|
| Donor and grant reports | Counted outputs, because funders demand them |
| Logframe indicator tables | Indicators already defined, baselined and measured |
| Attendance and registration records | Sign-in sheets, enrolment lists, case files |
| Annual accounts | Expenditure by programme, beneficiary counts, staff and volunteer numbers |
| Procurement and distribution logs | Meals served, kits delivered, sessions run |
| Partner reporting | Schools, clinics and municipalities counting you in their own reports |
Use your audience's vocabulary alongside your own
A programme may be called "Community Resilience Initiative III" internally. People search for "free food parcels for families in Leeds."
Use both, in that order: plain language first, formal name second. Semantic retrieval bridges concepts well. It cannot bridge a gap that was never written down.
What schema markup should a nonprofit use?
A nonprofit should publish Organization plus NGO markup on its homepage, with legal name, identifier, address, area served and sameAs links to its register and rating profiles. Schema.org provides the NGO type and the nonprofitStatus property. Structured data does not buy a citation. It removes ambiguity, which is a quieter and more durable advantage.
A workable homepage implementation:
{
"@context": "https://schema.org",
"@type": ["Organization", "NGO"],
"@id": "https://www.example.org/#organization",
"name": "Hope Children Foundation",
"legalName": "Hope Children Foundation Inc.",
"url": "https://www.example.org/",
"logo": "https://www.example.org/logo.png",
"description": "A nonprofit providing after-school education and family support in Chicago.",
"foundingDate": "2008",
"nonprofitStatus": "https://schema.org/Nonprofit501c3",
"identifier": "12-3456789",
"address": {
"@type": "PostalAddress",
"streetAddress": "120 W Madison St",
"addressLocality": "Chicago",
"addressRegion": "IL",
"postalCode": "60602",
"addressCountry": "US"
},
"areaServed": {
"@type": "AdministrativeArea",
"name": "Cook County, Illinois"
},
"sameAs": [
"https://www.linkedin.com/company/example",
"https://projects.propublica.org/nonprofits/organizations/123456789",
"https://www.charitynavigator.org/ein/123456789"
]
}Non-US organizations use the same structure, swap nonprofitStatus for the closest applicable value or omit it, and rely on identifier plus sameAs pointing at the national register entry. A UK charity should carry its Charity Commission register page in sameAs and its charity number in identifier.
Which schema type goes on which page?
The homepage carries Organization plus NGO. Programme pages use Service, articles and news use Article, events use Event, team pages use Person, location pages use Place, and the donation page uses DonateAction. Each type has a short list of properties that must be present, shown in the table below.
| Page type | Schema type | Must include |
|---|---|---|
| Homepage | Organization + NGO |
legalName, identifier, address, areaServed, sameAs |
| Programme pages | Service |
provider, areaServed, audience, eligibility text |
| Articles and news | Article |
datePublished, author, publisher |
| Events | Event |
startDate, location, organizer |
| Team | Person |
jobTitle, worksFor |
| Locations | Place |
address, openingHours |
| Donation | DonateAction |
recipient, target |
Should a nonprofit use FAQPage markup?
Use FAQPage markup only where a page has real FAQ content, and do not expect a Google FAQ rich result from it. Google restricted FAQ rich results to authoritative government and health sites in 2023 and, according to its FAQPage documentation, stopped showing them in Google Search on 7 May 2026. FAQPage remains a valid Schema.org type.
The general principle applies to all schema: mark up what is true because it is true. Any claim that a specific markup type "increases AI citations" is, as of October 2026, an assertion without public evidence behind it.
How do you keep a nonprofit's identity consistent across the web?
Build one Entity Source of Truth, a single page of canonical facts with one owner, and audit every external profile against it. Identical sentences are not required. Identical facts are. Entity consistency is the most neglected and highest-leverage work in this guide, and it costs nothing but attention.
A search for the organization's own name may return five versions of that name.
| Where | Name as published |
|---|---|
| Website | Hope for Children Foundation |
| Statutory register | HOPE CHILDREN FOUNDATION, INC. |
| Hope Foundation USA | |
| Hope4Children | |
| Funder database | Hope Children Fdn |
A human reconciles those names instantly. A machine has to decide, on evidence, whether they are one organization or five.
