Caption compliance and accommodation request timelines: why reactive captioning structurally fails the ADA interactive process

The ADA interactive process for digital content: what the obligation actually is

The ADA’s interactive process requirement appears most explicitly in the employment accommodation context under Title I, where it describes the back-and-forth between employer and employee to identify an effective accommodation. The Title II context — which covers public entities including public universities, state agencies, and many healthcare organisations — uses slightly different language but imposes a structurally similar obligation: the covered entity must take steps to ensure that individuals with disabilities have access to programs and services, and that process must involve engagement with the individual requesting access.

For digital training video content, the interactive process plays out differently than for a physical workspace accommodation. There is no “let’s explore several options” stage in the same way — the effective accommodation for a deaf or hard-of-hearing learner who cannot access audio is synchronized text: captions. The interactive process element is primarily about documentation of engagement (did the organisation acknowledge the request, communicate what would be done, and follow through) rather than about identifying which accommodation to provide.

This is a critical distinction for L&D teams. Some compliance managers treat the interactive process as a gate: we receive the request, we engage with the learner about the options, we decide on an accommodation approach, then we produce it. In a physical accommodation context — a learner requesting a different chair, a different seating arrangement, a different delivery format for printed materials — that sequence can take weeks and still be compliant. For digital captioning, the sequence is compressed: there is no “explore options” stage because the option is captioned video; the question is only how quickly you can deliver it.

The Department of Justice’s 2024 web accessibility rule for state and local governments (Title II), which took effect in 2026 for larger public entities, codified WCAG 2.1 Level AA as the accessibility standard for digital content and services. The effective communication obligation that underlies the captioning requirement is not new — it predates the 2024 rule by decades — but the 2024 rule removed ambiguity about the standard that must be met. A public university that captions its training videos at 85% accuracy is not meeting the effective communication standard, regardless of whether a learner has requested accommodation or not. The accommodation request framework is a secondary layer on top of a primary obligation to publish accessible content.

Understanding this layered structure matters for timeline analysis. The primary obligation — caption all training video at WCAG accuracy before publication — means that an accommodation request, when it arrives, should find nothing left to do on the captioning side. The video is already captioned. The accommodation response is a delivery question, not a production question. The L&D team points the learner to the captioned version and documents the interaction. That is a same-day or next-day response, and it is the only response structure that is reliably compliant with the timeline obligation.

The alternative structure — caption on request — means that an accommodation request finds a production gap. The captioned version does not exist. The L&D team must commission captioning, wait for delivery, upload the captioned video to the LMS, and then notify the learner. Every step in that sequence takes time, and the total of those steps frequently exceeds what enforcement guidance treats as “prompt.” See how to build a caption compliance programme for L&D teams for the programme structure that eliminates this gap.

What “prompt” and “undue delay” mean in enforcement practice

The ADA does not specify a number of days for responding to accommodation requests in the digital content context. This absence of a statutory deadline is frequently misread as flexibility. It is not. The absence of a specific deadline means that courts and enforcement agencies apply a reasonableness standard — and the reasonableness standard, when applied to digital content that can be captioned, is measured in days, not weeks.

OCR (the Department of Education’s Office for Civil Rights) handles the majority of captioning complaints against educational institutions. OCR resolution agreements from the past decade consistently reflect a response-time expectation of five to ten business days for accommodation requests involving captioned digital content, with some agreements specifying tighter windows for content that is required for academic deadlines or professional licensing requirements. The ten-business-day benchmark is sometimes described in practitioner guidance as a “safe harbor” for routine accommodation responses, but it is not a statutory standard and it should not be treated as a target. It is a threshold below which enforcement agencies have generally not found unreasonable delay in the specific factual contexts of those resolution agreements.

DOJ enforcement in the employment and public accommodation contexts — which provides interpretive context for Title II even where the specific standard differs — has addressed digital accommodation delay in several consent decrees. The consistent thread is that delay caused by an organisation’s own failure to build accessible systems (rather than delay caused by a genuinely complex or ambiguous accommodation determination) is treated as a systemic compliance failure rather than an excusable process delay. An L&D team that routinely takes three weeks to produce captioned training video in response to accommodation requests is not managing a complex accommodation question; it is operating a production backlog. Enforcement agencies treat those differently.

