Captioning new-employee onboarding training at scale: ADA Title I from day one, multi-system delivery (HRIS, LMS, intranet), and the onboarding vocabulary problem

Why onboarding video is different from other training content

The three dimensions that make onboarding captions uniquely high-stakes

Most L&D caption programmes prioritise content by audience size or compliance obligation. Under that framework, an all-hands compliance training module watched by every employee annually ranks higher than a department-specific tool training module watched by twenty people. Onboarding video breaks this prioritisation model because it scores high on every dimension simultaneously: legal exposure, audience breadth, vocabulary sensitivity, and delivery complexity.

The legal exposure dimension is distinctive. Every other category of training content involves employees who already have an established accommodation record with HR. If a veteran employee is deaf and the organisation does not have their compliance training captioned, the employee can be expected to have an accommodation on file and the HR team will have been managing that accommodation proactively. New hires are different. A new hire who is deaf may have disclosed their hearing status during the hiring process, or may have received an offer without any accommodation discussion, or may be starting with a hearing aid or cochlear implant that functions adequately in most contexts but not for video audio. The organisation cannot assume a caption-accommodation relationship exists. ADA Title I requires accessible onboarding from the first day — not after the accommodation request is processed, not after the accommodation is set up, but from day one.

The audience breadth dimension is also distinctive. Every person who joins the organisation watches the core onboarding content. Over a five-year period, every current employee has encountered the onboarding library at least once — which means every caption error in the onboarding library has been seen by the organisation's entire employee base. Caption errors in specialised training content are seen by a subset. Caption errors in onboarding are seen by everyone.

The vocabulary sensitivity dimension is the most operationally significant. Onboarding content introduces vocabulary that is uniquely dense and uniquely unfamiliar: company product names, internal acronyms, team codenames, tool names, process designations, leadership titles. A veteran employee in the finance team may have never watched the engineering onboarding module, but a new engineer watching it in their first week is encountering every one of those terms for the first time. If the caption reads "AWS Lambda function" correctly, the new hire learns the term correctly. If the caption reads "AW S lam bda function" because the ASR system split the tokens incorrectly, the new hire notes the discrepancy without necessarily knowing which version is right.

The first-impression signal

Onboarding is the organisation's first extended communication with every new employee. Caption quality in onboarding content sends a signal about the organisation's commitment to accessibility. An organisation whose onboarding captions are accurate, synchronised, and present on every video communicates that accessibility is operationalised, not merely stated in policy. An organisation whose onboarding captions have a CC button that shows no text, or captions that are a full sentence behind the audio, or captions that render the company name incorrectly, communicates the opposite — regardless of what the Employee Handbook says about the organisation's commitment to inclusion.

This is not an abstract cultural point. Research on employee onboarding experience consistently identifies the first week's communications as disproportionately influential on long-term engagement. A new hire who is deaf or hard of hearing and encounters poor captions in their first day of onboarding training — the moment when they are most dependent on the organisation to provide accessible materials — is receiving a signal about how the organisation manages accessibility in practice. The hidden cost of caption correction labour analysis estimates 1–2 hours per video-hour of caption correction work. Organisations that absorb that cost proactively, before the onboarding library is deployed, are making both a compliance and a culture investment.

Scale: onboarding video volume at different organisation sizes

The volume of onboarding video content scales with organisation size but not linearly. A 50-person organisation may have a 3–5 hour onboarding library: a company welcome video, a benefits overview, a security and compliance module, a tool walkthrough series. A 500-person organisation typically has 15–30 hours of onboarding content: role-specific onboarding tracks for engineering, sales, support, and ops; a leadership speaker series; a culture and values curriculum; a department-specific tool and process training library. A 5,000-person organisation may have 100+ hours of onboarding content across dozens of role tracks, business units, and geographic regions.

The caption burden scales with content volume. For an organisation with 30 hours of onboarding video and a 1.5-hour-per-video-hour caption correction cost, captioning the full onboarding library represents 45 hours of caption work — a week-plus of dedicated L&D time, or a significant batch job for a captioning vendor at $0.50–$3.00 per minute. The caption compliance programme guide recommends treating the onboarding library as a priority tier in the initial caption audit — it should be fully captioned before the general training library, because the legal exposure is higher and the audience is universal.

ADA Title I and the day-one caption obligation

ADA Title I vs Title II: the employment distinction

The Americans with Disabilities Act has two titles relevant to caption compliance. ADA Title I governs employment — it prohibits disability discrimination by employers with 15 or more employees in any aspect of employment, including job application procedures, hiring, advancement, discharge, compensation, job training, and all other terms, conditions, and privileges of employment. ADA Title II governs public entities — state and local government agencies and programmes. ADA Title II had a major deadline in April 2026 for web content accessibility (WCAG 2.1 AA compliance for many government entities) that has received significant attention in the L&D accessibility community. But for private employers, ADA Title I is the governing framework for employee training content, and it has been in effect — at full enforcement — since 1990.

The distinction matters for onboarding content. When an L&D team at a private-sector employer captions their training library for compliance purposes, they are operating under ADA Title I — the reasonable accommodation framework — not Title II. The Title I framework does not set a date-based requirement for web content accessibility (unlike Title II's 2026 web content rule). Instead, it establishes a duty to provide reasonable accommodation that is triggered by employee disclosure or accommodation request. The practical implication: an employer who has not proactively captioned onboarding content is not in violation until a new hire with a hearing disability requests an accommodation and the employer fails to provide one.

However, this framing is misleading as a compliance strategy for two reasons. First, the "wait for the request" approach presupposes that the request will come with enough advance notice to produce accurate captions before the new hire's onboarding begins — which is rarely true. A new hire who starts on Monday and discloses a hearing accommodation need at their offer stage may or may not give the L&D team enough time to caption the onboarding library before their first day. A new hire who starts Monday and mentions on day one that they are hard of hearing has given zero advance notice. Second, EEOC guidance and Title I case law have increasingly recognised that employers in industries and roles where video-based training is common have a duty to maintain accessible training materials proactively, not just to respond to individual accommodation requests reactively. The DOJ/EEOC pattern of enforcement actions in the 2020s, particularly against educational institutions and tech employers with video-heavy training programmes, reflects a "systemic barrier" theory: an organisation whose training library is structurally inaccessible is maintaining a systemic barrier to employment even if no individual accommodation request has been filed.

The day-one obligation in practice

What does "day-one" obligation mean in practice? Under ADA Title I, an employer may not delay accommodation once it is identified as necessary. Courts have consistently rejected employer arguments that accommodation was provided "as quickly as reasonably practicable" when the delay was caused by organisational processes that should have anticipated the accommodation need. For onboarding video captioning, the operative question is: could the organisation have reasonably anticipated that some portion of its new hires would need captioned video? The answer is always yes — every organisation hires people who are deaf or hard of hearing, the Department of Labour estimates 3.6% of the US workforce reports significant hearing difficulty, and video-based onboarding is a foreseeable context for that need to arise.

The practical standard: every piece of mandatory onboarding video content — anything a new hire is required to complete as a condition of their employment or as part of their first-week programme — should be captioned before it is assigned to any new hire. "Optional" onboarding content (supplementary culture videos, leadership speaker recordings that are nice-to-have but not required) follows a second-tier captioning priority. The accessibility coordinator playbook recommends establishing a formal "mandatory onboarding content" list in the LMS content management system, using that list as the first caption priority tier, and completing that tier before any new hire's start date.