What is an Entity Source of Truth?
An Entity Source of Truth is a one-page record of an organization's canonical facts: public name, legal name, abbreviation, registration number, website, headquarters, founding year, one-sentence mission, population served, service geography, executive director and the only official donation URL. One person owns the page and reviews it annually.
| Field | Canonical value |
|---|---|
| Public name | Hope Children Foundation |
| Legal name | Hope Children Foundation Inc. |
| Abbreviation | HCF |
| Registration number | 12-3456789 |
| Website | https://example.org |
| Headquarters | Chicago, Illinois, United States |
| Founded | 2008 |
| Mission, one sentence | After-school education for children aged 6 to 14 |
| Population served | Children aged 6 to 14 |
| Service geography | Cook County, Illinois |
| Executive Director | Jane Smith |
| Only official donation URL | https://example.org/donate |
Audit every external profile against the Entity Source of Truth. A founding year of 2008 on the website, 2011 in a register and 2009 on LinkedIn is not a cosmetic inconsistency. It is an instruction to a machine to treat the entity as unreliable.
Which registers and seal bodies verify a nonprofit, by country?
Verification lives in three kinds of source: the statutory register, a seal or rating body, and funder or grant databases. The specific bodies differ by jurisdiction. Most guides on this topic assume a US organization and stop at Form 990. The table below maps the sources for the United States, the United Kingdom, Ireland, German-speaking Europe, the Netherlands, France, the EU level and the Western Balkans.
| Region | Primary statutory register | Seal, rating or transparency body | Other strong sources |
|---|---|---|---|
| United States | IRS Form 990 filings | Charity Navigator, Candid | ProPublica Nonprofit Explorer, state AG registries |
| England and Wales | Charity Commission register | Fundraising Regulator | 360Giving GrantNav and UKGrantmaking (free, open), Companies House |
| Scotland | OSCR | Scottish Fundraising Adjudication Panel (Good Fundraising Guarantee) | 360Giving, Companies House |
| Ireland | Charities Regulator | Triple Lock (Charities Institute Ireland) | Revenue CHY listing |
| Switzerland | Cantonal Handelsregister; Stiftungsverzeichnis for foundations | Zewo quality seal, awarded to NPOs meeting its 21 standards | Cantonal tax-exemption listings, SwissFoundations |
| Germany | Vereinsregister; Freistellungsbescheid | DZI Spenden-Siegel; Deutscher Spendenrat transparency certificate | Initiative Transparente Zivilgesellschaft |
| Austria | Vereinsregister; Spendenbegünstigungsliste | Österreichisches Spendengütesiegel | Firmenbuch for gGmbH structures |
| Netherlands | KVK; ANBI status register | CBF Erkenning | Belastingdienst ANBI publication duty |
| France | Journal Officiel des Associations | Comité de la Charte / Don en Confiance | RNA, SIRENE |
| EU level | EU Transparency Register, recording who represents which interests at Union level and with what resources | not applicable | CORDIS and the Funding and Tenders Portal for EU-funded projects |
| Western Balkans | National NGO registers, for example APR in Serbia | rarely available | Donor project databases, public procurement portals, municipal partner listings |
Prioritize the sources in this order:
Statutory register → seal or rating body → major funder listings → LinkedIn → partner and coalition sites → media → social profiles.
For a nonprofit that has received EU, UN or large institutional funding, the CORDIS or donor project page is one of the strongest third-party entity signals it owns. Most organizations never link to it.
Fix what you control. For what you do not control, most registers and rating bodies have a correction process. The process is slow and worth it.
What should you do when AI gets your organization wrong?