For private-sector employers subject to ADA Title I, the EEOC’s enforcement posture on digital accommodation delay has aligned with the DOJ’s Title II posture in practice. A 2023 EEOC guidance document on remote work accessibility stated explicitly that an employer cannot use the time required to produce an accommodation as a reason to delay the accommodation indefinitely, and that for accommodations with known, bounded production timelines (like captioning), the employer is expected to have resources in place to meet the timeline or to explain why it does not.

The practical implication: if your current caption-on-request process takes two to three weeks from receipt of accommodation request to delivery of captioned video, you are operating outside the range of what enforcement guidance treats as “prompt.” Whether any specific complaint would result in a finding of unreasonable delay depends on facts that vary case by case. But the structural exposure is real, and it is the exposure that a pre-captioned programme eliminates entirely.

The undue delay standard also interacts with the nature of the content. A learner requesting captions for an optional enrichment video that has no deadline faces a different timeline calculus than a learner requesting captions for a mandatory compliance training that must be completed before a regulatory filing deadline, or a new employee requesting captions for onboarding content that determines when they can begin client-facing work. Courts and enforcement agencies have consistently found that delay is more egregious when it has material consequences for the learner: a new employee who cannot complete onboarding because captioning is pending has suffered a quantifiable harm (delayed start date, potentially delayed compensation) that a learner waiting for a supplemental resource has not. L&D programmes that serve employees in regulated industries, employees with certification or licensing requirements, and employees in roles with structured onboarding timelines face heightened exposure for captioning delays in those specific contexts.

For a full picture of the enforcement environment, see the ADA Title II enforcement reality check for 2026, which covers OCR complaint volume, resolution timelines, and the systemic vs. individual violation distinction that shapes how enforcement agencies prioritise cases.

Section 504 of the Rehabilitation Act: timeline obligations for educational institutions

Section 504 of the Rehabilitation Act applies to any institution that receives federal financial assistance, which includes all public universities, most private universities, many hospitals and healthcare systems, and a large fraction of healthcare workforce training programmes. Where Title II covers government entities as such, Section 504 creates obligations for institutions through their acceptance of federal funding. For most educational institutions, both obligations apply simultaneously, and the effective communication standard for captioning is the same under both frameworks.

The Section 504 grievance procedure requirement is particularly relevant for L&D teams. Institutions subject to Section 504 are required to have a designated 504 coordinator, a published grievance procedure for disability accommodation complaints, and a process for resolving those grievances promptly. The grievance procedure must include specific timelines for each stage of the process. Unlike the general “prompt” standard applicable to individual accommodation requests, the grievance procedure requirement creates an institutional obligation to publish and adhere to specific timelines.

For captioning accommodation requests, the Section 504 grievance timeline matters in a specific way: if a learner submits an accommodation request that is not met (because the captioned version does not exist), the learner can escalate to the grievance process. At that point, the institution’s failure to caption is no longer an informal compliance gap — it is a formal grievance that triggers OCR reporting obligations, internal review obligations, and potentially a formal findings document. The institutional record at that point documents not just the delay but the absence of a programme that would have prevented it.

OCR’s resolution process for Section 504 captioning complaints typically proceeds in two phases. In the first phase, OCR determines whether the institution has complied with the effective communication standard for the specific content at issue. In the second phase, if a violation is found, OCR requires the institution to enter a resolution agreement that specifies both the remediation of the specific content (captions the specific video) and systemic remediation (creates or strengthens a programme that prevents future violations). The systemic remediation component — the programme piece — is what L&D teams who have never received a complaint do not see, and it is the part that is most directly relevant to this post.

Resolution agreements from OCR captioning cases at universities consistently require institutions to do the same things that a well-designed proactive caption compliance programme already does: inventory all training video content, establish accuracy standards (typically 99% or better), create a captioning workflow for new content that meets the standard before publication, and establish a monitoring process to verify ongoing compliance. These are the elements of a caption compliance programme for L&D teams and the maturity stages described in the accessibility maturity model for L&D captioning. Institutions that build these elements proactively do not face them as remediation obligations after a complaint — they have them already, which is the point.