Section 508 for federal contractors and government-adjacent organisations

Federal contractors and sub-contractors must comply with Section 508 of the Rehabilitation Act for ICT (information and communications technology) used in performance of the contract. Section 508 has been interpreted to cover employee-facing systems and training content when those systems are used in the performance of federal contracting work. For a defence contractor whose onboarding content includes role-specific training for contract employees, Section 508 compliance — which aligns to WCAG 2.1 AA for web content, per the 2017 Access Board refresh — applies to that onboarding content directly. The US caption compliance matrix covers the Section 508 applicability in detail; for onboarding content, the practical outcome is the same as ADA Title I: synchronised captions at 99%+ accuracy for all video content used in employee training.

EAA and AODA for organisations with EU and Canadian employees

Organisations with European employees operating under the European Accessibility Act (EAA, enforceable since June 2025) have a separate compliance layer for digital services used in employment contexts. The EAA coverage of employment-related digital services has been interpreted with variation across member states, but organisations operating training programmes for EU employees should treat WCAG 2.1 AA as the baseline standard for all employee-facing digital content, including onboarding video. Similarly, organisations with Canadian employees covered by the Accessibility for Ontarians with Disabilities Act (AODA) are subject to AODA's WCAG 2.0 Level AA requirements for internet and intranet content used in employment. For most organisations operating across North America and Europe, applying the 99%+ accuracy synchronised caption standard to all onboarding content is the defensible approach across every applicable legal framework.

The accommodation request trigger vs proactive captioning: cost comparison

The reactive accommodation approach — wait for a new hire to request an accommodation, then caption the onboarding library — is more expensive than the proactive approach in almost every scenario where the accommodation request comes with less than two weeks of advance notice. Captioning a 20-hour onboarding library at a batch rate of $0.75 per minute costs approximately $900. Captioning the same library on a rush basis (48–72 hours) for a new hire starting Monday typically costs 2–3x the standard rate — $1,800–$2,700 for the same content. Captioning it reactively under EEOC complaint threat, with legal fees and remediation timeline costs included, costs multiples more. The half-FTE analysis applies here: the L&D team's internal caption correction labour on a rush timeline compresses weeks of work into days, with accuracy implications. Proactive captioning is both the legally defensible approach and the economically efficient one.

Onboarding content taxonomy: what needs captions and why

Mandatory compliance content (priority tier 1)

Every new hire is required to complete a set of mandatory compliance modules as a condition of employment. These typically include: workplace harassment and discrimination prevention (required by law in California, New York, Illinois, and many other states for employees and supervisors), workplace safety orientation (OSHA-required for industries with physical hazard exposure), information security and data handling training (often required by SOC 2, ISO 27001, or HIPAA compliance programmes), and company-specific compliance content (anti-bribery and corruption, conflict of interest, code of conduct).

Mandatory compliance content is the highest-priority tier for onboarding caption work for two reasons: (1) completion is a legal requirement, which means a new hire who cannot access the content due to absence of captions may be prevented from fulfilling a legal compliance requirement, and (2) the content is typically the first thing a new hire encounters, making any caption failure an immediate first-impression event. Many organisations source mandatory compliance content from third-party vendors (ComplianceWire, Navex One, Skillsoft, OpenSesame) — the captioning obligations for purchased compliance content are separate from internally produced content and are covered in the third-party compliance training captioning analysis.

Role-specific technical training (priority tier 1)

Role-specific technical training — tool walkthroughs, product demonstrations, systems access and configuration guides — is typically the largest volume category of onboarding content for technical roles. An engineering onboarding programme may include 10–15 hours of technical setup walkthroughs, architecture overview presentations, and tool-specific training videos. A sales role onboarding may include CRM walkthroughs, product demo scripts, and competitive positioning videos. This content is high-priority because it is directly linked to the new hire's ability to perform their role — an uncaptioned technical walkthroughdelays the new hire's ability to get set up and contribute, with direct productivity consequences.

Role-specific technical training is also the category most likely to have high vocabulary density. Engineering onboarding introduces SDK names, framework identifiers, internal platform names, and repository naming conventions. Sales onboarding introduces product names, feature names, competitive product names, and pricing plan labels. Each of these is a high-failure-rate term for ASR systems without glossary context. The glossary-biased captioning approach for engineering terms is directly applicable here — the onboarding glossary for each role track should be built from the role-specific technical content script before captioning begins.

Company welcome and culture content (priority tier 2)

Company welcome videos (CEO message, founder story, values presentation), culture and mission content, team introduction videos, and employee spotlight content are a standard component of onboarding programmes. This content is not typically mandatory in the compliance sense — a new hire's employment does not depend on watching the CEO welcome video — but it is part of the organisation's structured onboarding experience and should be captioned. The priority is tier 2 rather than tier 1 because it does not create an immediate compliance failure if temporarily uncaptioned, but it should be in the caption queue within the first onboarding programme cycle.

Culture and welcome content frequently has the highest production quality in the onboarding library — it is the content most likely to have been produced by an external creative agency, with professional narration and high production value. Paradoxically, high-production-quality narration is not always easier to caption accurately. Professional narrators speak at a faster pace, often with regional accent variation, and may use a teleprompter script that was written for impact rather than for ASR-friendly sentence structure. The auto-captions accuracy analysis notes that professionally produced video is not necessarily easier to auto-caption than conversational content — the error mode shifts from vocabulary accuracy (lower on professional narration) to timing accuracy (higher on fast-paced narration).

Benefits and HR process content (priority tier 2)

Benefits overview videos, HRIS and HR portal walkthroughs, payroll and time-tracking setup videos, and expense system walkthroughs are a standard onboarding content category. These are priority tier 2 because they are practically important (a new hire who cannot access the benefits enrolment walkthrough may miss open enrolment deadlines) but not typically compliance-critical in the ADA sense for the first day. Benefits content often references specific plan names, carrier names, and portal product names that are high-failure ASR terms — "BlueCross BlueShield" (typically rendered correctly), "Fidelity NetBenefits" (split-token risk), "HSA" (acronym expansion risk). A benefits glossary pass before captioning is appropriate.

Manager and team introduction content (priority tier 3)

Many organisations produce manager introduction videos, team overview videos, and "meet your colleagues" content as part of onboarding. These are typically short-form (2–5 minutes), conversational in tone, and have lower ASR vocabulary risk than technical content. Priority tier 3 reflects that this content is supplementary and has lower compliance exposure. However, if a new hire with a hearing disability specifically relies on visual content to understand their team structure and manager communication style, uncaptioned team introduction content creates a social-access barrier that, while not strictly an ADA Title I legal failure, is an inclusion failure.

eLearning modules vs video assets: the authoring tool distinction

Onboarding programmes frequently combine raw video assets with interactive eLearning modules built in authoring tools. A compliance module may be a Storyline course with branching scenario and quiz questions. A tool walkthrough may be a Captivate software simulation. A product overview may be a Rise 360 course with embedded video and interactive timeline elements. Each of these has a different caption workflow from raw video. The eLearning authoring tool caption guide covers the Storyline, Rise, Captivate, Lectora, and iSpring workflows in detail. The key point for the onboarding programme: the authoring tool content in the onboarding library requires a separate captioning process from the video library, and that process should be part of the onboarding content audit and caption production plan.

The multi-system delivery problem: HRIS, LMS, and intranet

Why the same onboarding video appears in three systems

A common architectural reality in mid-market organisations is that new-hire onboarding content is distributed through three separate platforms, each owned by a different team, each serving a different function in the onboarding flow:

The HRIS onboarding portal handles preboarding and first-day tasks: offer acceptance, document signing, personal data entry, benefits selection, and introductory video content. New hires access this portal before their first day, sometimes weeks before — which means it is the first place a new hire encounters company video content, and the first place where caption failures become visible. HRIS platforms (Workday, BambooHR, Rippling, Greenhouse) are built around workflow and document management, not media management. Video in HRIS onboarding portals is typically delivered via embedded links to externally hosted video.

