Captioning compliance for nonprofits: Section 504, ADA Title I and Title III, and how grant-funded organisations plan a caption programme on restricted budgets

The nonprofit compliance landscape: why three statutes can apply simultaneously

Corporate L&D teams generally have one primary captioning compliance obligation: ADA Title I (employment training for employers with 15+ employees), occasionally supplemented by state-level digital accessibility laws or sector-specific regulatory requirements. The compliance picture for nonprofits is structurally different because the nonprofit sector exists at the intersection of multiple federal funding streams, community service mandates, and employment law — each of which carries independent accessibility obligations that overlap without replacing each other.

The three statutes that most commonly apply to nonprofit training and educational video are:

• Section 504 of the Rehabilitation Act of 1973 — applies to any organisation that receives federal financial assistance. The key feature of Section 504 is that it is triggered by the funding relationship, not by the size of the organisation or the nature of its programmes. A two-person nonprofit that receives a $5,000 federal community development grant is subject to Section 504 for its entire program. A 500-person nonprofit that has never received any federal funding is not subject to Section 504 at all (though it may be subject to ADA Title I and potentially Title III).
• ADA Title I — applies to employers with 15 or more employees. For nonprofits, this covers all training provided to employees as part of the employment relationship: onboarding, compliance training, job-specific skill development, professional development, and any other training that a reasonable employee would understand to be part of their employment. The 15-employee threshold counts all employees, full-time and part-time — a nonprofit with 10 full-time staff and 8 part-time staff has 18 employees for Title I purposes.
• ADA Title III — applies to places of public accommodation. The critical question for nonprofits is whether the organisation operates a “place of public accommodation” under Title III's 12 enumerated categories. Community centres, arts and recreation facilities, educational institutions, social service establishments, and places of exercise or recreation are all on the list — and many nonprofits operate one or more of these.

The three statutes have different coverage triggers, different enforcement mechanisms, and different remedies. Section 504 is enforced through the federal agency that administers the funding (typically via OCR complaint at the Department of Education, HHS, HUD, or DOJ depending on the funding source). ADA Title I is enforced through the EEOC and through private right of action in federal court. ADA Title III is enforced by the DOJ and through private right of action in federal court, with some states providing broader remedies through state-level equivalents.

Crucially, compliance with one statute does not guarantee compliance with the others. A nonprofit that is subject to all three must satisfy all three independently, though in practice the standards overlap significantly — WCAG 2.1 AA is the technical standard that courts, OCR, and the DOJ consistently apply to video captioning quality under all three statutes. See our caption compliance matrix for a side-by-side comparison of how Section 508, ADA, and WCAG interact across different organisation types.

This guide focuses specifically on the nonprofit context. Many of the general principles covered in our guide to building a caption compliance programme apply here, but the specific budget constraints, funding structures, and organisational characteristics of the nonprofit sector create challenges that require a distinct approach.

Section 504: the federal financial assistance trigger and what it covers

Section 504 of the Rehabilitation Act reads: “No otherwise qualified individual with a disability in the United States shall, solely by reason of her or his disability, be excluded from the participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.”

Three elements of this language determine Section 504's applicability:

• “Program or activity receiving Federal financial assistance” — this is the trigger. If any program or activity of the organisation receives federal financial assistance, Section 504 applies. Critically, the 1988 Civil Rights Restoration Act expanded the scope of Section 504 to apply to the entire organisation (not just the specific program receiving funding) when the entity is an educational institution, hospital, or other type of program receiving federal assistance. Courts have broadly interpreted “program or activity” to mean the entire operations of an entity that receives federal funding.
• “Federal financial assistance” — this includes grants, loans, contracts (other than a contract of insurance or guaranty), property, technical assistance, and any other arrangement by which the federal government provides something of value. It does not include tax exemptions (so 501(c)(3) status alone does not trigger Section 504), but it includes a very broad range of funding vehicles that many nonprofits receive.
• “Qualified individual with a disability” — the protected class. For training video, the relevant population is anyone who participates in the organisation's programs, activities, or employment and has a disability that is substantially limited by hearing impairment. The law also covers people who are associated with someone with a disability, and people who are regarded as having a disability.

The practical effect of Section 504 for nonprofits is that any organisation that receives any form of federal financial assistance must ensure that its programs and activities — including training and educational video content — are accessible to people with disabilities. This is not a best-effort standard or a “reasonable steps” standard: it is a nondiscrimination standard. A video that a hearing-impaired staff member or programme participant cannot access because it lacks captions is a form of discrimination under Section 504.

The regulatory implementation of Section 504 is found in the regulations of each federal agency that distributes financial assistance. The Department of Education's Section 504 regulations (34 CFR Part 104) cover educational institutions receiving Department of Education funding. HHS Section 504 regulations (45 CFR Part 84) cover health and human services organisations receiving HHS funding. The DOJ's Section 504 regulations cover DOJ-funded activities. Each agency's regulations have been updated in recent years to explicitly incorporate web and digital accessibility standards.

OCR (the Office for Civil Rights at the Department of Education and at HHS) is the primary enforcement agency for Section 504 complaints against educational institutions and health and human services organisations. OCR complaint investigations routinely result in resolution agreements that require organisations to audit and remediate digital and video content. See our guide to state-level accessibility law for how state civil rights agencies add to these federal obligations.

Which types of federal funding trigger Section 504

The question nonprofits most commonly get wrong is the scope of what counts as federal financial assistance. Many organisations incorrectly assume that Section 504 only applies to organisations that receive direct grants from a federal agency. The actual scope is much broader.

Federal grants that directly trigger Section 504

Any direct grant from a federal agency triggers Section 504 for the recipient organisation's entire programme. Common examples:

• AmeriCorps grants (from the Corporation for National and Community Service) — one of the most common federal grants in the nonprofit sector. A nonprofit that hosts AmeriCorps members receives federal financial assistance and is subject to Section 504 for all of its programs.
• CDBG (Community Development Block Grant) — HUD-administered grants to local governments that are typically subgranted to nonprofits. A nonprofit that receives a CDBG subgrant from a city or county is receiving federal financial assistance and is subject to Section 504.
• Title I and Title II education grants — any nonprofit providing educational services under Title I of ESEA, or receiving federal special education funds under IDEA, is subject to Section 504 under the Department of Education's regulations.
• Head Start and Early Head Start grants — Head Start grantees are directly subject to Section 504 and have been required since 2016 to ensure that all programme content is accessible to participants with disabilities, including parents and family members attending programme activities.
• SAMHSA, HRSA, and other HHS grants — any nonprofit providing substance abuse services, mental health services, community health services, or HIV/AIDS services under an HHS grant programme is subject to Section 504 under HHS regulations (45 CFR Part 84).
• DOJ and OVW grants — nonprofits providing victim services, legal aid, or reentry services under Department of Justice or Office on Violence Against Women grants are subject to Section 504 under DOJ regulations.
• NSF and NIH research grants — universities and research institutions (including those with nonprofit status) that receive NSF or NIH grants are subject to Section 504 for their entire institutional program.