The documentation retention obligation under Section 504 is specific: records related to accommodation requests and their resolution must be retained for at least three years from the date of the activity, the date of the complaint, or the date of the OCR investigation, whichever is later. For L&D teams that have been managing captioning accommodation requests through informal channels (email, Slack messages, verbal conversations), this retention requirement is a gap in itself. The documentation section below addresses how to close it.

State law analogues: California, New York, and Texas

Federal law sets a floor for accommodation timeline obligations, but several states impose timelines and obligations that exceed the federal standard or that apply to entities not covered by federal frameworks. California, New York, and Texas collectively represent a large share of the enterprise L&D market and have distinct state accessibility frameworks that L&D compliance managers should understand.

California

California’s Unruh Civil Rights Act and the California Disabled Persons Act both apply to places of business and service providers, with the result that private employers and training providers face accommodation obligations under state law that extend beyond the ADA’s employment and public accommodation frameworks. California’s Fair Employment and Housing Act (FEHA), enforced by the California Civil Rights Department (CRD), applies the same interactive process obligation as the ADA but has been interpreted by California courts and the CRD to impose a more active engagement requirement: the employer or service provider must initiate or respond to accommodation discussions within a timeframe that reflects the urgency of the learner’s situation, not just within the employer’s administrative convenience.

California courts have found FEHA violations in cases where employers delayed accommodation responses by two to three weeks without adequate justification, in factual contexts that ADA cases might have resolved more charitably. For California employers with L&D programmes, the state framework creates an incrementally shorter effective response window than the federal floor — and the litigation risk for FEHA violations is higher than for federal ADA violations because California’s private right of action framework is more accessible to individual plaintiffs than the federal exhaustion-of-remedies process.

California also has specific requirements under Education Code sections applicable to community colleges and the California State University and University of California systems. These institutional frameworks have their own accommodation timelines, grievance procedures, and monitoring obligations that parallel but do not fully merge with the federal Section 504 structure. L&D teams at California public higher education institutions are subject to both the federal framework and the state Education Code requirements, which can impose tighter timelines in specific contexts (particularly for academic-deadline-sensitive content).

New York

New York’s Human Rights Law (NYHRL) applies to employers with four or more employees (versus the ADA’s fifteen-employee threshold) and to places of public accommodation with broad coverage. The NYHRL’s reasonable accommodation obligation has been interpreted by the New York State Division of Human Rights and by the courts as requiring a “good faith effort” to identify and provide effective accommodations, with the good faith assessment based in part on how promptly the process was initiated and concluded.

New York City has additional requirements under the New York City Human Rights Law (NYCHRL), which is explicitly interpreted more broadly than both state and federal law. The NYCHRL imposes an obligation to engage with the accommodation request “with all deliberate speed” — a phrase that New York courts have applied in ways that make two-to-three-week response times for digital content accommodation difficult to defend. The City Commission on Human Rights has issued guidance on digital accessibility accommodation that treats response timelines as a component of the good-faith engagement assessment.

For employers and training providers with operations in New York City, the practical implication is that the NYCHRL’s “all deliberate speed” standard applies a timeline pressure closer to the California FEHA standard than to the federal ADA floor — and applies it to a much broader range of employers (including small employers not covered by the ADA’s size thresholds) and service relationships (including some vendor relationships and contractual training obligations that would not trigger ADA coverage).

Texas

Texas has a more limited state accessibility framework than California or New York, but the Texas Labor Code’s discrimination provisions apply to employers with fifteen or more employees and incorporate the ADA’s interactive process obligation with similar timeline standards. More relevant for L&D teams in Texas is the Texas Accessibility Standards framework applicable to state agencies and state-funded programmes, which incorporates Section 508 requirements and has been updated to reference WCAG 2.1 AA as the applicable standard for digital content. State agencies in Texas that produce training video are subject to the Texas accessibility framework as well as federal Section 504 and Title II obligations, creating a multi-layered compliance structure with overlapping documentation requirements.