The LMS hosts the structured onboarding curriculum: courses, assessments, completion tracking, and compliance records. New hires are assigned onboarding curricula on day one or in preboarding. The LMS is where L&D owns the content, where caption tracks are typically managed, and where the caption workflow is most mature. However, an onboarding video that was embedded in the HRIS portal during preboarding and then also assigned as part of an LMS onboarding course creates two separate instances of the video with separate caption management requirements.

The company intranet hosts process documentation, team wikis, how-to guides, and culture content. Video content on the intranet is typically published by individual contributors or team leads — not by the L&D team — and is delivered via embedded links from video hosts (Loom, Vimeo, YouTube, Microsoft Stream). Intranet video is the most likely category to have no caption management process at all, because the people publishing it (engineers writing runbooks, ops leads recording process walkthroughs) are not thinking about caption compliance when they post a Loom recording in a Confluence page.

The synchronisation problem

When an onboarding video appears in all three systems and the caption text contains an error, the error must be corrected in all three systems independently. In the LMS, the correction means replacing the caption sidecar file and republishing the course. In the HRIS portal, the correction means updating the caption configuration at the video host (Vimeo, YouTube) where the video is hosted. In the intranet, the correction means finding who published the page, notifying them, and having them update the embedded video's host-level caption. These are three separate processes involving three separate teams, with no automated propagation between them.

The synchronisation problem is worst during onboarding programme updates. When an onboarding video is reshot because a product name changed, the new version must be deployed to all three systems. The LMS update is typically managed by L&D. The HRIS update requires coordination with HR. The intranet update requires finding every page where the old video was embedded — which may not be tracked — and updating each instance. The LMS migration caption checklist documents the content-tracking requirements for video assets across systems; the same tracking discipline applies to onboarding content distributed across HRIS, LMS, and intranet.

Caption ownership governance

The three-system architecture creates a caption ownership problem: who is responsible for ensuring that captions exist, are accurate, and are current in each system? The answer varies by organisation, but the most common failure mode is that each team assumes the other team is handling it. L&D assumes that any video HR puts in the HRIS portal has already been captioned by HR. HR assumes that any video the L&D team produced has been captioned by L&D. IT assumes that intranet video is handled by whoever published it. None of these assumptions is reliable.

The governance solution is a caption ownership matrix: for each category of onboarding content, a single team owns the caption production and maintenance obligation. L&D owns caption production for LMS-hosted onboarding courses and HRIS-embedded L&D-produced video. HR owns the HRIS onboarding portal caption verification (confirming that any video HR configures in the portal has a caption track at the video host). The intranet owner (typically IT or Communications) establishes the caption standard for intranet-published video and either enforces it through a publishing workflow or accepts a caption audit responsibility. The caption programme governance policy template includes a content ownership matrix template that can be adapted for the three-system onboarding architecture.

HRIS onboarding portal caption delivery

How HRIS platforms deliver video

HRIS platforms are fundamentally workflow tools. Their onboarding portals are built around task assignment, document management, and progress tracking — not media management. Video content in HRIS onboarding portals is delivered through one of three mechanisms: (1) an embedded iframe pointing to a video hosted on an external platform (Vimeo, YouTube, Wistia, Loom), (2) a hyperlink to an external video or LMS course, or (3) in a small number of enterprise HRIS implementations, a native video upload feature. The caption management implications differ for each mechanism.

For iframe-embedded video, caption quality is 100% dependent on the caption configuration at the external video host. The HRIS platform does not read, process, or modify the caption track. If the Vimeo video has a correct caption file uploaded, the Vimeo player embedded in the HRIS portal shows the CC button and correct captions. If the Vimeo video does not have a caption file, the HRIS portal shows no CC button. The HR team's role is to verify that every video they embed in the HRIS portal has been captioned at the video host before embedding — not to manage caption files in the HRIS itself.

Workday HCM onboarding portal

Workday HCM (the human resources information system) is distinct from Workday Learning (the LMS module). Workday HCM's onboarding portal delivers preboarding tasks, new-hire forms, and introductory content through a "New Hire Tasks" workflow. Video content in Workday HCM onboarding tasks is typically delivered as an embedded link or a task that opens an external URL. Workday HCM does not host video files directly — videos are embedded from Vimeo, YouTube, Wistia, or other external hosts via the embedded content action type in the task configuration.

Caption workflow for Workday HCM onboarding tasks: before configuring an onboarding task with an embedded video, confirm that the video host has a caption track uploaded and verified for the video. For Workday Learning content that is linked from an HCM onboarding task (the common integration pattern), caption management happens in Workday Learning — the LMS — and the HCM task simply links to the Learning assignment. For externally hosted video (not in Workday Learning), caption management happens at the video host (Vimeo, YouTube, Wistia).

A common misconfiguration: an organisation stores onboarding welcome videos in a SharePoint/OneDrive library and embeds them in Workday HCM onboarding tasks via the SharePoint video URL. Microsoft Stream auto-captions the video, but auto-caption accuracy for content with company-specific vocabulary is below the WCAG 99% threshold without human review and correction. The caption correction workflow for SharePoint/Stream-hosted video is covered in the intranet section below.

BambooHR

BambooHR's onboarding portal (Onboarding in BambooHR) assigns preboarding tasks to new hires. Tasks can include: send forms (electronic signature), provide information (data entry), complete an external activity (hyperlink to an LMS course or external URL), or view a file (document viewer). There is no native video upload or video hosting in BambooHR's onboarding task types. Video content in BambooHR onboarding is delivered either as a hyperlink (the new hire clicks a link that opens the video externally) or as an embedded video via a rich text block in a custom training section.

Caption workflow for BambooHR onboarding: any video linked from a BambooHR onboarding task must be captioned at the video host. BambooHR has no caption management capability. The HR team's workflow is: (1) ensure the video is hosted on a platform that supports caption tracks (Vimeo, Wistia, Loom, YouTube), (2) upload a caption file to the video host before linking the video in the BambooHR task, (3) verify caption display in the video player at the linked URL. The verification step is important because BambooHR task links open in a new browser tab — confirm that the video at the linked URL shows a CC button and functional captions before deploying the task to a new hire cohort.

Rippling

Rippling's onboarding workflows support document delivery, task assignment, and app provisioning. Video content in Rippling onboarding is delivered via app integrations (Rippling has integrations with LMS platforms including TalentLMS and Cornerstone) or via hyperlinks in custom task text. Like BambooHR, Rippling does not host video files. Caption management for video linked from Rippling onboarding tasks is the same as for BambooHR: manage captions at the video host, verify display before deploying to new hire cohorts.

Greenhouse (ATS with onboarding)

Greenhouse ATS extended its platform with an onboarding module (Greenhouse Onboarding, formerly called Greenhouse Welcome). New hires access Greenhouse's preboarding portal for document signing and introductory content before their start date. Video in Greenhouse Onboarding is delivered via Greenhouse's "Prep Sections" — custom content blocks that can include text, images, and embedded links. Like other HRIS-category platforms, Greenhouse does not host video directly. Embedded video comes from external hosts. Caption management is at the video host level.

A Greenhouse-specific note: Greenhouse Onboarding is often populated by recruiters and people operations staff who are not part of the L&D team. The video content embedded in Greenhouse preboarding materials may be produced and managed by a team that has no caption workflow. Establishing a standard that all video content added to Greenhouse Onboarding must be hosted on an approved platform (one that supports caption track upload — Vimeo, Wistia, or YouTube, not a raw MP4 upload to a shared drive) and must have a caption file uploaded before being added is the governance solution.