Indirect federal funding that triggers Section 504

The Section 504 trigger also applies to indirect federal financial assistance — funding that passes through an intermediary (a state agency or local government) before reaching the nonprofit. Courts and OCR have consistently held that subgrants of federal funding count as federal financial assistance for Section 504 purposes:

• Medicaid waiver funding — nonprofits that provide services to Medicaid-eligible individuals through state Medicaid waiver programmes are receiving federal financial assistance (Medicaid is jointly funded by federal and state governments under Title XIX of the Social Security Act). Home health agencies, disability service providers, mental health outpatient providers, and nursing facilities receiving Medicaid payments are subject to Section 504.
• State-administered federal grants — WIOA (Workforce Innovation and Opportunity Act) funds, TANF funds, Child Care and Development Block Grant funds, and many other federal programmes are administered by states and subgranted to local nonprofits. The federal origin of the funds means Section 504 applies throughout the funding chain.
• Federal contracts — a nonprofit that holds a federal contract (not a grant) — for example, a contract with a federal agency to provide training services, research services, or social services — is also covered by Section 504 through the Rehabilitation Act's contractor provision, as well as potentially through Section 503 (which applies to federal contractors with contracts over $15,000).

What does NOT trigger Section 504

Tax-exempt status under IRC Section 501(c)(3) is not federal financial assistance for Section 504 purposes. The Supreme Court held in Grove City College v. Bell (1984) and subsequent cases that tax exemptions alone do not constitute federal financial assistance. A nonprofit organisation that:

• Has never received any federal grant, contract, or subgrant
• Does not participate in any federal funding programme (including indirect pass-through funding)
• Does not receive Medicaid, Medicare, or other federal healthcare payments
• Does not have any AmeriCorps, VISTA, or other federally funded volunteer or staff placements

…would not be subject to Section 504, though it may still be subject to ADA Title I and/or Title III.

In practice, the vast majority of professionally-staffed nonprofits receive some form of federal financial assistance. The organisation that truly has no federal funding connection at all is the exception, not the rule. Any nonprofit with any question about whether Section 504 applies should conduct a funding source audit before concluding it does not.

The full compliance matrix covers how Section 504 interacts with Section 508 (which applies when the federal government itself is purchasing or developing digital content), ADA, and WCAG across different organisation and funding types.

What Section 504 requires for training video and digital content

Section 504 does not specify a technical captioning standard by name in the statute itself — the statute prohibits discrimination, and implementing regulations require that programmes be accessible. The question of what technical standard satisfies the accessibility requirement has been answered through agency guidance, resolution agreements, and court decisions.

The WCAG 2.1 AA standard in Section 504 enforcement

OCR's current position (expressed in resolution agreements and voluntary resolution guidance since at least 2018, and reinforced by DOE's 2024 digital accessibility rules) is that WCAG 2.1 AA is the appropriate technical standard for web and digital content accessibility under Section 504. For video content, the relevant WCAG 2.1 AA requirements are:

• Success Criterion 1.2.2 (Captions - Prerecorded): Captions are provided for all prerecorded audio content in synchronized media. “Synchronized media” includes all training video. The standard requires captions that are synchronised with the audio, include all spoken dialogue, include non-speech information (laughter, sound effects relevant to comprehension), and identify speaker changes. OCR resolution agreements consistently treat DCMP (Described and Captioned Media Program) quality guidelines as the benchmark for caption completeness and synchronisation.
• Success Criterion 1.2.4 (Captions - Live): Captions are provided for all live audio content in synchronized media. This covers webinars, live training sessions, and any live-streamed organisational content. CART (Communication Access Realtime Translation) is the standard accommodation for live content; see our guide to CART captioning for live training for procurement and deployment detail.

Caption accuracy under Section 504

OCR has not published a numeric accuracy threshold in regulation, but resolution agreements consistently treat 99% accuracy as the operative standard, consistent with the DCMP guidelines and the position of the National Association of the Deaf. The reason for the 99% threshold is substantive, not arbitrary: at 80–85% accuracy (the typical range for AI auto-captioning on general content), a 30-minute training video contains approximately 900–1,350 errors. A person relying on captions as their primary means of accessing the content is receiving a materially degraded version of the training. See our analysis of the 99% accuracy standard and our error rate calculator for the arithmetic.

The technical accuracy standard is where auto-captions from YouTube, Vimeo, Zoom, and most LMS native captioning engines consistently fail. LMS-native auto-captions on technical nonprofit content — grant compliance vocabulary, programme-specific terms, regulatory agency acronyms, community service jargon — typically achieve 73–87% accuracy on the first pass. That is not WCAG 2.1 AA compliant captioning. See our analysis of which auto-caption platforms meet WCAG compliance standards for a current assessment of the major platforms.

The programme-wide scope of the Section 504 obligation

A critical distinction between Section 504 and the ADA for nonprofit compliance purposes is that Section 504's nondiscrimination obligation applies to the organisation's entire program, not just to the specific activities that a person with a disability participates in. A Section 504 OCR investigation can examine any training video the organisation has produced, regardless of whether a complaint was filed about that specific video. The organisation is expected to have an accessible programme — not to reactively caption content when a complaint arrives.

This means that a nonprofit subject to Section 504 should treat forward-looking captioning as a systematic programme-wide obligation, not as an accommodation to be provided when requested. The accommodation request model — acceptable under ADA Title I in some circumstances — is insufficient for Section 504 purposes where the organisation has control over the content it produces and distributes.

ADA Title I: workforce training obligations for nonprofits with 15+ employees

ADA Title I prohibits discrimination against qualified individuals with disabilities in all aspects of employment, including “job training.” The statute applies to employers with 15 or more employees, and there is no exception for nonprofit status.

How to count employees for the 15-employee threshold

The 15-employee count under Title I is based on the number of employees the employer has for each working day in each of 20 or more calendar weeks in the current or preceding calendar year. The count includes:

• Full-time employees
• Part-time employees (counted as employees, not fractionally)
• Employees on leave (including FMLA leave)
• AmeriCorps members serving at the organisation on a stipend (some courts have treated these as employees; others have not — if in doubt, include them in the count)
• Work-study participants in some circumstances

The count does not include:

• Volunteers (unpaid), including volunteer board members
• Independent contractors
• Interns who are not paid (though unpaid interns may be covered by Section 504 if the organisation receives federal funding)

The 20-calendar-week requirement means that a nonprofit that has 15 employees for only part of the year may still meet the threshold if the 20-week criterion is satisfied in the preceding year. A nonprofit that grew from 12 to 16 employees mid-year and had 15+ employees for 20 or more calendar weeks in that year is covered by ADA Title I from that year forward.

