Rolling out a captioning programme to your L&D team: workflow adoption, producer training, and the resistance patterns that stall programmes

Why captioning programme rollouts stall: the three structural gaps

The failure mode of most captioning programme rollouts is not dramatic. There is no moment at which the programme visibly collapses. What happens instead is gradual attrition: submission rates that start at 85% in the first month drop to 70% in month three and 55% in month six. The drop is not uniform — it concentrates in specific departments where producers were undertrained, in content types that the governance policy left ambiguous, and in workflows where the submission step was not integrated into the existing publishing process. Understanding why rollouts stall requires understanding the three structural gaps that cause the attrition, because the mitigation for each gap is structural, not motivational. Reminding producers that captioning is a compliance requirement does not fix a workflow that requires them to log into a separate portal. It just adds friction to a workflow that was already too friction-heavy to sustain.

Gap 1: Workflow integration

The most common cause of submission rate decay is workflow fragmentation: the captioning submission step is not integrated into the existing video publishing workflow, so it requires a separate action that producers must remember to take after they have already completed what feels like the "done" step of uploading the video to the LMS. In a typical unintegrated workflow, a producer records a video in their screencasting or recording tool, exports it, uploads it to the LMS, marks the module as published, and moves on to the next task. The captioning step — go to the vendor portal, upload the video file again, wait for delivery, download the SRT, go back to the LMS, locate the published video, upload the SRT — is a multi-step interrupt that requires the producer to context-switch after they have already completed the natural end of their workflow. Every step in that sequence is an opportunity for the submission to be deprioritised, forgotten, or explicitly skipped under time pressure.

The structural fix for workflow fragmentation is LMS-level integration: the captioning request is triggered at the point of LMS upload, not as a separate subsequent action. Some LMS platforms support native captioning vendor integrations that automate the submission step entirely — the producer uploads the video, the LMS sends it to the vendor, the SRT is returned to the LMS directly without the producer ever touching a separate portal. Where native integration is not available, the submission can often be simplified to a single-link action: click the "request captions" button that appears on the video upload confirmation page, which pre-populates the submission with the video file and the producer's contact details. The target workflow for any producer is: upload video, click one button, review captions when they arrive, mark as reviewed. Any workflow that requires more than that from a non-specialist producer will see submission rate decay. The LMS migration caption checklist covers the integration layer in detail for the most common platforms; the LMS-specific rollout notes section below covers what each platform's integration looks like from the producer's perspective.

Gap 2: Scope ambiguity

The second structural gap is scope ambiguity: when the governance policy does not precisely define what triggers the captioning requirement, producers interpret ambiguous cases in the direction that requires less work from them. This is not bad faith — it is a rational response to an ambiguous rule under time pressure. The producer who asks "does this 3-minute department update video need captions?" and cannot find a clear answer in the policy documentation will make an on-the-spot judgement call. Under time pressure, with a deadline on the module launch, the judgement call is almost always "probably not."

The ambiguities that cause scope creep in most organisations follow a predictable pattern: short videos (under 5 minutes) that producers classify as informal content; videos created for internal-only audiences that producers assume are exempt from public accessibility law; videos that are "temporary" placeholders pending a more polished version; videos in languages other than the primary LMS language; videos that contain no spoken content except background music; and videos that are third-party embeds rather than native LMS uploads. Each of these categories will exist in your organisation, and producers in each category will need a written decision — not a policy principle, but a decision tree: "if the video is under 2 minutes and contains no instructional content, is it in scope?" with a yes/no answer and the responsible authority. The content gate clause of the governance policy covers this in detail; the rollout training must translate the policy language into a one-page quick-reference decision tree that producers can consult without reading the full policy document.