Texas also has specific requirements for higher education institutions under the Texas Education Code, including accommodation process obligations for students with disabilities that parallel the federal Section 504 structure but are enforced through the Texas Higher Education Coordinating Board. The practical effect for L&D programmes at Texas public universities is a state-level grievance obligation that runs alongside the federal OCR reporting obligation, with separate retention requirements under state public records law that may exceed the federal three-year minimum.

The key point for compliance managers in all three states: federal law sets the floor, but state and local frameworks can impose tighter timelines, broader coverage, and more accessible enforcement paths for complainants. A caption compliance programme calibrated to the federal floor may still generate state-level exposure, particularly in California and New York City. The caption compliance self-assessment checklist includes prompts to identify applicable state frameworks and adjust programme timelines accordingly.

Emergency captioning: what the turnaround actually is

Emergency captioning refers to the expedited production of captions for a specific video after a need has been identified — typically after an accommodation request, a compliance audit finding, or an internal deadline that was not anticipated when the video was originally produced without captions. Understanding what emergency captioning realistically delivers is essential for honest assessment of whether a reactive compliance programme can meet the timeline obligations described above.

Human captioning at WCAG accuracy

Human captioning at 99%+ word error rate accuracy — the standard required by WCAG 2.1 AA and referenced in DCMP guidelines — is performed by trained human captioners who listen to the audio, transcribe the content, time-stamp the caption segments, and review for accuracy before delivery. The production time for human captioning at this accuracy level is typically four to eight times the video duration for routine content (a forty-minute video takes two and a half to five hours of captioner time) plus review and formatting time. This is a throughput constraint that cannot be fully overcome by adding captioners, because each captioner works on one video at a time and the marginal cost of throughput acceleration is linear.

In practice, most professional human captioning vendors operating at WCAG accuracy quote three to five business days as their standard turnaround for new clients and two to three business days as an expedited or rush turnaround at premium pricing. Some vendors offer next-day or same-day rush services at significant premium, but these services are typically available only for short content (under fifteen minutes), require submission before an early-morning cutoff, and are subject to availability constraints that mean they cannot be reliably committed to in advance for an arbitrary video of arbitrary length.

The three-to-five business day window for human captioning does not account for the administrative time on the L&D side: identifying the accommodation request has been submitted, routing it to the person who can action it, uploading the video to the vendor’s submission system or sharing a download link, receiving the captioned file, uploading it to the LMS, and notifying the learner. Each of those administrative steps adds time, and in a typical L&D team structure where captioning is not a primary responsibility, each step may require coordination across multiple people. The total time from accommodation request submission to learner access to captioned content is routinely seven to ten business days for teams that have never had to do it under time pressure — and that is a best-case scenario assuming no vendor availability issues, no LMS upload complications, and no approval gates.

AI captioning with quality assurance

AI captioning services — including Whisper-based platforms and the ASR engines used by major video platform providers — can produce a first-pass caption file in minutes to hours, even for long-form content. The speed advantage is real and meaningful. The accuracy limitation is equally real: general-purpose AI captioning on content with technical vocabulary, multiple speakers, accented speech, or poor audio quality consistently falls below 95% accuracy on word error rate, which is below the WCAG 2.1 AA threshold of 99% and below the DCMP guidance threshold.

AI captioning with a quality assurance pass — where a human reviewer corrects the AI output against the audio, particularly for technical terms, proper nouns, and time-stamp alignment — can achieve WCAG accuracy in twenty-four to forty-eight hours for typical training video content. This assumes that the QA reviewer has access to a terminology glossary or sufficient domain knowledge to identify and correct errors on technical vocabulary, that the original audio quality is adequate for reliable AI transcription, and that the QA reviewer is available and not queue-delayed by other work.

For organisations with a glossary-aware captioning system like GlossCap, the AI accuracy on technical vocabulary is substantially higher before QA, which reduces the QA load and can compress the twenty-four-to-forty-eight-hour window. The glossary bias means that product names, SDK symbols, and institutional terminology that would normally require heavy human correction are handled correctly in the first-pass output. This is the production-efficiency argument for glossary-biased captioning beyond its learning-effectiveness benefits: it changes the math on how quickly AI-assisted captioning can reach WCAG accuracy, which matters directly when accommodation request timelines are in play.