What ADA Title I requires for training video

ADA Title I requires that employers provide “reasonable accommodations” to qualified individuals with disabilities, unless doing so would cause undue hardship. For an employee with a hearing impairment, the reasonable accommodation for a training video without captions is — at minimum — an equivalent accessible alternative. In practice, courts and the EEOC have consistently found that captioning pre-existing training videos is a reasonable accommodation that does not constitute undue hardship for employers of the size covered by Title I.

The employer can satisfy its Title I obligation reactively (by providing captions when an employee makes an accommodation request) or proactively (by captioning all training video as a matter of programme design). The reactive model is legally permissible under Title I but creates operational problems: the employee with a disability may be unable to complete the training on the same schedule as their colleagues, which itself can be a form of disparate treatment if the training delay affects advancement or retention. A proactive captioning programme avoids the accommodation request workflow entirely and treats accessible content as the default.

For nonprofits subject to both Section 504 and ADA Title I, the Section 504 programme-wide obligation (described above) effectively requires the proactive approach. An organisation that is compliant with Section 504 in its training content is also compliant with ADA Title I for the same content. The reverse is not necessarily true: an organisation that takes a reactive, accommodation-based approach to Title I may still be in violation of Section 504.

Our guide to building a caption compliance programme and the accessibility coordinator playbook cover the governance structure for managing this proactive obligation at different organisational sizes. For nonprofits, the challenge is that there is rarely a dedicated accessibility coordinator — the training manager or HR director often carries this responsibility in addition to other duties.

The undue hardship analysis for small nonprofits

ADA Title I allows an employer to claim “undue hardship” as a defence to providing a specific accommodation if the accommodation would impose significant difficulty or expense in light of the employer's financial resources, the nature of its operations, and the overall impact. The undue hardship analysis considers:

• The nature and cost of the accommodation
• The overall financial resources of the facility and the organisation
• The type of operation, including its structure, workforce, and geographic separateness
• The impact of the accommodation on the operation of the facility

For professional captioning of a single training video, courts have consistently found that undue hardship is not established for any employer covered by Title I (i.e., any employer with 15+ employees). Professional captioning costs $1.00–$3.00 per minute, or $60–$180 for a 60-minute training video. No court has found that this cost rises to “undue hardship” for an organisation with 15 or more employees.

For small nonprofits with tight budgets, the undue hardship argument is not a viable compliance strategy. The appropriate response to budget constraints is to prioritise which content gets captioned first (the triage framework described below), to use grant funding for caption costs, and to deploy the most cost-effective captioning workflow available. See our analysis of the hidden FTE cost of manual caption correction — many nonprofits are already spending more on manual caption correction than a systematic captioning programme would cost.

ADA Title III: which nonprofits are public accommodations and what it requires

ADA Title III applies to “places of public accommodation” — a category that covers a much broader range of nonprofit activities than most training teams realise. The statute lists 12 categories of public accommodations, several of which describe common nonprofit operations:

• Category 5: Educational institutions — includes nurseries, elementary/secondary/undergraduate/postgraduate private schools, and other places of education. A nonprofit that runs a continuing education programme, a community language school, a tutoring centre, or a certification training programme is an educational institution for Title III purposes.
• Category 7: Social service establishments — includes day care centres, senior citizen centres, homeless shelters, food banks, adoption agencies, and other social service establishments. A food bank that runs cooking classes, a day care centre that runs parent training sessions, or a senior centre that runs digital literacy workshops is a social service establishment offering training to the public.
• Category 9: Places of recreation — includes parks, zoos, amusement parks, and other places of recreation. Youth sports organisations and recreational programmes operated by nonprofits fall under this category.
• Category 10: Places of exercise or recreation — explicitly includes gymnasiums, health spas, bowling alleys, and golf courses. The YMCA and similar nonprofit fitness organisations are squarely within Title III.
• Category 12: Miscellaneous establishments — the catch-all category has been interpreted by courts to reach a broad range of venues and service providers that don’t fit the other 11 categories but offer services to the public.

What Title III requires for training and educational video

ADA Title III requires that goods, services, facilities, privileges, advantages, and accommodations be provided to individuals with disabilities on a full and equal basis. For video content offered as part of a programme or service to the public, this means that the video must be accessible to individuals with hearing impairments. Captions are the primary means of making video accessible for this population.

The Title III obligation for nonprofits covers:

• Volunteer orientation and training video — a nonprofit that trains volunteers through video content is offering a programme to the public. If a prospective volunteer is deaf or hard of hearing, the volunteer training programme must be accessible. This is often the most immediately practical compliance obligation for nonprofits: volunteer training video is almost always produced in-house, is almost never captioned, and is often watched by a diverse population that includes people with hearing impairments.
• Community programme participant training — a job training nonprofit that provides workforce development video to programme participants, a community health nonprofit that provides patient education video, or a youth development nonprofit that provides programme orientation video is offering educational content to the public under Title III. That content must be accessible.
• Educational programming open to the public — continuing education programmes, public workshops, community lectures, and similar programming that is open to members of the public rather than just employees or enrolled students. A nonprofit that posts recordings of its community workshops to YouTube or Vimeo is distributing public accommodation content online, which courts have consistently held falls under Title III's scope.

The DOJ's position on web accessibility under Title III, articulated in the 2024 Title II web accessibility final rule and in longstanding DOJ guidance, is that WCAG 2.1 AA is the appropriate technical standard. While the 2024 final rule technically applies to Title II (state and local governments), the DOJ has simultaneously stated in multiple contexts that it views WCAG 2.1 AA as the appropriate standard for Title III entities as well. See our ADA Title II captioning guide for the public entity context.

Title III and the “readily achievable” standard

Unlike Title I (which uses an “undue hardship” defence) and Section 504 (which imposes a nondiscrimination standard), Title III requires removal of access barriers only if “readily achievable” — meaning easily accomplishable and able to be carried out without much difficulty or expense. The “readily achievable” standard explicitly considers the financial resources of the entity and is generally understood to be a lower threshold than “undue hardship.”

In the video captioning context, courts have found that captioning a training video that is being produced or already exists in digital form is “readily achievable” for virtually any organisation that falls within Title III's scope. The cost of captioning an existing video using a professional caption vendor — $60–$180 for a 60-minute video — is not “much difficulty or expense” in any realistic sense. The readily achievable standard does not provide a practical escape route from the obligation to caption video content.

What the readily achievable standard does affect is the scope of required back-catalogue remediation. A nonprofit that has 500 uncaptioned videos from the past decade is not required under Title III to caption all 500 simultaneously; it is required to caption them as the barrier removal becomes readily achievable, prioritising the content most critical to programme access. This is the basis for the triage framework described later in this guide.