Find the source of the error and correct it, because the generated answer itself cannot be edited. The correction protocol has five steps: detect the error, diagnose its source, fix the evidence, publish an explicit status statement and escalate impersonation. Visibility without accuracy is not a win. For nonprofits, accuracy is the highest-stakes part of the subject, and almost nobody writes about it.
| Failure mode | What it looks like | Who it harms |
|---|---|---|
| Stale programme | A service closed in 2023 described as current | Someone in crisis arrives at an empty building |
| Wrong eligibility | Incorrect age range, income threshold or catchment | People self-exclude, or are turned away |
| Entity blending | Merged with a similarly named organization | Your reputation, their reputation, both |
| Wrong donation route | An old campaign page or retired platform surfaced | Donors, and your income |
| Impersonation | A fraudulent donation site named in your place | Donors, directly |
| Invented specifics | A plausible number or quote you never published | Your credibility with funders and press |
A five-step correction protocol
1. Detect. Add an accuracy column to the benchmark prompt tracking described in the measurement section. Record not just whether the organization was mentioned but whether the description was correct.
2. Diagnose the source. Ask the system to cite its source, then open it. Fabrications with no source are the minority.
The error is usually real and lives on the organization's own site, an outdated PDF or an old press release.
3. Fix the evidence, not the model. The answer cannot be edited. The source can be corrected, the register entry updated, a dated correction published, and the current fact made unambiguous and easy to retrieve.
4. Publish an explicit status statement. For discontinued programmes, keep the URL live and replace the content:
"The Riverside Drop-in Centre closed in June 2024. Services previously delivered there are now available at [location]. Referrals should go to [contact]."
A status statement gives retrieval systems a current, dated fact to prefer over a cached one. Deleting the page and returning a 404 is worse, because it leaves stale third-party copies unchallenged.
5. Escalate impersonation. A fraudulent donation site appearing in AI answers about the organization is a trademark and platform-abuse issue, not an SEO issue. Report it to the platform, the payment provider and the national fraud body, and publish a clear statement naming the only official donation URL.
Set a quarterly reminder. Accuracy drift is silent, and by the time a beneficiary reports it, the answer has been wrong for months.
How can a two-person team measure AI visibility?
A two-person team can measure AI visibility in about 45 minutes a month using four layers: Google AI impressions, AI referral traffic, ten benchmark prompts run in two AI systems, and a quarterly entity accuracy check. Most AI measurement advice prescribes work nobody will do twice. The four-layer version is calibrated to what a small team can sustain.
| Layer | What it tells you | Time | Cadence |
|---|---|---|---|
| 1. Google AI impressions | Whether Google surfaces you at all | 10 min | Monthly |
| 2. AI referral traffic | Whether anyone clicks through | 5 min | Monthly |
| 3. Benchmark prompts | Whether you are named, cited and described correctly | 30 min | Monthly |
| 4. Entity accuracy | Whether the description holds up | 20 min | Quarterly |
Layer 1: Google AI impressions (10 minutes, monthly)
The Search Console generative AI performance report shows impressions, pages, countries, devices and dates for AI responses, AI Mode and AI Overviews. It does not include click data.
The missing click data matters. A nonprofit can report appearance, not performance. Do not let anyone build a board narrative that treats AI impressions as traffic.
Record monthly: total AI impressions, top five pages, trend against last month.
Layer 2: AI referral traffic (5 minutes, monthly)
OpenAI appends utm_source=chatgpt.com to links in ChatGPT, which makes those visits identifiable. In GA4, build one saved segment containing:
chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, claude.ai
Record monthly: sessions, and whether any converted.
The segment undercounts. Not every AI-mediated visit carries a parameter, and some arrive as direct traffic. Treat the figure as a trend line, not a census.
Layer 3: benchmark prompts (30 minutes, monthly)
Benchmark prompts are the layer the other three cannot replace.
Write ten prompts, not fifty. They should be the questions the organization's actual audiences ask, in their words:
- "What organizations help refugees find work in Zurich?"
- "Where can I get free counselling for teenagers in Manchester?"
- "Which charities run after-school programmes in Cook County?"
- "Is [your organization name] a legitimate charity?"
Run the same ten prompts in two systems, not five. Pick the two that matter for the audience. Use a fresh session each time.