ADP Workforce Now

ADP Workforce Now's onboarding feature delivers documents, orientation content, and task workflows. Video content is typically delivered via embedded links. ADP does not provide native video hosting for onboarding content. For organisations using ADP for HRIS and a separate LMS for onboarding training, the ADP onboarding portal typically contains links to the LMS rather than standalone video content. Caption management is at the LMS level for LMS-linked content, and at the video host level for any standalone video linked from ADP tasks.

LMS caption delivery for onboarding content

The LMS as the primary caption management system for onboarding

For L&D-produced onboarding content, the LMS is the primary caption management system. The LMS is where structured onboarding courses live, where caption tracks are uploaded alongside video assets, and where the L&D team has administrative control over the full content library. The enterprise LMS caption audit methodology covers the full audit process; for onboarding content specifically, the LMS audit should identify every video asset in every onboarding-tagged course or learning path and verify that each has a caption track with a recent quality verification date.

Onboarding content in the LMS requires the same caption workflow as any other training content — with one additional step: the onboarding vocabulary glossary pass. Because onboarding content introduces company-specific vocabulary at high density, the caption quality gate before publishing an onboarding video course should include a spot-check of vocabulary accuracy against the organisation's term list. A video that passes a general caption quality check (98–99% accuracy on general language) may fail the onboarding-specific quality check if it has 2–3 vocabulary errors on company product names or internal acronyms that a new hire will encounter for the first time.

Workday Learning

Workday Learning is the LMS module within the Workday HCM ecosystem. It manages learning content, assignments, completions, and compliance records for Workday HCM customers. Workday Learning caption management supports caption file upload (SRT/VTT) alongside video assets in the Content section of the Learning Administration module. For SCORM content uploaded to Workday Learning (common for onboarding modules built in Storyline or Rise), caption data is embedded in the SCORM package and does not require separate Workday Learning configuration — the package delivers its own caption track.

Workday Learning integration with Workday HCM onboarding: a common configuration is to assign onboarding learning plans in Workday Learning at the point of hire in Workday HCM. The new hire's Workday HCM account triggers an automated assignment of an onboarding learning plan in Workday Learning. Caption management for this configuration is fully within Workday Learning — the L&D team manages caption tracks in the Learning Content library, and the HCM-triggered assignment delivers the content to the new hire with the L&D-managed caption track.

TalentLMS

TalentLMS supports caption file upload for video course content. In a TalentLMS video course unit, the Caption field accepts SRT file upload. The uploaded caption file is associated with the video and displayed in the TalentLMS video player for learners. For SCORM content uploaded to TalentLMS, the SCORM package carries its own caption data — a Storyline SCORM package with SRT-imported captions will display those captions when launched in TalentLMS, without requiring a separate TalentLMS caption track upload.

TalentLMS also supports video content delivered via YouTube embed. For a YouTube-embedded video in a TalentLMS course unit, caption management is at the YouTube level — the TalentLMS course unit inherits the YouTube player, which displays captions if they have been configured in YouTube Studio. This hybrid architecture (some video assets in TalentLMS's own video player with SRT upload, some as YouTube embeds with YouTube-managed captions) creates a caption management complexity: the L&D team must track which videos are managed in TalentLMS and which are managed in YouTube, and ensure that caption updates are applied in the correct system.

Cornerstone OnDemand

Cornerstone OnDemand (CSOD) is a widely used enterprise LMS and talent management platform. Cornerstone OnDemand caption management is available through the Learning Content administration interface. For video content uploaded to Cornerstone's video library, the platform supports caption track management — an SRT or VTT file can be associated with each video asset. Cornerstone also supports transcripts, which satisfy WCAG 1.2.1 for audio-only content but do not satisfy 1.2.2 for synchronised video content unless paired with a synchronised caption track.

Cornerstone's "Playlists" and "Learning Paths" features are commonly used for onboarding curricula. A new hire assigned to an onboarding learning path in Cornerstone works through a curated sequence of courses, videos, and assessments. For this workflow, the caption management responsibility is consistent: the L&D or platform admin team manages caption tracks for each video asset in the Cornerstone video library, and every video asset in an onboarding learning path must have a caption track before the path is activated.

Cornerstone LMS SCORM delivery: for SCORM content in Cornerstone, the SCORM package carries its own caption data. The Cornerstone SCORM player renders the course content in an iframe — if the SCORM package was built in Storyline with correct caption configuration, those captions display in the Cornerstone SCORM player. However, Cornerstone's SCORM player has historically had JavaScript execution restrictions that can interfere with some Storyline caption rendering implementations. Testing SCORM onboarding content in the Cornerstone player (not just in Articulate Review 360) before deployment is essential.

Docebo

Docebo supports caption file upload for video content in the Content Library. In Docebo's media player, an uploaded SRT or VTT file provides the caption track. Docebo also supports integration with external video platforms (Vimeo, YouTube, Wistia) via its E-Content integration system. For integrated external video content, caption management is at the video host level.

Docebo's onboarding-specific feature set includes Learning Plans (structured onboarding curricula) and automatic course assignment based on user attributes (department, role, hire date). The hire-date-based assignment is particularly relevant for onboarding: Docebo can automatically assign onboarding courses to users who joined within the last 30 days, for example. This automation means that any uncaptioned video in the onboarding-assigned courses will be automatically delivered to new hires — the auto-assignment amplifies the caption compliance exposure if the onboarding library is not fully captioned before the automation runs.

Kaltura

Kaltura is an enterprise video platform used by many organisations as both a media library and a video delivery layer within LMS integrations. Kaltura MediaSpace supports caption track management through the Captions and Enrich feature in the media management interface. Caption tracks can be uploaded (SRT, DFXP/TTML), generated via Kaltura's built-in ASR service (Kaltura REACH), or imported from a third-party captioning service integration.

Kaltura's integration with LMS platforms (including Moodle, Canvas, Blackboard, and others via Kaltura LTI) means that caption management in Kaltura MediaSpace is reflected in the LMS player when content is delivered via the Kaltura LTI integration. For an organisation using Kaltura as the video library and an LMS (Cornerstone, TalentLMS, etc.) as the course delivery platform, caption management at the Kaltura level propagates automatically to the LMS delivery — unlike the three-system synchronisation problem described earlier, this is an architecture where a single caption correction at the Kaltura level updates the caption track everywhere the video is delivered via the Kaltura integration.

Kaltura REACH (the built-in ASR captioning service) offers machine-generated captions that require human review before use in onboarding content. Kaltura REACH accuracy varies by content type and vocabulary density. For onboarding content with high company-specific vocabulary, REACH-generated captions should be treated as a starting draft, not a deployable caption track. The vocabulary correction step — applying the onboarding-specific glossary to the REACH output before quality review — significantly reduces the review burden.

Company intranet video caption delivery

The intranet as an onboarding delivery surface

Most organisations direct new hires to the company intranet during their first week as a source for process documentation, team information, and contextual knowledge. The intranet contains a significant volume of video content — recorded team presentations, process walkthrough videos, tool how-to recordings, leadership messages — that is produced by individual contributors and teams outside of the L&D and HR functions. This content is frequently not tracked in the LMS, not included in caption audits, and not managed by anyone with a formal caption responsibility.