Even at twenty-four to forty-eight hours for AI plus QA, the emergency captioning path is tight when measured against an accommodation request that arrives on a Monday for a training that the learner must complete by Wednesday. The administrative steps on both sides consume time that the production window does not have. The learner who cannot complete the training by its deadline because captioning was not available has experienced the exact harm that the ADA accommodation obligation exists to prevent — and the fact that the organisation acted in good faith by commissioning emergency captioning does not retroactively remedy the delayed access.

The emergency captioning analysis leads to the same structural conclusion that the timeline analysis leads to: the only way to guarantee compliance with the accommodation request timeline obligation is to have the captioned version ready before the request arrives. For the full argument for building a programme that achieves this, see the large-scale caption backlog remediation playbook, which addresses how to close the gap between the existing uncaptioned content library and the pre-captioned state that accommodation compliance requires.

The pre-captioned programme as the only reliable compliance path

The structural argument for a pre-captioned programme — one that captions all training video before publication and has a remediation plan for the existing uncaptioned backlog — follows directly from the timeline analysis above. When an accommodation request arrives and the captioned version already exists in the LMS, the accommodation response is a delivery and documentation task that can be completed in hours. When an accommodation request arrives and the captioned version does not exist, the response requires a production cycle that takes days — and that production cycle may not complete within the legally required window.

The pre-captioned programme also eliminates a category of risk that reactive programmes cannot eliminate: the accommodation request that arrives at a bad time. Emergency captioning depends on vendor availability. A vendor that is backlogged during a high-demand period (course completion seasons, audit deadlines, the period around a new hire cohort starting) may not be able to deliver within the normal turnaround window. The learner who submits an accommodation request when the captioning vendor is at capacity faces an extended delay that the organisation cannot control once it has committed to a reactive model.

Pre-captioned programmes also eliminate the quality-under-pressure problem. Emergency captioning, by definition, is performed under time pressure. QA steps get abbreviated, glossary review is skipped when the reviewer does not have time, and the result is captions that meet some but not all of the accuracy standard. An organisation that routinely produces emergency captions under timeline pressure is systematically producing captions below WCAG standard — which means it is producing accommodation responses that are technically non-compliant, even when the intent is compliance. The pre-captioned programme, where captions are produced in the normal workflow with adequate QA time, does not have this quality-under-pressure failure mode.

The transition from a reactive model to a pre-captioned model requires addressing two things simultaneously: establishing a new-content captioning workflow that ensures all new training video is captioned before publication, and executing a backlog remediation programme that captions the existing library of uncaptioned content. These are different challenges. The new-content workflow is a process design question; the backlog remediation is a resource and prioritisation question. For most L&D teams with a content library of more than a few hundred hours, the backlog is the larger challenge — and the accommodation risk during the remediation period is real, because accommodation requests do not wait for the remediation programme to complete.

The backlog remediation strategy should prioritise content based on accommodation risk: content that is required for role qualification, compliance certification, or new hire onboarding creates the highest accommodation exposure because delays in accessing it create the most concrete harm. Content that is optional, supplemental, or scheduled for retirement in the near term creates lower priority. A triage framework that uses these criteria to sequence the remediation backlog can significantly reduce the compliance risk exposure during the transition period, even before the remediation is complete.

The annual review process for the caption compliance programme should include a checkpoint on the new-content workflow: is every piece of training video being captioned before publication, or have exceptions accumulated that are creating a new backlog faster than the remediation programme is closing the old one? The caption programme annual review process provides a structured framework for this checkpoint and for the broader compliance posture review that the accommodation timeline obligation requires.

Interactive process documentation: what to document, when, and for how long

The documentation obligation that runs alongside the accommodation timeline obligation is the part of the interactive process that most L&D teams have not systematically addressed. The documentation requirement is not optional: it is the evidentiary basis for demonstrating that the organisation engaged with the accommodation request promptly and in good faith. In the event of an OCR complaint or EEOC charge, the documentation is what the organisation presents to show what happened and when. Organisations that cannot produce documentation of their accommodation process response face a significant evidentiary disadvantage, regardless of whether the accommodation was actually provided on time.