Gap 3: Accountability absence

The third structural gap is the absence of a monitoring mechanism that makes non-compliance visible. In a well-designed captioning programme, every video published to the LMS without a caption file generates a gap in the submission record. That gap is reported monthly to the accessibility coordinator, who reviews it, contacts the producer, and either resolves it or escalates it. The producer knows that publishing without captions produces a record. That record is reviewed. There is a consequence — not necessarily a punitive one, but at minimum a follow-up contact and a required remediation. In most rollouts, this monitoring mechanism does not exist in the first six months. The baseline submission tracking has not been configured, the accessibility coordinator is not yet running the monthly review process, and the manager escalation path has not been tested. In the absence of monitoring, producers who are time-pressed will calculate that the probability of negative consequence from skipping the caption step is low, and they will be right. The structural fix is to configure the monitoring mechanism — submission tracking by producer and by department — before launch, run the first monthly review in the first month of live operation even if the submission rate is 100%, and ensure that the first non-compliance follow-up happens within the first 90 days so that producers learn that the monitoring is real.

The stakeholder map before you start

A captioning programme rollout involves more stakeholders than the producers who will use the workflow daily. Understanding who each stakeholder group is, what their stake is, and what action you need from them before or during rollout is prerequisite work for the change management plan. The stakeholder map for a mid-market L&D organisation typically has six groups.

Video producers (the primary actors)

Video producers are the highest-volume actors in the captioning programme. They are the people who create the videos, submit them for captioning, review the output, and upload the caption files to the LMS. In a mid-size L&D organisation (200–1,000 employees), there are typically 10–40 video producers with varying degrees of technical sophistication and video production frequency. The producer population is heterogeneous: some producers create ten or more videos per month and will notice every friction point in the captioning workflow; others create one or two per quarter and will need to be retrained each time they use the workflow. The rollout must account for both populations. High-frequency producers need a frictionless, automated-as-possible workflow; low-frequency producers need a workflow reference they can find and follow without external help when they return to it after a three-month gap.

The producer population is also departmentally distributed. L&D may have five or six staff producers who become fluent with the workflow early and maintain it well. But HR, Compliance, Sales Enablement, and Operations may each have one or two part-time video producers who create training content as a secondary responsibility. These secondary producers are the highest-risk population for submission rate decay: they have less stake in the captioning programme, less contact with the accessibility coordinator, and more competing priorities. Identifying them by department before launch and ensuring each has a trained contact for workflow questions is one of the highest-leverage pre-launch actions.

LMS administrators (the integration gatekeepers)

LMS administrators control the technical configuration of the platform where caption files are delivered and displayed. They are not typically video producers, and they may or may not be part of the L&D team. Their stake in the captioning programme is primarily technical: they need to ensure that the LMS is configured to accept SRT or VTT files in the correct format, that the captioning vendor's integration is configured correctly, and that the caption display settings meet the WCAG 2.1 AA contrast and timing requirements. They also typically control the submission tracking reports that the accessibility coordinator will use for monthly monitoring. Getting LMS administrator cooperation early — ideally in the foundation phase before any producer is trained — is a prerequisite for a functioning workflow. An LMS that is not configured to display captions correctly will undermine producer confidence in the workflow: if a producer reviews and approves captions that then display incorrectly in the published module, they will lose trust in the process and become resistant to future submissions.

Department leads and L&D managers (the proxy influence layer)

Department leads and L&D managers are not part of the captioning workflow operationally, but they set the cultural norm that determines whether producers prioritise the workflow under time pressure. A producer whose manager has communicated — explicitly and visibly, not just in a policy email — that caption compliance is a required component of video publication will prioritise the caption step differently from a producer whose manager has never mentioned it. This is the proxy influence layer: the people who shape the daily prioritisation calculus of the producers who are doing the work. The change management plan must include a brief (30-minute) briefing for department leads before launch that covers three things: what the programme requires from their producers, what the monitoring mechanism is (so that they understand non-compliance will be visible), and how to handle the first few cases where a producer under time pressure asks for an exception. Department leads who understand the exception procedure and can communicate it clearly to their producers are significantly more effective than department leads who refer producers to "check with the accessibility coordinator."