The scale of uncaptioned intranet video is typically much larger than organisations estimate. A Confluence or SharePoint intranet at a 200-person organisation may have hundreds of Loom or Vimeo videos embedded in team wiki pages, how-to documentation, and project-specific content. Most of these were recorded without captions, embedded without caption consideration, and have never been reviewed for accessibility. When a new hire with a hearing disability is directed to the intranet as part of their onboarding, they encounter a large volume of uncaptioned video content that is de facto part of their onboarding experience even if it is not part of the formal onboarding programme.

SharePoint / Microsoft Stream

Microsoft 365 organisations use SharePoint and Microsoft Stream for intranet video content. Microsoft Stream (the new SharePoint-integrated Stream, as of 2023) automatically generates captions for uploaded video content using Azure Cognitive Services ASR. Auto-generated captions are available for new hires as an opt-in CC track. The accuracy of these auto-generated captions for content with company-specific vocabulary is comparable to other ASR systems: 75–88% on technical vocabulary without glossary context, well below the WCAG 99% threshold.

SharePoint/Stream provides a caption editing interface: in the Stream portal, video owners can edit the auto-generated caption transcript directly in the browser. The edited, corrected caption file can then be downloaded as an SRT or saved to the video. For onboarding-relevant SharePoint videos, the caption editing workflow is: (1) identify the video in SharePoint/Stream, (2) access the auto-generated transcript via the video's transcript panel, (3) review and correct vocabulary errors — especially company product names, tool names, and internal acronyms, (4) save the corrected transcript. The correction is reflected immediately in the Stream video player for any subsequent viewer.

SharePoint organisational video governance: for large organisations, governing caption quality on SharePoint intranet video requires a policy that requires video publishers to review and correct auto-generated captions before publishing, rather than relying on the auto-generated draft. This is enforced through a publishing workflow (a SharePoint site policy or a communication from IT/Communications requiring caption review) rather than through automatic technical enforcement, since SharePoint does not block video publication with uncorrected captions by default.

Confluence (Atlassian)

Confluence is the most widely used team wiki and intranet platform in technology organisations. Video in Confluence is delivered via embedded media — typically Loom (for screen recording content), Vimeo, YouTube, or a Loom/Wistia embed. Confluence does not host video files directly and does not have native caption management. Caption quality for Confluence-embedded video is entirely dependent on the caption configuration at the video host.

Loom is the dominant video source for intranet video in technology organisations — it is the tool engineers, product managers, and operations staff use to record quick walkthrough videos and embed them in Confluence documentation. Loom generates auto-captions for all videos and provides a caption editing interface in the Loom editor. Loom's auto-caption accuracy for technical vocabulary is in the 75–85% range. Loom provides a caption correction UI in the Loom editor that allows word-by-word correction. For onboarding-relevant Loom videos embedded in Confluence, the caption correction process is: (1) access the Loom video from the embed URL, (2) edit captions in the Loom editor (correction is word-level, not full-text replacement), (3) save the corrected captions. The correction is immediately reflected in any Confluence page where the Loom video is embedded — this is an architectural advantage of the video-host model over the SCORM model.

The practical governance challenge for Confluence: engineers who record Loom walkthroughs and embed them in Confluence runbooks are not thinking about caption quality. A policy that requires caption review for Confluence pages tagged as "onboarding" or "new hire" content provides a scoped, manageable governance boundary — rather than requiring caption review for all Confluence content (which is operationally impractical), it focuses caption quality requirements on the content that new hires are most likely to encounter.

Notion

Notion is used as a company wiki and onboarding portal by many early-stage and growth-stage organisations. Video in Notion is delivered via embedded blocks (Loom, Vimeo, YouTube, or generic video embeds). Like Confluence, Notion does not host video files and has no native caption management. Caption quality for Notion-embedded video is dependent on the video host configuration. The governance approach is the same as for Confluence: require caption-reviewed video for any Notion pages designated as part of the new hire onboarding experience.

Guru and similar knowledge base platforms

Guru (the knowledge management platform) is used by many sales and support organisations for onboarding documentation. Guru cards can embed video from external hosts. The caption governance approach is the same as for Confluence and Notion: manage captions at the video host level. Guru's onboarding-specific use case — "new hire Guru cards" that are assigned in onboarding workflows — makes the caption governance question particularly relevant, because Guru cards assigned as part of formal onboarding are analogous to LMS content in their compliance exposure even if they are not hosted in the LMS.

The onboarding vocabulary problem

Why new hires are the worst audience for ASR vocabulary errors

Caption accuracy problems from ASR vocabulary errors are a universal challenge for training content. An experienced employee who encounters an ASR error on a company product name has the vocabulary context to recognise the error and mentally substitute the correct term. They know "Gloss Cap" is a misrecognition of "GlossCap." They know "AWS S3 bucket" is correct even if the caption reads "A W S S 3." They have years of context to apply a correction.

A new hire in their first week has none of this context. They are forming their vocabulary model of the organisation's products, tools, and processes in real time, from the content they are consuming. If the caption reads "Gloss Cap" instead of "GlossCap," the new hire may learn "Gloss Cap" as the correct form. If the caption reads "the GLC portal" when the product is "the GlossCap Console," the new hire may spend weeks using the wrong name before a colleague corrects them. If the caption reads "our JIRA project H-ZAP" when the project is named "HZAP," the new hire learns the wrong designation for a tool they are about to use daily.

The downstream consequences of vocabulary errors in onboarding captions are not trivial. An employee who learns incorrect product names during onboarding will use those names in customer conversations, support tickets, and documentation — propagating the error beyond the individual. An employee who learns incorrect process names will look confused in team meetings when the correct name is used, creating a social friction that a new hire is especially poorly positioned to correct without appearing uninformed. The onboarding vocabulary problem is a training-effectiveness problem with ongoing organisational consequences, not just a compliance nicety.

The vocabulary categories unique to onboarding content

Four categories of vocabulary are uniquely dense in onboarding content compared to other training content:

Product names and version identifiers. Every organisation has product names that combine common words in uncommon ways, use non-standard capitalisation, or abbreviate to acronyms that do not match the expected expansion. "GlossCap" is not a natural language token. "HubSpot" compounds two words without a space. "GitLab" versus "GitHub" is a single-letter distinction that ASR systems often swap. "Kubernetes" (frequently rendered "koober netties" or "cube er net ease" by ASR systems with insufficient domain exposure). Onboarding content introduces the full product taxonomy in the first week — this is when vocabulary accuracy matters most and when the new hire has the least ability to detect errors.

Internal acronyms and project codenames. Every organisation has internal acronyms that have no natural-language precedent: "HZAP" (a JIRA project), "TLDR" (used literally in meeting references), "GTM" (Go-To-Market, not a mapping tool). Project codenames are frequently proper nouns with arbitrary letter combinations: "Project Falcon," "Operation Mercury," "Sprint Cobalt." ASR systems encountering these for the first time produce unpredictable outputs. The correct approach is to build the onboarding glossary from the script content — every term that is used in onboarding narration and is not a standard English word should be in the glossary.

Tool and platform names. The onboarding week introduces new hires to the organisation's full software stack: CRM, HRIS, LMS, project management, engineering tools, design tools, communication tools. Many of these have names that are high-failure ASR terms: "Salesforce" (typically rendered correctly), "Asana" (rendered as "a sana" or "as ana" with incorrect token boundaries), "Figma" (rendered as "fig ma" or "fig mah"), "Okta" (correctly rendered but often split as "OK ta"), "Notion" (correctly rendered — standard English word), "Zapier" (rendered as "zap ee er" or "zay pee er" with uncertainty on the penultimate vowel).