What to document

Every captioning accommodation request should generate a written record at each of the following points: (1) the initial request, including the date and time received, the identity of the learner (if not anonymous), the content requested, and the reason for the request if provided; (2) the organisation’s acknowledgment of the request, including the date sent and the projected response timeline communicated to the learner; (3) any interim status communications, particularly if the projected timeline cannot be met; (4) the provision of the accommodation, including the date the captioned content was made available to the learner, the format provided, and the method of notification; and (5) the outcome — whether the learner was able to access and complete the content with the accommodation provided, and any follow-up needed.

For organisations that have been managing captioning requests informally, creating this documentation structure requires a process change rather than just a record-keeping change. The intake point matters: if learners submit accommodation requests through the LMS’s built-in accommodation request function, the timestamp and content are automatically logged. If learners submit requests by email to their manager or L&D contact, the informal channel has to be converted to a formal record. The conversion step is often missed: someone responds to the email and arranges the captioning, but the original email and the response are not retained in a system that meets the three-year retention requirement.

Where to keep it

The accommodation record should be maintained in a system that (a) assigns a unique identifier to each request, (b) timestamps each stage of the response, (c) can be exported for audit or investigation purposes, (d) restricts access to appropriate personnel, and (e) supports the three-year retention period without manual intervention. Many LMS platforms have built-in accommodation request tracking functions that meet some of these requirements. Where the LMS does not, a simple shared case-management system or even a structured spreadsheet with appropriate access controls and a documented retention policy can serve the function.

The accommodation record should be stored separately from the general training record for the learner. The disability-related information in an accommodation request is sensitive medical or disability status information that should not be accessible to supervisors, colleagues, or other personnel who do not have a need to know. The separation of the accommodation record from the training record is a privacy obligation, not just a best practice.

How long to keep it

The minimum retention period under Section 504 is three years from the later of the date of the activity (the accommodation request and response), the date of any related complaint, or the date of any related OCR investigation. The ADA itself does not specify a retention period for accommodation records, but EEOC recordkeeping regulations require that employment records relevant to potential charges be retained for a minimum of one year, and records for known charges or litigation be retained until final disposition plus three years. In practice, the most conservative and broadly defensible approach is a five-year retention period for all captioning accommodation records, held from the date of final resolution of the request.

The retention requirement is worth flagging specifically for organisations that are transitioning from informal to formal accommodation tracking: there is no mechanism to retroactively create documentation for accommodation requests that were handled informally and not recorded. If an OCR complaint relates to an accommodation request from two years ago that was handled through Slack and not documented, the documentation simply does not exist. The lesson is forward-looking: implement the documentation system before the next accommodation request arrives, not after.

Eight failure modes in reactive captioning compliance

The structural argument for pre-captioned programmes is reinforced by the specific failure modes that reactive programmes encounter. These are the patterns that appear in OCR resolution agreements, in EEOC findings, and in the compliance reviews that organisations conduct after receiving a complaint. Understanding them explicitly is useful for organisations that are still operating reactively and are assessing their exposure.

Failure mode 1: Late notice

The accommodation request arrives, but the L&D team does not receive it promptly. The learner submitted a request through an HR system, a supervisor, or a building accessibility coordinator — and the request sat in an inbox or a queue before reaching the person who can action the captioning. The timeline clock started when the learner submitted the request, not when the L&D team received it. Organisations that do not have a direct, monitored intake channel for captioning accommodation requests consistently lose days to internal routing before the captioning production even begins.

Failure mode 2: No documentation of the request

The learner mentioned in a meeting, or sent a message to their manager, that they needed captions. The manager arranged the captioning informally and the issue was resolved — but no record was created. When an OCR investigation is triggered by a subsequent complaint, the organisation cannot demonstrate that it had an effective accommodation process because no record exists of the prior accommodation. The history of accommodations being provided is invisible, and the investigator sees only the complaint, not the pattern of prior compliance.

Failure mode 3: Wrong timeline assumption

The L&D team believes that “we are in the process of captioning it” satisfies the accommodation obligation as long as it is eventually completed. This assumption is incorrect for content that has a deadline. A new employee who needs captioned onboarding content to begin client-facing work has been denied access to the accommodation — regardless of whether the captions eventually appear three weeks later — if the access delay affected their employment timeline. The process being underway does not retroactively remedy the period of delay.