The WCAG 2.1 AA 99% accuracy standard and why auto-captions fail it

The technical standard for caption quality under Section 504, ADA Title I, and ADA Title III has converged on WCAG 2.1 AA Success Criterion 1.2.2, interpreted in light of the DCMP quality guidelines. The DCMP guidelines specify that captions must be:

• Verbatim (all spoken dialogue included)
• Accurate (consistent with what was actually said)
• Synchronised (no more than two seconds offset between audio and caption text)
• Complete (non-speech information relevant to comprehension is included)
• Properly formatted (speaker identification, sound effects, line breaks at natural phrase boundaries)

The 99% accuracy standard that OCR applies in resolution agreements, and that the National Association of the Deaf recommends, reflects the substantive requirement that captions be a functionally complete substitute for the audio track. At 99% accuracy on a 30-minute video with approximately 6,000 words, there are approximately 60 errors — the equivalent of one small mistake per 100 words. At 90% accuracy — a reasonable estimate for LMS-native auto-captions on general professional content — there are approximately 600 errors, enough to materially degrade comprehension of technical or instructional content.

Why auto-captions fail on nonprofit-specific content

The accuracy problem is worse for nonprofit content than for general corporate content because nonprofit training vocabulary includes terms that appear at low frequency in ASR model training corpora. Major ASR engines are trained primarily on general web text, news broadcasts, and widely available spoken content. Nonprofit-specific vocabulary — grant administration acronyms, regulatory agency names, programme-specific terminology, social services jargon, and community-specific terms — appears rarely in those corpora, and the phoneme sequences are unfamiliar to the decoder.

Common failure patterns in nonprofit training video:

• Grant and regulatory agency names: SAMHSA (Substance Abuse and Mental Health Services Administration) renders as “Sam-sha” or “Sam Shaw.” HRSA (Health Resources and Services Administration) renders as “Hirsa” or “Ersa.” CDBG renders as “C.D.B.G” or is simply mangled to an unrecognisable string. WIOA (Workforce Innovation and Opportunity Act) renders as “Wyoa” or “Wye-oh-ah.”
• Programme-specific terminology: Most nonprofits have proprietary programme names, initiative titles, and internal system names that have no presence in any public dataset. A case management programme called “Navigator” will transcribe correctly; a programme called “PathFinder360” or “THRIVE Connect” will fail at the word level because the decoder has never encountered those phoneme sequences associated with those word forms.
• Social services vocabulary: Terms like “wraparound services,” “trauma-informed care,” “motivational interviewing,” “strengths-based approach,” “harm reduction,” and “adverse childhood experiences” are well-established in the social services sector but appear at low frequency in general ASR training data. The phrase “adverse childhood experiences” may decode correctly on its own, but the acronym ACEs renders as “aces” (as in card game) rather than as the acronym when it appears in the rapid speech patterns of a training module.
• Legal entity names and regulatory references: For nonprofits in regulated sectors (healthcare, substance abuse treatment, child welfare), regulatory references like “Title XIX,” “42 CFR Part 2,” “45 CFR Part 164,” and “HIPAA Security Rule” may decode with errors. Section citations (e.g., “Section 1915(c) waiver”) are particularly prone to failure because the decoder handles numeric strings and parenthetical references inconsistently.
• Proper nouns for community partners, funders, and regulatory bodies: Every nonprofit organisation has a network of named community partners, government agencies, and funders whose names appear in training content. “The Kresge Foundation,” “United Way of Greater Cleveland,” “DCFS” (Department of Children and Family Services — varies by state), and similar names fail consistently when they don’t appear in the ASR model’s training corpus.

The solution is a custom glossary that encodes the canonical forms of all nonprofit-specific terms, acronyms, programme names, and regulatory references that appear in the organisation's training content. A glossary applied at the decoding stage — rather than as post-processing text substitution — raises first-pass accuracy from the 73–87% range to the 96–99%+ range for nonprofit-specific vocabulary. See our guide to glossary architecture for AI captions and our glossary maintenance workflow guide for how to build and maintain this vocabulary infrastructure.

Nonprofit vocabulary challenges: sector jargon, grant terminology, and programme-specific terms

Beyond the accuracy failures described above, nonprofit training content presents vocabulary challenges that require systematic glossary development rather than ad hoc corrections. The scope of the glossary problem varies by subsector but is significant across all nonprofit categories.

Vocabulary by nonprofit subsector

Health and human services

Healthcare-adjacent nonprofits (community health centres, hospice providers, disability service organisations, substance abuse treatment providers) use clinical and regulatory vocabulary that is covered in detail in our medical training captioning guide. The specific challenge for nonprofits in this sector is that they often provide training at a lower technical level than hospital systems — community health workers and case managers, not physicians — but still use enough technical vocabulary to create significant ASR accuracy problems. HIPAA terminology, diagnostic code categories (DSM-5, ICD-10), and substance use disorder treatment terminology (MAT, MOUD, buprenorphine, naloxone, Vivitrol) all require glossary encoding.

Workforce development

Nonprofits providing workforce development services under WIOA or similar programmes use labour market and regulatory vocabulary that is specific to the sector: “One-Stop Career Center,” “Individual Training Accounts,” “Eligible Training Providers,” “OSHA 10,” “sector-based training,” and “soft skills” vs. “hard skills” are generally well-handled by ASR. More problematic are the programme-specific terms: the name of the specific track (“Healthcare Career Pathways”), the name of the internal assessment tool (“WorkKeys”), and the names of partner employers and industry certifications.

Housing and community development

Nonprofits providing affordable housing development, community development, or housing counselling services use regulatory and financial vocabulary specific to the sector: HUD programme names (HOME Investment Partnerships, Low-Income Housing Tax Credit, Section 8 HCV, Section 811), financial terminology (LIHTC, AMI, bond financing, GP/LP structure), and compliance vocabulary (HQS inspection, HAP contract, environmental review). These terms appear at near-zero frequency in general ASR corpora and fail consistently without a glossary.

Youth development and education

Youth-serving nonprofits (after-school programmes, mentoring organisations, youth sports) use programme and pedagogical vocabulary that requires careful glossary treatment: evidence-based programme models (“Big Brothers Big Sisters Mentoring,” “Positive Youth Development,” SEL frameworks), assessment terminology (“Strengths & Difficulties Questionnaire,” “DESSA”), and regulatory vocabulary for programmes receiving federal funding (21st CCLC, TRIO programmes, Title IV-E for youth in care).

Arts and culture

Arts organisations (museums, performing arts venues, arts education nonprofits) train staff and volunteers in technical vocabulary that is highly specific: lighting and sound terminology for performing arts, conservation and curatorial vocabulary for museums, grant reporting vocabulary for arts funders (NEA terminology, state arts agency programme names). These terms appear at extremely low frequency in ASR training data and fail at high rates without a glossary.