Internal process names. Organisations have named processes that do not appear in any external reference: "The Monday Scrub" (a weekly review process), "the QBR deck" (a quarterly business review document — rendered as "the Q B R deck" with possible space insertion), "the GSD framework" (rendered as "the G S D framework" or "the gist framework" depending on context). These process names are the hardest to correct in ASR output because they have no standard form outside the organisation — there is no web reference for the ASR system to align to, and the glossary is the only mechanism for ensuring correct rendering.

Building the onboarding glossary

The organisational caption glossary described in the customer glossary architecture guide is a starting point for onboarding content. But onboarding content requires a supplementary onboarding-specific glossary pass for three reasons: (1) onboarding content introduces vocabulary that may not yet be in the general glossary (new product names not yet in external documentation), (2) the vocabulary density is higher in onboarding content than in any other training category, and (3) the consequence of vocabulary errors is highest in onboarding content.

The onboarding glossary build process:

Step 1: Extract the onboarding script corpus. Gather the narration scripts (or slide notes) for all onboarding video content. For content that was recorded without a script, transcribe the narration (using the ASR output as a starting draft) and identify every term that looks potentially problematic — non-standard capitalisation, compound words, abbreviations, acronyms, proper nouns.

Step 2: Cross-reference with the company term list. The company's internal term list (often maintained in a brand guidelines document, a style guide, or a Confluence/Notion page) contains the authoritative versions of product names, brand terms, and standard capitalisation. Cross-reference the onboarding script corpus against this term list. Any term in the onboarding scripts that appears in the term list goes directly into the onboarding glossary with its authoritative form.

Step 3: Identify role-specific technical vocabulary. For role-specific onboarding tracks (engineering onboarding, sales onboarding, etc.), the technical vocabulary in that track may not be in the company-wide term list. Engineering onboarding scripts contain framework names, CLI commands, and repository naming conventions that the marketing team's brand guidelines will not include. Build role-specific glossary extensions by reviewing the scripts with a domain expert from that function — a senior engineer for the engineering onboarding track, a senior sales rep for the sales onboarding track.

Step 4: Apply the glossary to the captioning workflow. The glossary-biased decoding approach applies the glossary at the ASR stage — terms in the glossary are weighted to appear correctly in the transcript before timing is applied. For caption workflows that run ASR first and apply glossary correction as a post-processing step, the glossary correction should be applied before timing is finalised, not after, to allow for sentence-level re-timing if a multi-word correction changes the token count significantly.

Step 5: Maintain the glossary as onboarding content updates. Onboarding content is updated whenever products, processes, or tools change — which is frequently in growth-stage organisations. The onboarding glossary must be updated in sync with onboarding content updates. A product name change that is reflected in a reshot onboarding video must also be reflected in the glossary before the updated video is captioned. Building the glossary update into the onboarding content update workflow — a checklist item in the "update onboarding video" process — is the reliable mechanism for keeping glossary and content in sync.

The first-cohort feedback loop

After the onboarding library is captioned and glossary-corrected, the first cohort of new hires who go through the onboarding programme is the best source of residual vocabulary error identification. New hires are reading captions carefully (they are learning the vocabulary) and they are primed to notice when the caption text does not match what they expected. Establishing a simple caption error reporting path for new hires — mentioned in the onboarding welcome communication, with a single-step reporting mechanism (an email alias, a Slack message to the L&D team) — converts the first cohort from a passive audience into an active QA contributor.

The feedback should feed into the caption QA workflow described in the caption QA methodology guide. Caption errors reported by new hires are triaged by type (vocabulary error, timing error, missing caption, synchronisation failure) and severity (affects meaning, affects vocabulary learning, cosmetic), corrected in the caption file, and propagated to all delivery systems. The glossary is updated with any vocabulary error that reflects a gap in the glossary coverage. Over the first 2–3 onboarding cohorts, the glossary correction rate drops toward zero as residual errors are eliminated — this is the compounding accuracy improvement described in the glossary architecture guide.

Production workflow at scale

Centralised vs decentralised onboarding production

Onboarding video is produced in one of two models: centralised (L&D produces all onboarding content through a formal production process) or decentralised (functional teams produce their own onboarding content — engineering onboarding is made by the engineering team, sales onboarding by the sales team — and L&D provides quality and compliance oversight). Most organisations of 200+ employees operate in a hybrid: L&D owns core compliance content and the "welcome and culture" layer, while functional teams produce role-specific training content with L&D providing templates, standards, and review.

The decentralised model creates caption governance complexity. If the engineering team is producing its own onboarding videos, the L&D team must ensure that the engineering team's production process includes a caption step. This requires either (1) a mandatory captioning step in the content production workflow enforced through an LMS publishing gate (content cannot be published to the onboarding learning path without a caption track), or (2) a post-production caption review step that L&D performs on all content before it enters the onboarding library, regardless of which team produced it. Option 1 is technically cleaner; option 2 is more realistic for organisations that do not have the LMS configuration authority to enforce workflow gates.

Caption priority tiers in production

For organisations captioning an existing onboarding library (rather than building caption-first), the caption production should follow the content priority tiers established in the taxonomy section above:

Priority tier 1 (address before any new hire cohort): All mandatory compliance modules, all role-specific technical training required for basic job function performance, all content assigned through the HRIS preboarding workflow. A new hire who encounters uncaptioned content in this tier on their first day creates an immediate ADA Title I compliance exposure.

Priority tier 2 (address within first 30 days): Company welcome and culture content, benefits and HR process content, supplementary role-specific training not required for basic job function performance. Content in this tier creates ADA compliance exposure if a new hire requests an accommodation for it, but does not create a day-one compliance failure.

Priority tier 3 (address within 90 days): Supplementary culture content, optional enrichment resources, manager and team introduction videos. Content in this tier has lowest compliance exposure but should be captioned to complete the accessible onboarding experience.

Script-first vs transcript-first production

For new onboarding content, the caption production model should be script-first: the narration script is written before recording, and the script is used as the source for caption text, with timing generated by syncing the script to the recorded audio. This is more efficient than transcript-first (where ASR is run on the recorded audio and the transcript is corrected) for two reasons: (1) the script is already 99%+ accurate for vocabulary because it was written by a human, and (2) the glossary-corrected script requires only timing validation, not vocabulary correction.

For legacy onboarding content that was recorded without a script (common for Loom walkthroughs and informal onboarding videos), the transcript-first approach is necessary: run ASR on the recorded audio, apply the onboarding glossary correction, validate timing, and produce the caption file. The remote workforce async video captioning guide covers the workflow for informally recorded video content at scale; the same workflow applies to informally produced onboarding content.

Captioning at scale: batch processing and vendor workflow

For organisations captioning a large onboarding library (20+ hours), batch processing is more efficient than video-by-video captioning. Batch captioning workflow:

Step 1: Inventory and triage. Run the LMS caption audit on the onboarding library. Identify every video asset in onboarding-tagged courses, its current caption status (captioned, uncaptioned, partially captioned, auto-captions not corrected), and its priority tier. Export this as a batch manifest with video titles, durations, and caption status. The enterprise LMS caption audit methodology covers the audit process in detail.

Step 2: Script or audio export. For video assets where scripts exist, compile the scripts into a batch manifest. For video assets without scripts, export the audio track from each video. Batch audio export is available from most LMS platforms via admin download; for videos hosted on Vimeo/Wistia/YouTube, audio extraction can be done via standard media tooling. Organise audio files in a naming convention that maps to the video titles in the batch manifest.

Step 3: Run batch ASR with glossary pre-seeding. Submit the audio batch to the captioning vendor with the onboarding glossary attached. GlossCap applies the glossary at the ASR decoding stage — the glossary terms are weighted in the ASR model before the transcription is produced, rather than being applied as post-processing text replacement. This produces lower residual error rates, especially for multi-word terms where post-processing text replacement introduces sentence-level timing disruption.