The accessibility coordinator (the programme owner)

The accessibility coordinator is the programme owner who bears the operational accountability for caption compliance. In the rollout phase, their primary responsibilities are producer training, monitoring setup, exception handling, and escalation management. They are the person producers contact with workflow questions, the person who reviews the monthly submission report, the person who follows up on non-compliance, and the person who approves or denies exception requests. In organisations that do not have a dedicated accessibility coordinator, this role is typically assigned to a senior L&D manager or the L&D director, which creates a capacity problem: the director has limited time for the operational details of monthly submission review and individual producer follow-up. If the accessibility coordinator role is not dedicated, the rollout plan must explicitly define the time allocation — typically four to eight hours per month in the first six months — and protect it from competing priorities. Rollouts where the accessibility coordinator is nominally designated but operationally overloaded will fail to maintain the monitoring mechanism, which then allows the accountability gap to open.

Legal and compliance (the external pressure source)

Legal and compliance are not operational participants in the captioning programme, but they are the source of the urgency that makes the programme necessary and the authority that makes it fundable. In the rollout phase, legal's primary role is to validate the compliance threshold — confirming that the accuracy standard and content scope in the governance policy meet the organisation's legal obligations under ADA Title II, Section 508, or WCAG 2.1 AA — and to serve as the escalation backstop for cases where a producer's non-compliance creates a legal exposure. Legal does not need to be involved in day-to-day programme operations, but their written validation of the compliance framework before launch gives the accessibility coordinator the authority position they need when producers challenge the scope of the requirement. An accessibility coordinator who can say "legal has reviewed and approved this content gate" has a much stronger position than one who can only say "this is what I think the policy means."

IT and procurement (the vendor access and integration layer)

IT and procurement control access to the captioning vendor's systems, the SSO and data-sharing agreements that determine how the vendor integrates with the LMS, and in some organisations the budget approval process for the ongoing vendor spend. In organisations with strict vendor onboarding requirements, the IT procurement process can take 4–8 weeks and should be started in the foundation phase, not during the producer training phase. The most common rollout delay caused by IT is SSO configuration: if the LMS vendor integration requires SSO credentials for the captioning vendor's portal, and SSO configuration requires an IT security review, starting that process after the programme is otherwise ready to launch will create a 4–8 week gap during which producers are trained but cannot use the workflow. Starting SSO configuration in week 1 of the foundation phase, in parallel with governance policy finalisation, is the standard mitigation.

Phase 1: Foundation (weeks 1–4)

The foundation phase prepares the infrastructure that the producer training phase depends on. Going into producer training without a functioning LMS integration, a finalised governance policy, and a configured monitoring mechanism means that producers will be trained on a workflow that does not yet work, which creates a trust deficit that is hard to recover from. The four foundation phase deliverables are: a finalised governance policy, a completed baseline audit, a functioning LMS-to-vendor integration, and a configured glossary.

Finalise the governance policy before training anyone

The governance policy must be written, approved by the appropriate authority (typically the L&D director and legal), and published in an accessible location before any producer is trained. Producer training references the policy for the content gate decision tree and the exception procedure; if the policy is still in draft or under revision during producer training, producers will encounter inconsistencies between what they are trained on and what the published document says, and they will lose confidence in the programme's stability. The process of finalising the policy — particularly the content gate clause and the exception procedure — often surfaces disagreements between legal, the L&D director, and the accessibility coordinator about scope that need to be resolved before they are presented to producers as settled decisions. The governance policy template post covers the four clauses in detail; the practical note for the rollout is that the policy finalisation process should include a review pass specifically for producer-facing language: is the content gate decision tree comprehensible to a producer who is not an accessibility specialist? Are the exception procedure steps clear enough that a producer can follow them without calling the accessibility coordinator?

Complete the baseline audit

The baseline audit establishes the scope of the existing uncaptioned content backlog and the starting compliance rate. This matters for the rollout for two reasons. First, the rollout change management plan needs to address the backlog explicitly: producers who are trained on the new workflow for new content will inevitably ask about the existing uncaptioned content library. If the answer is "we're remediating it on a priority schedule and it will be done by [date]," producers accept that and focus on the new-content workflow. If the answer is vague, producers will experience cognitive dissonance — "we're told to caption new content carefully, but the existing library is 40% uncaptioned and nobody is addressing it" — which undermines the programme's credibility. Second, the baseline audit identifies which producers and departments already have content quality gaps, which allows the training plan to prioritise high-risk departments for closer follow-up in the live period. The enterprise LMS caption audit methodology covers the audit process; the relevant output for rollout planning is a producer-by-department compliance heatmap that the accessibility coordinator uses to sequence the live-period monitoring.