Building the nonprofit glossary: a practical starting list

A nonprofit building its first caption glossary should prioritise these categories:

1. The organisation's own name and all programme names — including acronyms, abbreviations, and informal shortened forms used by staff in training video
2. Names of all funders referenced in training content — government agencies, private foundations, individual donors whose names are mentioned
3. Key regulatory frameworks that apply to the organisation — the statutory references, CFR citations, and regulatory body names specific to the organisation's sector
4. All partner organisation names — community partners, referral networks, subcontractors, coalitions
5. Technical vocabulary specific to the subsector — clinical terms, financial terms, legal terms, technical terms that appear in training content
6. Internal system and tool names — case management software, LMS name, financial system name, any proprietary tools mentioned in training

The glossary starts small and grows with each caption session. An organisation producing 10 hours of training video per year will typically encounter 200–400 unique vocabulary items that require glossary encoding over the first year — a manageable number that grows more slowly in subsequent years as the canonical forms are established. See our glossary maintenance workflow for how to manage the growth and update cycle.

Documenting the accommodation request process for video content

Even with a proactive captioning programme, nonprofits must have a documented process for handling accommodation requests related to video accessibility. This serves two purposes: it provides a clear pathway for individuals who encounter uncaptioned content (whether legacy content or content produced before the captioning process catches up), and it demonstrates to OCR and courts that the organisation takes its accessibility obligations seriously.

The accommodation request workflow for video content

A functional accommodation request process for video content should include:

1. A published point of contact — the organisation's accessibility coordinator (or the person fulfilling that role, even if it is not a dedicated position) should be named in the organisation's accessibility statement, on training portal pages, and in any standard disclosure language on pages where video content is hosted. An email address and a response time commitment should be published. “Email accessibility@organisation.org for accessibility assistance with any training content; we will respond within 3 business days” is an example of an appropriate published commitment.
2. A standard intake form or request method — requests for captioned versions of specific videos should be logged, including the date of request, the content requested, the requester’s role (employee, volunteer, programme participant), and the timeline for delivery. This log is part of the accommodation request record that OCR would review in a complaint investigation.
3. A defined SLA for caption delivery — the organisation should commit to a maximum timeline for delivering a captioned version of uncaptioned content upon request. 5–7 business days is a reasonable SLA for professional captioning of a standard training video. For content requested in connection with a training that is scheduled in the near term, a shorter SLA or a real-time accommodation (CART, a transcript, or a written alternative) may be required.
4. A standard accessible format offer — the accommodation offer should not be limited to captions. Depending on the content and the requester’s needs, alternatives may include a full transcript, an audio-described version, or written text equivalent. The organisation should ask the requester what format best meets their needs rather than assuming that captions are the only acceptable accommodation.
5. A process for escalation and dispute resolution — if the requester is not satisfied with the accommodation offered, there should be a clear process for escalating the request. For employees, this connects to the organisation’s standard ADA reasonable accommodation procedure. For programme participants and volunteers, this should connect to the organisation’s grievance procedure.

The accommodation request process complements but does not replace the proactive captioning obligation under Section 504. Organisations subject to Section 504 are expected to provide accessible content proactively; the accommodation request process is a safety net for cases where proactive captioning has not yet reached specific content, not a substitute for a systematic captioning programme.

For a comprehensive view of the governance structure that connects accommodation requests to the captioning workflow, see our accessibility coordinator playbook and our guide to caption programme governance policy.

Grant proposal language: funding a caption programme through federal grants

The most practical solution to the nonprofit captioning budget problem is to fund the caption programme through the federal grants that trigger the Section 504 obligation in the first place. This is not only permitted under federal grant accounting rules — it is arguably the most appropriate way to structure the cost.

Why captioning is an allowable grant cost under 2 CFR Part 200

The Uniform Guidance (2 CFR Part 200) governs all federal grant expenditures and provides the framework for determining whether a cost is allowable as a charge to a federal grant. Under the Uniform Guidance, a cost is allowable if it is:

• Necessary and reasonable for the performance of the federal award
• Allocable to the federal award
• Consistent with policies applied uniformly to both federally-funded and non-federally-funded activities
• Consistent with Generally Accepted Accounting Principles
• Not a cost that is specifically disallowed under 2 CFR Part 200 or the terms of the award

Captioning training video that is used to train staff and volunteers who work on a federally-funded programme is “necessary and reasonable” for the performance of that award when the captioned video is used to train people working on the award. It is also a Section 504 compliance cost — captioning is required by law for organisations receiving the federal funds. The direct costs of captioning video used in programme activities are allocable to the federal award.

Three ways to structure caption costs in a grant budget

Option 1: Direct programme cost

If the captioned training videos are used specifically to train staff working on the grant-funded programme, the captioning cost can be included as a direct programme cost in the budget narrative. Example language:

Video captioning services — $2,400: This organisation is subject to Section 504 of the Rehabilitation Act by virtue of its receipt of federal financial assistance. Pursuant to Section 504 and WCAG 2.1 AA (the technical standard applied by OCR), all training video used in programme activities must be closed-captioned to 99% accuracy. The project will produce approximately 20 training videos during the grant period, totalling approximately 30 hours of content. Professional captioning at $80/hour = $2,400. These costs are directly allocable to the programme activities described in Section [X] of this proposal.

Option 2: Organisation-wide compliance cost in indirect cost pool

If the organisation has an approved indirect cost rate (negotiated with the federal cognizant agency), captioning costs that benefit the organisation’s entire programme — including the federally funded programme — can be included in the indirect cost pool. Captioning of general staff training (onboarding, compliance, professional development) that benefits all staff across all programmes, including those funded by federal grants, is an appropriate indirect cost because it benefits all organisational activities proportionally. If the organisation does not yet have a negotiated indirect cost rate, it may use the de minimis rate (10% of modified total direct costs) without negotiating, but cannot include captioning as a direct cost under the de minimis rate unless it is directly allocable to specific programme activities.

Option 3: Accessibility line item in programme budget with cross-reference to Section 504

Some federal agency programme officers are unfamiliar with captioning as a budget category and may question a standalone captioning line item. The most effective approach is to include a brief compliance explanation in the budget narrative that connects the captioning cost to the Section 504 obligation that the grant itself creates:

Accessibility compliance — $3,200: [Organisation name] is subject to Section 504 of the Rehabilitation Act because it receives federal financial assistance. Section 504 and WCAG 2.1 AA require that all programme training materials, including video content, be accessible to individuals with disabilities. This line item covers professional closed captioning ($1.50/minute × 2,100 minutes of training video produced during the grant period) to satisfy this statutory obligation. These costs would not exist but for the programmatic activities funded by this award and are directly allocable to the programme budget.

Foundations and private funders

Private foundation grants are not subject to Section 504 (since Section 504 only applies to federal financial assistance), but many private funders are increasingly attentive to equity and accessibility in the programmes they fund. Including a modest caption programme line item in a private foundation proposal, with a brief explanation of the accessibility benefit for participants and volunteers with hearing impairments, is generally well-received. The cost per grant proposal is small relative to total programme budgets, the accessibility benefit is concrete and verifiable, and the argument aligns with equity values that most private funders espouse.

Some funders — particularly those in the disability rights, health equity, and civil rights spaces — have published explicit requirements that grantees provide accessible training and educational content. Reviewing the accessibility requirements in funder programme guidelines before finalising the budget is a good practice for any nonprofit managing multiple funding streams.