Track one row per prompt per run:
| Date | Prompt | System | Mentioned? | Cited URL | Position in answer | Description accurate? | Named instead |
|---|---|---|---|---|---|---|---|
| Y / N | 1st, 2nd, 3rd... | Y / partly / N |
Do not change the prompts. The entire value is comparability over time. A prompt set run unchanged for four quarters is worth more than a hundred prompts run once.
Layer 4: entity accuracy (20 minutes, quarterly)
Ask each system: "Tell me about [your legal name]." Check mission, geography, eligibility, leadership, status and donation route against the Entity Source of Truth. Log every error.
What does the board report look like afterwards?
The board report moves from one traffic figure to four measures: organic sessions, Google AI impressions, AI referral sessions, and benchmark prompt results with their accuracy. A report in that form tells a board whether the organization is being named, whether the description is correct and what was fixed. A board can act on that.
Instead of:
Organic traffic is down 8 percent.
The report can say:
Organic sessions fell 8 percent. Google AI impressions rose 31 percent. AI referral sessions rose 18 percent. We were named in 6 of our 10 benchmark prompts, up from 2 last quarter, and the description was accurate in 5 of those 6. The one inaccuracy concerned our closed Riverside site and has been corrected.
What does a 30/60/90-day AI visibility plan look like?
The plan has three phases. Days 1 to 30 establish the entity and remove ambiguity. Days 31 to 60 build the evidence by turning claims into dated facts. Days 61 to 90 secure independent corroboration and measure the change against the day 1 baseline. A nonprofit does not need a GEO department for this work. It needs sequence.
Days 1 to 30: establish the entity
Objective: remove ambiguity.
| Task | Owner | Done |
|---|---|---|
| Run the 90-minute audit, record the score | Comms | |
| Fix robots.txt | Web / agency | |
| Confirm no firewall returns 403 to search crawlers | Web / agency | |
| Confirm the Search Console generative AI control includes your site | Comms | |
| Fix indexing errors and sitemap gaps | Web / agency | |
Publish Organization + NGO schema on the homepage |
Web / agency | |
| Rewrite About: legal name, number, location, mission, population, geography in the first 200 words | Comms | |
| Build the Entity Source of Truth and assign an owner | Director | |
| Claim and correct your national register entry and one seal profile | Finance / Director | |
| Write ten benchmark prompts and record the baseline | Comms |
Days 31 to 60: build the evidence
Objective: turn claims into facts.
Rework pages in this order: About → Programmes → Impact → Locations → Eligibility → Donate → Annual reports.
Use the pattern: answer first → evidence → source → date → detail.
Replace every "thousands of families transformed" with a sentence like:
"Between January and December 2025, 3,418 families received emergency food assistance through six distribution points in Cook County. Source: 2025 annual report, page 14."
Then publish three to five pages answering the real questions the organization's audiences ask, using their words in the headings.
Days 61 to 90: corroborate and measure
Objective: get independent confirmation.
- Audit every external profile against the Entity Source of Truth and correct what you control.
- Request corrections where you do not.
- Pursue legitimate mentions from funders, coalition partners, local authorities, universities, sector directories and local media. A single funder page listing your legal name, grant purpose and amount is worth more than a dozen guest posts.
- Rerun the ten benchmark prompts and compare day 1 with day 90.
Judge the result on four things, in this order: accuracy, citation, source diversity, volume. More mentions of a wrong description is a worse outcome than fewer mentions of a right one.
What is still unproven about AI visibility?
Three claims remain unproven: that llms.txt is used by any major provider for retrieval or ranking, that any "AI citation score" or guaranteed placement in AI answers exists, and that a specific word count, heading structure or "GEO format" causes citations. Crawler roles, Google's reporting limits and Google's reach figures are documented. The effects of structured data and third-party corroboration are reasonable inferences.