The backlog remediation decision — remediate all uncaptioned content before launch, or launch the new-content workflow first and address the backlog on a priority schedule — is a resource allocation question that depends on the size of the backlog and the available remediation budget. Most organisations choose the phased approach: launch the new-content workflow for all newly published content immediately, and remediate the existing backlog over 6–18 months prioritised by content priority score (mandatory compliance training first, audience size second, content age third). The governance policy should document the backlog remediation timeline so that it is an explicit programme commitment, not an informal intention.

Configure the LMS-to-vendor integration

The LMS integration should be functioning end-to-end — including the return delivery of the caption file to the LMS, not just the outbound submission — before producer training begins. This means running at least five end-to-end test submissions through the complete workflow: a producer-role user submits a test video, the vendor receives and processes it, the caption file is returned to the LMS, the producer-role user reviews and approves it, and the caption displays correctly in the published module. Running these tests before training reveals workflow gaps that are very difficult to correct during the training phase: submission confirmation emails that go to the wrong address, SRT files that are returned in a format the LMS rejects, caption display settings that do not meet WCAG contrast requirements. The LMS-specific rollout notes section below covers the most common integration issues by platform.

Build the initial glossary

The captioning vendor's glossary — the domain-specific vocabulary list that biases the ASR model toward the organisation's technical terms — should be seeded before the first production submission. A glossary that is empty at launch means that the first batch of production captions will have a higher error rate on domain vocabulary than the glossary-assisted captions that will be the norm after several months of operation. Producer trust in the captioning workflow is established in the first weeks of operation; if the first captions they review have a high error rate on obvious domain vocabulary (product names, regulatory acronyms, technical terms from the training content), producers will lose confidence in the review step and either spend disproportionate time correcting captions or stop reviewing carefully and upload low-quality files. The glossary architecture post covers the term sourcing and ingestion process; for rollout planning, the practical note is to seed the glossary with at minimum 30–50 high-frequency domain terms before the first production submission, sourced from the L&D content library's most common training topics.

Phase 2: Producer training and workflow adoption (weeks 5–8)

Producer training is the highest-leverage activity in the rollout. Done well, it establishes a workflow habit that producers maintain with minimal subsequent intervention. Done poorly — too comprehensive, too abstract, not grounded in the producer's actual LMS and actual content — it creates a training event that producers remember as a compliance checkbox rather than a practical skill. The training design principles that produce sustained workflow adoption are different from the principles that produce content knowledge. The goal is not that producers understand why WCAG 2.1 AA matters or what the legal history of ADA captioning compliance is. The goal is that every producer can perform three actions independently without external help: identify whether a video they are about to publish is in scope, submit a video for captioning, and review the returned caption file before upload.

The three things every producer needs to know

The content of the producer training should be scoped to exactly three capabilities, not expanded beyond them. Additional context — the compliance history, the accuracy methodology, the vendor selection rationale — can be provided as supplementary reference material for producers who want it, but it should not be part of the required training for producers who need only to operate the workflow.

Capability 1: Identify whether a video is in scope. The producer needs to be able to apply the content gate decision tree to any video they are about to publish and reach a confident yes/no answer without consulting the accessibility coordinator. The training should walk through at least six example scenarios — a short informal update video, a formal compliance training module, a video with no spoken content, a video embedded from an external platform, a video in a foreign language, a video that is a revised version of a previously captioned video — and show the decision tree applied to each. The producer should leave training confident that they can apply the same logic to edge cases they have not seen.