Restricted budgets and grant modifications

For an existing grant where captioning was not included in the original budget, a budget modification to add a captioning line item is typically approvable when the captioning is necessary for Section 504 compliance. Budget modifications for compliance costs are generally viewed favourably by programme officers because the alternative is a compliance violation that creates legal risk for the organisation and potentially for the federal agency. The request should clearly state the statutory basis for the captioning requirement and the allocation basis for charging the cost to the award. See our caption programme budget planning guide for the full cost modelling approach.

Free and reduced-cost captioning options for budget-constrained nonprofits

For nonprofits that cannot immediately fund a professional captioning programme through grants or operational budgets, there is a continuum of captioning options from free to professional that can be deployed strategically based on the content type and compliance exposure.

Free options (and their limitations)

YouTube auto-captions with manual correction

YouTube's auto-caption engine is free and, for general professional content, typically achieves 85–92% accuracy. For nonprofit-specific vocabulary (programme names, grant terminology, regulatory acronyms), accuracy drops to 73–80% on first pass. Manually correcting YouTube auto-captions in YouTube Studio is free, but the correction time is significant: correcting from 85% to 99% accuracy on a 60-minute video requires approximately 45–60 minutes of editor time. This is not “free” once staff time is accounted for — it is simply an allocation of labour cost rather than a vendor payment. See our hidden FTE cost analysis for the full accounting.

The YouTube approach is appropriate for low-compliance-exposure content (general programme updates, non-mandatory informational content) where the cost of professional captioning is genuinely prohibitive and the audience is unlikely to include individuals who rely on captions. It is not appropriate for required training content, content that has received an accessibility complaint, or content that must meet WCAG 2.1 AA compliance standards.

Zoom auto-captions for live training

Zoom's live auto-captions are free on all Zoom plans and provide real-time caption generation for live training sessions. Accuracy is similar to YouTube — generally adequate for general conversation but failing significantly on technical vocabulary. For live mandatory training with participants who rely on captions, CART (professional human real-time captioning) is the appropriate option; see our CART captioning guide. Zoom auto-captions are appropriate for informal meetings and optional training sessions where they provide a degree of accessibility even if not at WCAG 2.1 AA standard.

DCMP (Described and Captioned Media Program)

The DCMP provides free captioning and described video for educational media for K-12 students with sensory disabilities. Eligibility is limited to educational content for K-12 audiences, and the organisation must register as an approved submitter. For nonprofits that produce educational video content for K-12 students (after-school programmes, tutoring organisations, youth development nonprofits), DCMP is a legitimate free captioning resource. See our K-12 captioning guide for more on DCMP eligibility and submission requirements.

University partnerships

Some universities with disability services or communications departments offer captioning as a community service or as a training practicum for students learning captioning and CART skills. A nonprofit in a university community might explore whether such a partnership is available. The quality of student captioning varies significantly and typically does not meet WCAG 2.1 AA standards without substantial editing, but it can be a cost-effective option for non-compliance-critical content.

Reduced-cost professional captioning

Nonprofit pricing from established vendors

Several professional captioning vendors offer discounted rates for registered nonprofits. The discount is typically 10–25% off standard rates. Given that professional captioning rates range from $1.00 to $3.00 per minute of video, nonprofit rates typically fall in the $0.80–$2.25/minute range depending on the vendor and service tier. For accurate vendor comparison, see our captioning vendor pricing comparison.

Bulk and subscription pricing

Nonprofits that produce significant captioning volume (10+ hours per month) can negotiate bulk pricing or subscription arrangements that substantially reduce per-minute costs. A subscription arrangement at a fixed monthly cost for a defined volume provides budget predictability, which is particularly valuable for grant-funded organisations that need to include captioning costs in a grant budget as a defined line item. See our vendor SLA review checklist for what to review before signing a subscription contract.

AI-assisted captioning with human review at the threshold vocabulary

The most cost-effective approach for nonprofits with established glossaries is AI captioning with a custom glossary applied at the decoding stage, followed by human review focused on the specific vocabulary categories most likely to fail. If the glossary correctly handles 95% of the problematic vocabulary, the human review task shrinks dramatically — the reviewer is correcting 1–2% of words rather than 10–15%. This is the model that GlossCap's caption tool is built around: glossary-biased AI captioning that reduces the manual correction burden to the minimum required to reach WCAG 2.1 AA standard.

Back-catalogue remediation on a restricted budget: the triage framework

Most nonprofits that begin their captioning compliance work have a significant back-catalogue of uncaptioned video content. Attempting to caption everything immediately is neither financially feasible nor legally required. The appropriate strategy is a triage framework that prioritises content based on compliance risk, content value, and current access demand.

For the general principles of large-scale back-catalogue remediation, see our back-catalogue remediation playbook. The nonprofit-specific considerations are:

Tier 1: Immediate remediation required

The following content should be captioned immediately, regardless of budget constraints, because the legal exposure from leaving it uncaptioned is significant:

• Any content for which an accommodation request has been received. Once an individual has requested a captioned version of a specific video, the organisation is on notice that an inaccessible version exists and that a person who relies on captions is trying to access it. Continuing to provide an uncaptioned version after a request has been received is a clear accommodation failure. The standard SLA for responding to an accommodation request with a professional captioned version is 5–7 business days; same-day or next-day response may be required if the content is needed for an imminent training or programme activity.
• Required compliance training. Training that employees are required by law or organisational policy to complete — sexual harassment prevention, workplace safety, data privacy compliance, mandatory reporter training — must be accessible to all employees, including those with hearing impairments. Compliance training that cannot be completed by an employee due to inaccessibility is both a Section 504/Title I failure and a compliance failure (the employee cannot document completion of required training).
• Content included in active Section 504 or ADA complaints or OCR investigations. If the organisation is the subject of a complaint or investigation related to video accessibility, the specific content at issue must be remediated as a priority, and the organisation should establish that its entire forward-going captioning programme is compliant as part of the resolution.
• Onboarding content that new employees with disabilities are attempting to complete. If a new employee with a hearing impairment is in onboarding and the onboarding video content is uncaptioned, the onboarding content must be captioned immediately. This is a Title I obligation that does not wait for a systematic programme to be established.

Tier 2: Near-term remediation (within 90 days)

The following content should be captioned within 90 days of beginning the compliance programme:

• All mandatory training for current employees and active volunteers. Training that employees and volunteers are expected to complete as a condition of their role should be captioned systematically. This is the bulk of the Title I compliance obligation for most nonprofits.
• All training video for current programme participants. For nonprofits subject to Section 504, training and educational video distributed to programme participants is part of the accessible programme obligation. Participant-facing content accessed through an LMS or shared video platform should be captioned as a near-term priority.
• Content with significant ongoing access demand. Video content that receives regular views in the LMS access logs, is embedded in active training courses, or is linked from current-use learning resources has ongoing accessibility exposure. Content that has received zero views in 18 months can be addressed in a later tier.