Step 4: Human review pass for vocabulary. After ASR-plus-glossary processing, the caption output goes through a human vocabulary review. The reviewer focuses on: (a) any term in the onboarding glossary that appears in a potentially incorrect form, (b) any proper noun or technical term not in the glossary that appears suspicious, and (c) any segment where the caption timing is noticeably misaligned with the audio. The vocabulary review pass is faster than a full accuracy review because the glossary has already addressed the most common error sources.

Step 5: Upload to delivery systems. Upload corrected caption files to the LMS (SRT/VTT sidecar for video assets, SRT import for SCORM content in authoring tools). For HRIS-embedded video, upload to the video host (Vimeo, YouTube, Wistia). For intranet-embedded video, coordinate caption updates with the content owners at the video host. Document the upload date and caption quality verification date for each video in the caption tracking log.

Pre-go-live quality gate before the first cohort

Why onboarding content needs a go-live quality gate

Training video content in the general LMS library can have a caption error discovered by a single employee who watches the video in isolation and reports it to L&D. The error is corrected and the update affects one video in the library. Onboarding content is different: the same video is watched by every new hire in the same cohort, often in the same week. A vocabulary error in the first video a new hire cohort watches is a vocabulary error that 10, 20, or 50 people learn simultaneously. The cost of discovering the error post-deployment and correcting it — including notifying the cohort that the term was incorrect — is significantly higher than discovering it in a pre-go-live review.

The pre-go-live quality gate is a structured review of the onboarding caption library before the first new hire cohort is assigned the content. It is not a full accuracy review of every caption in every video (that would take as long as producing the captions). It is a targeted review of the vocabulary categories most likely to have residual errors after the glossary-plus-human-review production workflow.

Quality gate structure

The pre-go-live quality gate for onboarding content should include five checks:

Check 1 — Caption presence verification. Confirm that every video asset in the mandatory onboarding content (priority tier 1) has a caption track. Open each video in the delivery system, activate the CC button, and confirm that captions appear. This sounds obvious but is the most reliably skipped step in onboarding caption programmes — the assumption that "we uploaded the SRT, so captions must be there" is frequently incorrect when the SRT file has a formatting issue, when the video host requires a specific encoding that the uploaded file does not match, or when the SCORM package was published before the SRT import was completed.

Check 2 — Vocabulary spot-check. For each video in priority tier 1, review the caption text for the first 60 seconds and the 60 seconds around the highest-vocabulary-density segment (typically the product introduction, tool walkthrough, or compliance-term-heavy section). Verify that every term in the onboarding glossary that appears in that segment is rendered correctly. A 120-second spot-check covers approximately 3–5 minutes of caption text (at normal narration speed) and will surface the majority of vocabulary errors that survived the glossary and human review pass.

Check 3 — Timing accuracy check. For each video in priority tier 1, verify that captions are synchronised with the audio at the opening, middle, and closing segments. The opening caption should appear within 0.5 seconds of the narration start. The closing caption should disappear within 1 second of the narration end. In the middle segment, check that a caption line that begins at the 30-second mark of the video is actually synchronised to the narration at that point — timing drift from the beginning of a long video can accumulate to 2–5 seconds by the middle, which is a WCAG synchronisation failure.

Check 4 — Delivery system functional test. Test caption display in each delivery system where the onboarding content is deployed: the LMS player, the HRIS-embedded video player, and the intranet-embedded video player. A caption file that renders correctly in one system may not render correctly in another due to player differences. Test the CC button, confirm caption text appears and disappears correctly, and confirm that the caption text matches the audio content (not just that text appears, but that the correct text appears).

Check 5 — Mobile device test. New hires increasingly access onboarding content on mobile devices, particularly for content that is assigned before their first day (HRIS preboarding content). Caption display on mobile devices is frequently different from desktop — smaller text, different positioning, potential overlap with player controls. Test caption display on at least one iOS device and one Android device before deploying onboarding content to a new hire cohort.

Who conducts the quality gate

The pre-go-live quality gate is conducted by a combination of L&D and the accessibility coordinator (if the organisation has one). For organisations without a dedicated accessibility coordinator, the accessibility coordinator playbook recommends designating a caption quality reviewer role within L&D — a team member who takes point responsibility for caption quality across all content categories, including onboarding. This person conducts the pre-go-live gate review, maintains the onboarding glossary, and manages the post-cohort feedback loop.

The quality gate review should be documented: a checklist that records each check, the reviewer, the date, and the outcome (pass / needs correction). This documentation serves as evidence of the organisation's reasonable accommodation compliance process — if an ADA Title I complaint is filed by a new hire about inaccessible onboarding content, the documented quality gate review demonstrates that the organisation made a good-faith effort to provide accessible content before the new hire's start date.

Eight failure modes specific to onboarding content

Failure mode 1: HRIS preboarding video added without caption verification

The HR coordinator who sets up the HRIS onboarding workflow adds an embedded video link to a preboarding welcome task without verifying that the linked video has captions at the video host. The new hire who begins preboarding receives a welcome video with no captions — the first content they see from the company is inaccessible. Prevention: establish a workflow rule that any video embedded in an HRIS preboarding task must be confirmed as captioned at the video host by the person who configures the task, with a caption URL recorded in the task configuration notes.

Failure mode 2: LMS SCORM course published before authoring-tool SRT import

The L&D team publishes a Storyline onboarding module to SCORM and uploads it to the LMS before the SRT import step is completed in Storyline. The SCORM package in the LMS has no caption data. New hires complete the module without captions. The correction requires republishing the Storyline course with SRT imported, and re-uploading the SCORM package to the LMS. Prevention: the SCORM publish workflow should include caption verification as the final step before LMS upload — the Storyline preview must show a functional CC button with correct captions before the SCORM export is authorised.

Failure mode 3: Onboarding vocabulary glossary not built before captioning

The captioning vendor or internal captioning workflow runs ASR on onboarding content without the onboarding-specific glossary, producing captions with multiple vocabulary errors on company product names and internal acronyms. New hires in the first cohort encounter incorrect terms and learn them as correct. Prevention: the onboarding glossary build is a prerequisite task in the onboarding content production plan — captioning does not begin until the glossary is built and reviewed by the content owner.

Failure mode 4: Caption update applied to LMS but not to HRIS or intranet

A caption error is discovered and corrected in the LMS caption track. The L&D team does not notify the HR team (who manages the HRIS portal) or the intranet team (who manages the Confluence/SharePoint pages). New hires who access the video via the HRIS portal or the intranet continue to see the uncorrected caption. Prevention: caption corrections to onboarding content must include a cross-system update checklist — identify every delivery system where the video appears, apply the correction to each system, and document the update date in each system.

Failure mode 5: Auto-generated captions left uncorrected in SharePoint/Stream

Onboarding-relevant content on SharePoint has Microsoft Stream auto-generated captions at 75–85% accuracy. No one reviews or corrects the auto-generated captions before new hires are directed to the SharePoint page. The content is treated as "captioned" (the CC button exists) even though the caption accuracy is below the WCAG 99% threshold. Prevention: establish a policy that SharePoint/Stream auto-captions for onboarding-tagged content must be reviewed and corrected in the Stream caption editor before the page is linked in any new hire onboarding communication or HRIS task.