Capability 2: Submit a video for captioning. The submission workflow should be demonstrated using the producer's actual LMS, not a generic demonstration environment. If the organisation uses Kaltura, the training shows the Kaltura submission workflow. If it uses Cornerstone OnDemand, the training shows the Cornerstone workflow. The producer should perform the submission themselves during training on a test video from their own content library — not watch a demonstration — so that they have completed the full workflow at least once before they are expected to do it in production. The training should also cover what happens after submission: the expected delivery time, what the notification email looks like, and where to find the returned caption file in the LMS.

Capability 3: Review the returned caption file before upload. Caption review is the quality gate that the producer is responsible for. The training should establish clearly what "review" means: not a word-for-word accuracy verification of the entire transcript (that is the vendor's responsibility), but a pass that catches systematic errors — wrong product names, incorrect regulatory acronyms, speaker identification errors — and flags them for correction before the file is uploaded. The review should take no more than 10–15 minutes for a 30-minute video if the vendor has delivered an accurate file. The training should show what a good review pass looks like and what errors require correction versus what errors are below the threshold. The caption QA methodology post covers the review protocol in detail; the producer training should cover only the producer-role steps of that protocol, not the full accessibility coordinator QA workflow.

Training session design

The producer training session should be 30–45 minutes maximum for the required content, followed by 15–30 minutes of hands-on practice. Sessions longer than 60 minutes produce diminishing retention in the skills that matter operationally. The session should be conducted in the LMS environment the producer actually uses, with a real video from their own content library as the training example. Generic examples ("here is a sample video") produce less durable habit formation than examples drawn from the producer's actual work.

The session should be conducted in small groups (5–8 producers) rather than all-hands events for two reasons. First, small groups allow the trainer to observe each producer's hands-on practice and correct errors before they become habits. A producer who submits a video to the wrong queue in a 5-person session gets corrected immediately; a producer who does the same thing in a 40-person session may not be noticed. Second, small groups allow the content to be tailored to the LMS and content types used by that group. A group of Sales Enablement producers who use WorkRamp needs different specific instructions than a group of L&D producers who use Docebo, even if the underlying captioning workflow is the same. The LMS-specific notes section covers the training differences by platform.

Each training session should produce one persistent artefact: a one-page quick-reference card specific to the LMS the group uses, covering the content gate decision tree (5–7 bullet points), the submission steps (numbered, LMS-specific), the review steps (numbered, what to look for), and the contact for questions (the accessibility coordinator's name and email). This reference card is what low-frequency producers will use when they return to the workflow three months after their training session and cannot remember the specific steps. It should be stored in a location that producers can find without help — pinned in the team communication channel, linked from the LMS home page, or included in the new-hire onboarding checklist.

Champion producers

In each department with significant video production volume, identify one or two producers to train first as programme champions. Champion producers are trained before the general producer cohort, given a more complete view of the programme (including the exception procedure and the monitoring mechanism), and positioned as the first point of contact for questions from other producers in their department. The champion role is informal — it adds perhaps 30 minutes per month to the champion's responsibilities — but it provides three things the accessibility coordinator alone cannot: departmental visibility (the champion knows which producers are struggling before the submission report shows it), local credibility (producers trust a colleague who does the same work over a programme administrator they rarely interact with), and a distributed capacity buffer (the champion can answer basic workflow questions without routing them to the accessibility coordinator, who may be managing multiple departments simultaneously).

Recruiting champion producers before the general training rollout also allows you to test the training materials and the workflow with a small group before training the full population. Champions will surface the questions and failure points that the training session needs to address, giving you one revision cycle on the training materials before the full rollout.

Phased department rollout

The producer training should not be conducted as a simultaneous all-organisation event. A phased rollout — department by department over 4–6 weeks — produces better outcomes for three reasons. First, it allows the accessibility coordinator to manage the support demand from newly trained producers without being overwhelmed. When 40 producers are trained simultaneously, the accessibility coordinator receives 40 producers' worth of workflow questions in the first two weeks. When 8 producers are trained per week over five weeks, the question volume is distributed and manageable. Second, it allows the training materials to be refined between cohorts based on the questions each group raises. The questions that producers ask during training are a diagnostic of the scope ambiguities and workflow friction points that the materials have not addressed adequately. Third, it allows the monitoring mechanism to start with a smaller population where the accessibility coordinator can observe the first month of submissions closely, identify non-compliance patterns, and adjust the workflow before they affect the full producer population.