Tier 3: Planned remediation (within 12 months)

The following content should be scheduled for captioning within the next 12-month grant cycle or budget cycle:

• Historical training content that is still actively referenced. Legacy videos that are no longer in active training programmes but are kept in the library as reference resources for staff or volunteers should be captioned in the second tier of the remediation plan.
• Community programme content distributed externally. Video content embedded on the organisation's public website, YouTube channel, or Vimeo channel should be captioned as a Title III obligation. Public-facing content represents ongoing legal exposure for nonprofits that are public accommodations.

Tier 4: Archive or delete

Content that is no longer relevant, has received zero views in 24+ months, and is not required for compliance or reference purposes should be archived or deleted rather than captioned. Captioning content that will never be accessed again is not a productive use of the remediation budget. The archive/delete decision should be documented — if a question ever arises about why specific content was removed, the organisation should be able to explain that it was legacy content removed as part of a content library hygiene process, not that it was removed to avoid captioning it. Content that has been the subject of an accommodation request should not be deleted without first providing the accommodated version.

The triage assessment process takes a few hours but should be done before committing budget to any remediation work. The LMS access logs, the training department’s content inventory, and the accommodation request log are the primary inputs to the triage assessment. Our guide to LMS caption audit methodology provides the process for conducting a complete inventory of existing captions and identifying gaps.

Building a forward-looking caption process without a dedicated staff member

Most nonprofits do not have a dedicated accessibility coordinator or disability services staff member. The captioning compliance function is typically absorbed by an L&D manager, HR director, or communications director who has many other responsibilities. Building a sustainable captioning process in this environment requires systematising the workflow so that captioning is a routine step in video production, not an afterthought or a special project.

The minimum viable captioning process

A nonprofit with limited staff capacity needs a captioning process that can be executed consistently without significant oversight. The minimum viable process for a nonprofit producing 5–15 hours of training video per month is:

1. Define the production trigger: Every training video is submitted for captioning at the same time it is submitted to the LMS for publication. Captioning is part of the video publication workflow, not a separate downstream process. The person who uploads the video is responsible for submitting it for captioning; no additional step or approval is required.
2. Maintain the glossary: The organisation's caption glossary is a shared document (in the LMS or in a shared drive folder) that is updated by anyone who encounters a new term in the captioning review process. The glossary does not require a dedicated manager — it requires a defined update process that anyone in the training workflow can execute. Once per quarter, a designated person (the L&D manager or HR director) reviews the glossary for accuracy and removes obsolete terms.
3. Set a standard review step: For all captioned content, a designated reviewer (typically the subject matter expert who produced the content, or the training manager) reviews the draft captions for the specific vocabulary in their domain before the video is published. The reviewer is not expected to review the entire transcript — only the vocabulary that is likely to have failed on first pass (programme names, regulatory terms, proper nouns). The glossary guides the review by flagging the terms most likely to fail.
4. Log all accommodation requests: A single shared spreadsheet (in Google Sheets or equivalent) tracks all incoming accommodation requests, the date received, the content requested, the response provided, and the date of resolution. The log is reviewed quarterly by the person responsible for accessibility compliance, and any pattern of requests for the same content is treated as a signal that the content should be moved up in the triage priority queue.
5. Include captioning costs in every grant proposal: As described above, captioning costs are included as a line item in every federal grant proposal where training video is a programme activity. The budget narrative includes the statutory basis for the cost (Section 504 compliance). This ensures that the captioning programme is funded sustainably without requiring annual budget battles.

This five-step process can be executed by a part-time L&D coordinator or HR director without dedicated accessibility expertise. It does not require specialised software beyond a caption vendor relationship and a shared glossary document. It is consistent with the requirements of Section 504, ADA Title I, and ADA Title III. And it is scalable — as the organisation grows and produces more content, the process scales with additional vendor volume rather than requiring additional process redesign.

For a more detailed governance framework that includes policy documentation, KPI tracking, and annual review cycles, see our caption programme governance policy template, our QA methodology guide, and our annual review process guide.

Volunteer management of the caption review function

Some nonprofits have successfully engaged volunteers to perform caption review — the manual correction step after AI captioning. This requires training the volunteer on the caption review tool and on the organisation's glossary, and establishing a quality control check by a staff member before final publication. Volunteers who are retired professionals from relevant fields (healthcare, law, finance) can be particularly effective caption reviewers for technical content in their domain, since they have the vocabulary expertise to identify and correct errors that a non-specialist might miss.

Volunteer-supported caption review should be treated as a supplement to the captioning programme, not a substitute for it. The final quality review before publication must be performed by someone with enough context to verify the accuracy of the corrections. And volunteers should not be responsible for the accommodation request workflow — that function should remain with a staff member who can commit to SLA timelines and manage escalations.

Building a caption glossary for nonprofit-specific vocabulary

The glossary is the primary technical tool for achieving WCAG 2.1 AA caption accuracy on nonprofit-specific content. A well-constructed glossary eliminates the recurring manual corrections that consume the most staff time in the caption review process and is the factor that most differentiates a sustainable captioning programme from a perpetual manual correction project.

The four tiers of nonprofit vocabulary

Tier 1: Stable organisational vocabulary (build once, maintain rarely)

This tier covers vocabulary that changes infrequently and applies across all of the organisation's content:

• The organisation's legal name, common name, abbreviation, and all programme names
• Names of the organisation's key partner organisations
• Names of key funders (government agencies, foundations, major donors) that are referenced frequently in training content
• The organisation's internal systems and tools (LMS name, case management system, donor database, HR platform)
• Key regulatory frameworks that are stable (Section 504, WCAG 2.1 AA, HIPAA, ADA)

Tier 1 vocabulary is the foundation of the glossary. It is built in the first session of glossary development and reviewed annually for changes (organisation rebrand, programme restructuring, new major funder). For most nonprofits, Tier 1 contains 50–150 entries.

Tier 2: Programme-specific vocabulary (build per-programme)

Each programme within the organisation has its own vocabulary: participant population terms, programme model terms, outcome measurement terms, and sector-specific regulatory vocabulary. When a new programme launches or when the captioning programme expands to cover a new programme area, a Tier 2 glossary expansion is required. Tier 2 additions are typically 20–50 entries per programme area.

Tier 3: Grant and compliance vocabulary (build per-funder)

When a new grant is received, the grant agreement introduces new vocabulary: the specific programme model referenced in the award (by name), the reporting metrics (by name), and the federal agency's proprietary systems and terminology (online reporting portals, data systems, compliance checklist terminology). This tier is built at grant award and maintained through the grant period. Typically 10–30 entries per new grant.

Tier 4: Content-specific vocabulary (build per-video)

Individual training videos may introduce highly specific vocabulary — a guest speaker's name, a specific regulation that is the subject of the training, a product or system being trained on for the first time. Tier 4 entries are identified during the caption review process and may be one-time additions (for a specific video) or elevated to Tier 1–3 if they turn out to be recurring vocabulary. The reviewer should note the context for each Tier 4 addition so that the quarterly glossary review can determine whether the term should be retained or removed.