The field is full of confident claims and thin evidence. An honest guide should say where the line is.
| Status | Claim |
|---|---|
| Documented | Which crawlers exist, what each is for, and what blocking each one does |
| Documented | Google's generative AI reporting shows impressions, not clicks |
| Documented | Google's own reach figures for AI Overviews and AI Mode |
| Documented | Google-Extended does not affect Google Search inclusion |
| Reasonable inference | Structured data improves accurate entity resolution. Plausible and cheap, but no vendor publishes a causal claim |
| Reasonable inference | Third-party corroboration increases citation likelihood. Consistent with how retrieval is described, not independently measured |
| Unproven | llms.txt. No major provider has publicly confirmed using it for retrieval or ranking. Nearly free to publish; not a strategy |
| Unproven | Any "AI citation score" or guaranteed placement in AI answers. The systems are not deterministic and vendors do not expose mechanics |
| Unproven | That a specific word count, heading structure or "GEO format" causes citations |
| Actively changing | Crawler economics. Some CDNs now offer pay-per-crawl and bot monetization |
| Actively changing | Page-level AI controls in Google, expected under regulatory pressure by March 2027 |
| Actively changing | Whether AI referral parameters stay stable across providers |
A provider that sells AI visibility services without distinguishing between these categories has given the buyer an answer about its rigour.
Frequently asked questions
What is AI visibility for a nonprofit?
AI visibility is the ability of AI-powered search and assistant systems to correctly identify, retrieve, understand, name and cite a nonprofit when answering a relevant question. It combines technical accessibility, entity clarity, structured data, verifiable evidence and third-party corroboration. The target is a trusted entity and a usable source, not more traffic.
Is GEO replacing SEO?
No. GEO extends SEO. Retrieval systems still need discoverable pages, accessible content and trustworthy signals. What changes is the unit of competition: a nonprofit competes to be evidence inside an answer, not a link inside a list. Google itself treats GEO and AEO as vocabulary layered on top of ordinary SEO.
Can schema markup make ChatGPT cite a nonprofit?
No markup guarantees a citation. Schema markup helps machines resolve what an organization is and how its pages relate, and that is worth doing on its own merits. Treat any stronger claim as marketing: no public evidence shows that a specific markup type increases AI citations.
Should a nonprofit allow AI crawlers?
Allowing AI crawlers is a governance decision, not a technical one, and it should be made precisely. Blocking training crawlers is defensible. Blocking search crawlers removes the organization from the answers people already use to choose where to volunteer, refer and donate. Decide those two questions separately, in writing, at board level.
Does AI visibility matter for a small, local nonprofit?
AI visibility matters more for a small, local nonprofit, not less. Queries like "food banks near me," "autism support in Denver" and "refugee charities in Zurich" depend heavily on explicit organizational identity, geography and eligibility. Local nonprofits usually have the least entity noise to clean up and the most to gain from cleaning it.
Does this guide apply to nonprofits outside the United States?
All of the guide applies outside the United States except the specific databases. Substitute the national register and seal body using the register table in the entity consistency section. The underlying requirement, independent confirmation of who the organization is, is identical everywhere.
How often should a nonprofit audit its AI visibility?
Spend ten minutes monthly on measurement, run a full re-audit quarterly, and check entity accuracy every quarter. Keep the same benchmark prompts throughout, or the organization is measuring noise. A prompt set run unchanged for four quarters is worth more than a hundred prompts run once.
Where should a nonprofit start?
Start with the 90-minute audit. Find out what machines can reach, what they can verify, and where other organizations currently provide better evidence. Then fix the gaps in this order: access, entity, evidence, structure, corroboration, measurement. The nonprofits that will be visible in AI search are not the ones producing the most content. They are the ones easiest to discover, understand, verify, retrieve, quote and trust.
A technically accessible website is where the work starts, not where it ends. The organization's identity has to hold together across its own pages, structured data, national register, funders' databases, partners' sites and published results.
Think of the result as an evidence network.
| Asset | What it establishes |
|---|---|
| Your website | What the organization is |
| Your programme pages | What it does |
| Your dated impact data | What actually happened |
| Your structured data | How it all relates |
| Your external profiles | That someone else agrees |
| Your measurement | Whether any of it is being used |
The strategic question has changed from "how do we rank this page?" to "how do we become the clearest and most trustworthy source on this topic?"
Access → Entity → Evidence → Structure → Corroboration → Measurement.
Anything done out of that order is decoration.