The sequencing of departments should follow the baseline audit's compliance heatmap: departments with the highest existing caption compliance rates go first (they will adopt the new workflow more easily and provide positive social proof for later cohorts), departments with the highest new video production volume go second (their submission data will be the most informative in the first monitoring period), and departments with the most resistant producers or the most complex content types go last (by which time the workflow is stable, the champion network is established, and the accessibility coordinator has experience managing the exception procedure).

Phase 3: Live-period monitoring and calibration (weeks 9–16)

The live period is the eight weeks after the full producer population has been trained and the programme is operating in production. It is the highest-risk period for submission rate decay: the training event is recent but fading, the monitoring mechanism is new and may not yet be producing reliable data, and the producers who are most likely to find workarounds are testing the system. The live-period monitoring cadence exists to detect decay early — when it is recoverable with individual follow-up — rather than late, when it has become the de facto standard in resistant departments.

The monthly submission review

The accessibility coordinator should run a monthly submission review starting in the first month of live operation. The review has three outputs: the organisation-wide submission rate (total captioned videos published as a percentage of total videos published, excluding out-of-scope content); the submission rate by department; and the list of individual producers whose submission rate is below the threshold (typically below 90% for a producer with more than 10 videos published that month). The review should take no more than two hours per month once the monitoring mechanism is configured and the report is running automatically. In the first month, it may take longer as the accessibility coordinator calibrates the report to correctly identify out-of-scope content and exclude it from the denominator.

The department-level submission rate is more actionable than the organisation-wide rate because it identifies where the problem is localised. An organisation-wide rate of 85% that is composed of 100% compliance in six departments and 30% compliance in two departments is a very different problem from an organisation-wide rate of 85% that is uniformly distributed. The former requires targeted intervention in two departments; the latter requires a structural fix in the workflow or training that applies across all departments. The monthly review should always produce both the aggregate and the department breakdown.

First non-compliance follow-up

The first time a producer with a below-threshold submission rate appears in the monthly report, the follow-up should happen within five business days of the report date. The follow-up is not a compliance enforcement action — it is a workflow support contact. The accessibility coordinator (or the department champion, depending on the organisation) contacts the producer, confirms that they understand the workflow, asks if there are any friction points or questions, and offers to do a brief workflow refresher if needed. The framing is supportive, not punitive: "I noticed some of your videos this month don't have captions yet — can I help you get those submitted?" This framing produces producer cooperation rather than defensive resistance and is consistent with the change management goal of making the workflow habitual rather than merely obligatory.

If the producer's submission rate remains below threshold for two consecutive months after a supportive follow-up, the escalation path moves to their manager. The manager escalation should reference the governance policy's exception procedure and accuracy standard — "the policy requires captions for all training content meeting these criteria, and [producer name]'s content in [department] has not been meeting that standard" — and ask the manager to reinforce the expectation. In most cases, one manager escalation is sufficient to restore compliance; the monitoring mechanism has made the gap visible, and visibility is the primary behaviour-change mechanism for producers who were skipping the step under time pressure rather than out of principled resistance.

Exception handling in practice

The exception procedure — the mechanism by which producers request a formal deviation from the standard captioning requirement — will be invoked in the first weeks of live operation. The accessibility coordinator should expect exception requests for all four eligible categories: technical barrier exceptions (the video file is in a format the vendor cannot process), timeline exceptions (the content must be published before the vendor turnaround time allows), content-type exceptions (the video contains no instructional spoken content), and extraordinary-circumstances exceptions (force majeure or organisational disruption). The first round of exception requests reveals whether the exception procedure language is clear enough for producers to self-serve — to submit a request with the required information — or whether the coordinator must spend significant time clarifying the procedure in each case. If most first-round exception requests are incomplete or misclassified, the producer-facing exception request form needs revision before the programme reaches steady-state operation.