The complete glossary architecture, including file formats, maintenance workflow, and integration with major caption vendors, is covered in our glossary architecture guide. The maintenance cycle and update process are covered in our glossary maintenance workflow guide.

Special considerations: multilingual nonprofit content

Many nonprofits serve linguistically diverse communities and produce training content in languages other than English, or in English with significant code-switching (mixing two languages within a single utterance). Caption quality for multilingual content presents additional challenges beyond the vocabulary failures described above:

• Spanish: The largest non-English language in US nonprofit training content. ASR accuracy for standard spoken Spanish is generally high (85–92%), but community dialect features (Spanglish, regional pronunciation, rapid speech) degrade accuracy to 70–82% on first pass. Glossary-based correction is effective for proper nouns and technical terms, but idiomatic and dialectal features require human review at a higher rate than English content.
• Code-switching content: Content that mixes English and Spanish (or another language pair) within a single utterance is the most challenging category for ASR. The decoder switches between language models mid-utterance and often fails at the transition points. Human review is required for code-switching content at a higher rate than monolingual content. A separate glossary entry set for the most common code-switched phrases helps but does not fully resolve the accuracy challenge.
• ASL interpretation in training video: Training video that includes an ASL interpreter on-screen does not thereby have a caption track. The ASL interpretation is visual, not textual, and does not create a synchronised text caption that can be ingested by an LMS or read by a screen reader user. A training video with an ASL interpreter still requires a standard closed caption track for hearing-impaired viewers who do not use ASL and for platform accessibility compliance.

See our multilingual caption workflow guide for detailed treatment of multi-language content strategies.

FAQ

Our nonprofit only receives $10,000 in federal grants per year. Are we really subject to Section 504?

Yes. Section 504 does not have a minimum federal funding threshold. Any amount of federal financial assistance — including a single small federal grant — triggers Section 504 for the organisation's entire program. The size of the federal funding does not scale the compliance obligation: an organisation that receives $10,000 in federal grants has the same Section 504 obligation to provide accessible programming as an organisation that receives $10 million. The practical difference is in the “readily achievable” analysis under Title III and the financial resources factor in the ADA undue hardship analysis — but neither of these provides an exemption from Section 504 itself.

We have only 12 employees. Are we exempt from ADA Title I?

If you have fewer than 15 employees (counting all full-time and part-time employees, but not volunteers or independent contractors), you are not covered by ADA Title I. However, if you receive any federal financial assistance, you are still subject to Section 504. And if you operate as a public accommodation (community centre, performing arts venue, educational programme, etc.), you are still subject to ADA Title III, which has no employee count threshold. The 15-employee threshold exempts only the Title I employment training obligation; the other applicable statutes remain in effect.

Our board was told that 501(c)(3) status means we are exempt from ADA requirements. Is that correct?

No. Tax-exempt status under IRC Section 501(c)(3) provides no exemption from any provision of the ADA or Section 504. The ADA's exemptions are limited to specific categories of entities (religious organisations, private clubs, small employers) that do not map to 501(c)(3) status. A large 501(c)(3) with 200 employees is subject to ADA Title I. A 501(c)(3) that operates a community centre is subject to ADA Title III. A 501(c)(3) that receives federal grants is subject to Section 504. Tax exemption and ADA coverage are entirely separate legal frameworks.

Can we use YouTube's auto-captions to satisfy our Section 504 obligation?

No. YouTube's auto-captions typically achieve 85–92% accuracy on general content and significantly less on nonprofit-specific vocabulary. WCAG 2.1 AA requires approximately 99% accuracy, consistent with the DCMP quality guidelines. YouTube provides a caption editing interface that allows manual correction of auto-generated captions — if captions are manually corrected to 99% accuracy in YouTube Studio, the corrected captions can satisfy WCAG 2.1 AA. The issue is not which platform generated the initial captions; it is whether the final published captions meet the accuracy standard. Raw auto-captions at 85–92% accuracy do not meet the standard, regardless of platform.

A volunteer just told us they have hearing loss and asked for captions on our volunteer training videos. What do we need to do?

You need to provide captioned versions of the volunteer training videos. The obligation arises under both ADA Title III (if your organisation is a public accommodation and volunteers participate in a programme open to the public) and potentially Section 504 (if the organisation receives federal financial assistance). The appropriate response is: (1) acknowledge the request within one business day, (2) commit to a delivery timeline (5–7 business days is standard for professional captioning of a pre-existing video), (3) provide the captioned version by the committed date, and (4) log the request and resolution in your accommodation request record. Going forward, all new volunteer training videos should be captioned as a matter of programme design so that future volunteers with hearing impairments are not required to make accommodation requests for new content.

We want to include caption costs in our next federal grant proposal. How much should we budget?

Professional closed captioning typically costs $1.00–$3.00 per minute of video at standard rates, with nonprofit discounts available from major vendors bringing rates to approximately $0.80–$2.25 per minute. For budgeting purposes, $1.50 per minute is a reasonable estimate for AI-assisted captioning with human review at WCAG 2.1 AA quality, which is what most professional captioning vendors provide at the mid-range price point. If you expect to produce 20 hours (1,200 minutes) of training video during the grant period, budget $1,500–$1,800 for captioning at the $1.25–$1.50/minute range. Include the Section 504 statutory basis in the budget narrative: “captioning required as a Section 504 compliance cost for organisations receiving federal financial assistance.” This framing makes the line item easy for the programme officer to approve.

We have 300 hours of uncaptioned legacy training video. Do we have to caption all of it immediately?

No. The law does not require immediate retroactive compliance for all legacy content in all cases. The obligation is to not discriminate against individuals with disabilities in the provision of programme benefits — which means that content actively used in current programmes must be accessible. The appropriate strategy is the triage framework described in this guide: prioritise required compliance training, content for which accommodation requests have been received, and content with significant ongoing access demand. Archive or delete content that is no longer relevant. Plan a phased remediation for the remainder over 12–24 months as budget permits. Document the plan and the rationale for the prioritisation. An organisation with a documented, credible remediation plan in progress is in a substantially better compliance posture than one that has taken no action at all, even if the full catalogue is not yet captioned.

How do we handle training videos that include guest speakers who don’t know our vocabulary?

Guest speakers who are unfamiliar with your organisation's specific vocabulary (programme names, funder names, internal systems) are a significant source of caption accuracy problems because the speaker uses the terms correctly but pronounces them in ways that the ASR engine has never associated with those forms. The solution is to include the guest speaker section of the video in the standard caption review process with the subject matter expert from your organisation who knows what the guest speaker was referring to. The SME review can catch and correct the errors that the guest speaker's pronunciation created. For live events with guest speakers, a post-event caption edit (adding captions to a recording) is more reliable than real-time auto-captions for this reason.