Failure mode 4: Vendor SLA mismatch

The organisation has a vendor contract for captioning services, and the vendor SLA specifies five business days for standard delivery. When an accommodation request arrives on a Friday for content that must be completed by Tuesday, the vendor SLA cannot meet the timeline. The organisation has a contract, but the contract does not match the compliance requirement. Emergency rush delivery is available at a premium, but it was not in the procurement plan and requires approval that takes longer than the remaining time.

Failure mode 5: LMS delivery delay

The captions are produced on time, but uploading the captioned version to the LMS, assigning it to the learner’s course path, and notifying the learner adds two to three additional business days. The production timeline was met; the delivery timeline was not. This failure mode is specific to organisations where the LMS is managed by a different team than the captioning function, and where LMS content updates require a separate workflow with its own queue and approval process.

Failure mode 6: QA skipped under timeline pressure

To meet the accommodation deadline, the QA review of the AI caption output is abbreviated or skipped. The captions are delivered on time but are below WCAG accuracy. The accommodation has been provided in form but not in substance: the learner receives captions, but the captions do not meet the effective communication standard. If the learner complains about caption accuracy, the organisation has met the form of the obligation while failing its substance — a defensible position for some types of accommodation disputes, but not for captions where the accuracy standard is codified in WCAG 2.1 AA.

Failure mode 7: No escalation path

The accommodation request arrives, the captioning process is initiated, and then something goes wrong — the vendor is unavailable, the video file is corrupted, the LMS upload fails. The L&D team does not have a documented escalation path for captioning emergencies, and the person who would normally handle it is unavailable. The request sits until the person returns or someone figures out an alternative. There is no organisational procedure that moves the request forward automatically when the normal path is blocked.

Failure mode 8: Serial accommodation requests revealing a systemic gap

Multiple learners request captions for the same training video over a period of months. Each request is handled individually, each time by commissioning emergency captioning, each time just within or just outside the timeline. The pattern is visible in retrospect but was not addressed proactively because each individual accommodation was treated as a one-off event rather than as evidence of a systemic gap. When OCR investigates a complaint related to one of these requests, the pattern of prior requests is discoverable — and the failure to address the systemic gap proactively (by simply captioning the video after the first request) is evidence of a compliance programme that prioritised cost avoidance over effective accommodation.

The caption compliance self-assessment checklist includes diagnostic questions for each of these failure modes, allowing L&D teams to identify which ones they are currently exposed to before they appear in a complaint. The caption programme annual review process provides a structured review cadence for assessing whether the programme’s controls are functioning to prevent these failures, or whether they have degraded since the last review.

FAQ

The compliance argument for a pre-captioned programme: summary

The accommodation request timeline obligation is not new, but its practical consequences for L&D teams have become more concrete as the ADA Title II enforcement environment has intensified, as OCR complaint volume at educational institutions has increased, and as more organisations have learned through enforcement actions what “prompt” actually means in practice. The argument for a pre-captioned programme — one that captions all training video before publication, with a prioritised remediation programme for the existing backlog — is stronger on the accommodation timeline grounds than on the general compliance grounds, because the timeline obligation is the one that a reactive model structurally cannot meet.

The starting point for building or evaluating a pre-captioned programme is the caption compliance self-assessment checklist, which provides diagnostic questions for each element of the programme. The accessibility maturity model for L&D captioning describes the progression from a reactive accommodation-only approach to a proactive programme that meets the accommodation timeline obligation by design rather than by emergency response. The ADA Title II and training videos guide provides the immediate-action framing for teams that are starting from zero. And the large-scale caption backlog remediation playbook addresses the backlog challenge that every team with an existing content library must work through before the pre-captioned programme is fully operational.

The accommodation request is not the end of the timeline story — it is the beginning of a documentation record that follows the organisation for at least three years. Building the programme that eliminates the production gap is the compliance move. Building the documentation process that captures every request and response is the risk management move. Both are required. Neither is optional.