Failure mode 6: Onboarding content updated without updating caption files

A product name changes. The L&D team reshoot the relevant onboarding video segment. The new video is uploaded to the LMS and the HRIS portal. The caption file is not updated — the old SRT file (which contains the old product name) is re-associated with the new video. New hires now see the new video with the old product name in the captions — a vocabulary accuracy failure that is harder to detect than a missing caption because a CC button exists. Prevention: any onboarding content update triggers a caption file review step — confirm that the caption file matches the new video content, and regenerate or correct the caption file if the content has changed.

Failure mode 7: No caption error reporting path for new hires

Caption errors in onboarding content go unreported because new hires do not know how to report them, are reluctant to flag issues in their first week, or assume that the L&D team is already aware. Errors persist through multiple cohorts. Prevention: include a brief, one-sentence caption error reporting path in the new hire welcome communication — "If you notice any caption issues, email [captions@company.com] or message [slack channel]." The path should be frictionless and explicitly welcoming. Many new hires will use it — first-week new hires are disproportionately good caption error reporters because they are reading captions carefully and noting everything unusual.

Failure mode 8: CC button visible but nonfunctional due to LMS iframe restriction

A Storyline or Captivate SCORM onboarding module has correct caption configuration and displays correctly in Articulate Review or local SCORM test. When uploaded to the LMS (Cornerstone, Moodle, or a less common LMS with strict iframe policies), the JavaScript that controls caption button functionality is blocked by the LMS's Content Security Policy. The CC button appears in the player but does not respond to clicks, or clicks the button displays no caption text. New hires see a CC button and assume captions are available — they are not. Prevention: LMS SCORM functional testing (specifically caption button functionality) must be performed in the target LMS player, not just in the authoring tool preview. The pre-go-live quality gate check 4 specifically addresses this.

Seven-question FAQ

Does ADA Title I require proactive captioning of all onboarding content, or only when a new hire requests an accommodation?

ADA Title I's reasonable accommodation framework is technically triggered by an individual accommodation request — the employer's obligation to provide a specific accommodation arises when an employee or applicant discloses a disability and requests an accommodation. However, courts and the EEOC have increasingly applied a "systemic barrier" theory to employers whose training programmes are structurally inaccessible: maintaining a training library that cannot accommodate employees with hearing disabilities is itself evidence of discriminatory conditions of employment, even if no individual complaint has been filed.

The practical legal guidance from employment law practitioners is: treat ADA Title I as requiring proactive captioning of mandatory onboarding content. The risk of an accommodation request arriving with less than 24 hours of advance notice (day-one disclosure) is sufficiently high, and the cost of reactive captioning under that constraint is sufficiently greater than proactive captioning, that proactive captioning is the legally and operationally defensible approach. For non-mandatory onboarding content, the accommodation request trigger model is more defensible — but only if the response mechanism is rapid (same-day or next-day captioning available for requested content).

What is the caption standard for onboarding content — 99% accuracy or something lower?

The WCAG 2.1 SC 1.2.2 accuracy standard does not specify a percentage — it requires captions that "provide equivalent access to the audio content." The 99%+ accuracy benchmark comes from the DCMP (Described and Captioned Media Program) Captioning Key, which is the most widely cited technical accuracy standard for instructional media captions. The DCMP standard applies to synchronised media in educational and instructional contexts, which includes employee onboarding training. For ADA Title I compliance in an employment context, the defensible standard is 99%+ synchronised accuracy — the same standard that applies to any other training content. The 99% accuracy rationale covers why lower accuracy thresholds (95%, 98%) fail the "equivalent access" standard for training content with technical vocabulary.

Our onboarding videos are on Loom, which has auto-captions. Does that satisfy ADA Title I?

Loom auto-captions do not satisfy ADA Title I as delivered, for the same reason that any auto-generated captions without human review do not satisfy the standard: auto-caption accuracy on content with company-specific vocabulary is typically 75–85%, below the 99%+ accuracy required for equivalent access. The presence of a CC button and some caption text does not constitute compliance — compliance requires accurate, synchronised captions. Loom provides a caption editing interface where you can correct the auto-generated captions word-by-word. Loom captions that have been reviewed and corrected for accuracy — particularly for vocabulary errors on company product names and internal terms — can satisfy the standard. The workflow is: edit captions in Loom's editor after recording, correct vocabulary errors against the onboarding glossary, save the corrected captions. This satisfies the standard. Unreviewed auto-captions do not. The auto-captions compliance analysis covers this distinction in detail for all major platform auto-caption systems.

We use a third-party vendor for our mandatory compliance onboarding modules. Do we need to caption those separately?

The caption obligation under ADA Title I rests with the employer who assigns the training to the employee — not solely with the vendor who produced it. If you purchase compliance training modules from a vendor (ComplianceWire, Navex One, Skillsoft, OpenSesame) and assign them to new hires as mandatory onboarding content, you are responsible for ensuring those modules are accessible to employees with hearing disabilities. The practical question is whether the vendor-provided modules have WCAG-compliant captions. Most enterprise compliance training vendors include caption tracks; the quality varies. Review the vendor's caption tracks for accuracy before deploying to new hire cohorts, and raise any accuracy issues with the vendor as part of your procurement and contract review process. If the vendor cannot provide accurate captions, you have two options: caption the vendor content yourself (which may have contractual implications) or replace the vendor content with internally produced accessible content.

How do we handle onboarding content that is updated frequently? Do we recaption everything each time?

Not necessarily. Caption files are associated with specific video content — if a segment of an onboarding video is reshot, only the caption file for that segment needs to be updated, not the entire onboarding library. The key is tracking which videos were changed and which caption files correspond to which video versions. This is a content version management problem as much as a caption problem. The approach: (1) maintain a content version log for onboarding videos — every reshot, the new video replaces the old with a version date, (2) for each reshot video, regenerate or correct the caption file from the new audio, (3) update the caption file in all delivery systems (LMS, HRIS video host, intranet video host). For frequently updated content (quarterly product update videos, for example), building the caption production into the video production workflow — as a post-production step that is part of the video's "done" definition — is more efficient than treating captioning as a separate process that runs after the video is already deployed.

Can we use the same caption file for the LMS version and the HRIS-embedded version of the same video?

The SRT caption file content is the same for both systems — the caption text and timing data do not change based on the delivery system. What changes is where the file is stored and how it is associated with the video. For the LMS, the SRT file is uploaded to the LMS media library or course unit as a caption track. For the HRIS-embedded version (which is delivered via a video host like Vimeo or YouTube), the same SRT file is uploaded to the video host's caption management interface. The files have identical content — you produce one SRT file per video and deploy it to both systems. The failure mode to avoid is uploading a corrected SRT to the LMS but forgetting to update the video host (Vimeo/YouTube), resulting in the LMS and HRIS portal showing different caption versions of the same video.

What should we do if a new hire's first day is in 48 hours and the onboarding content is not captioned?

Address the highest-priority content first: mandatory compliance modules and role-specific technical training required for basic job function performance. These are the content categories where a day-one caption absence creates the most immediate ADA Title I exposure. For 48-hour turnaround captioning, the options are: (1) a rush captioning vendor (typically 2–3x standard rate, available from most enterprise captioning vendors), (2) internal captioning with a dedicated L&D team member doing vocabulary-corrected ASR review on priority-tier-1 content only, or (3) temporarily replacing video content with text-based equivalents (a written script or transcript document) while captions are being produced — a text document is not the same as a synchronised caption track but satisfies the "access to the content" requirement in the immediate term. Notify the new hire proactively — acknowledging the situation, providing the text alternative, and stating the timeline for full caption completion, demonstrates a good-faith accommodation effort. Document the notification and the timeline commitment. Then prioritise the full caption compliance programme build so this scenario does not recur.