The exception log — required under the governance policy's audit trigger clause — should be set up and populated from day one of live operation, not retroactively. Every approved exception should be logged with the video ID, the producer name, the exception category, the approval date, the approved exception duration, and the scheduled remediation date (if applicable). The exception log is a governance document: it creates the paper trail that demonstrates that exceptions were reviewed, approved by the appropriate authority, and time-limited. An exception log that is started three months after launch, when the accessibility coordinator realises that the audit trigger is approaching, will have gaps that are difficult to reconstruct.

Calibration during the live period

The live period will surface gaps in the workflow and the training that were not visible during the foundation and training phases. These gaps are calibration opportunities, not failures. Common calibration items in the first 8–12 weeks of live operation include: a content type that producers consistently misclassify as out-of-scope (requiring a governance policy clarification or a training update); an LMS-specific submission step that produces errors at higher than expected rates (requiring a workflow revision or additional training); a vendor turnaround time that is longer than the SLA in certain content types (requiring a vendor conversation or a timeline exception process revision); and a glossary gap that is causing systematic errors in a specific content domain (requiring glossary term additions). Each of these calibration items should be documented, addressed in the relevant artefact (governance policy, training materials, vendor SLA, glossary), and communicated to the producer population. Calibration communications — brief, specific, actionable updates to the workflow — build producer confidence that the programme is responsive and improving.

The four resistance patterns and their structural counters

Every captioning programme rollout encounters producer resistance. Resistance is not primarily a motivational problem — producers who understand why captioning matters and agree that it is a good idea still skip the workflow step when they are under time pressure and the step requires significant effort. The four resistance patterns that appear in virtually every rollout have structural causes, and they have structural counters. Motivational interventions — reminding producers of the legal requirement, making an accessibility values case, explaining the consequences of non-compliance — do not fix structural problems and should not be the primary response to resistance.

Pattern 1: "This is extra work" (workflow friction)

The producer who says "captioning is extra work" is describing a workflow integration failure. For a producer whose captioning submission requires leaving their normal publishing environment, uploading the video file to a separate vendor portal, managing a separate communication stream for the delivery, downloading and re-uploading the file, and returning to the LMS to complete the upload — the captioning step is genuinely extra work, not just different work. The resistance is accurate feedback about the workflow design, not evidence of producer non-compliance.

The structural counter is workflow integration: the captioning request should trigger automatically at the point of LMS upload, the vendor should return the file directly to the LMS, and the producer's role should be reduced to two steps — submit and review — that together take no more than 5–10 minutes for a standard video. Where full automation is not possible, the submission step should be integrated as a single-click action on the video upload confirmation page. The hidden correction labour cost post provides the business case for investing in workflow integration: the labour cost of manual workarounds — fragmented submission workflows, producer-side SRT downloads and re-uploads, high correction rates from late or incomplete submissions — is typically higher than the integration investment required to eliminate them.

Pattern 2: "My content is exempt" (scope ambiguity)

The producer who claims exemption for their content is almost always responding to genuine scope ambiguity in the governance policy, not attempting to evade a clear requirement. The cases that producers most commonly claim as exempt — short videos, internal-only audiences, no-narration screencasts, temporary placeholder content, third-party embedded videos — are legitimately ambiguous in many governance policies because the policy authors wrote the content gate clause for the modal case (a standard training module with narrated instruction) and did not enumerate every edge case. The producer is not wrong that the policy is ambiguous; they are wrong only in concluding that ambiguity means exemption.

The structural counter is a content gate decision tree with explicit coverage of the categories producers most commonly misclassify. The decision tree should be written as a series of yes/no questions — not a prose paragraph — and should terminate at a clear "in scope" or "out of scope" determination for every path. The six scenarios covered in the producer training (short informal update, formal compliance module, no-spoken-content video, external platform embed, foreign language video, revised version of captioned content) should all appear in the decision tree as example cases for the relevant branch. The decision tree should be a separate one-page document, not embedded in the full governance policy, so that producers can consult it as a reference tool without reading the full policy. Where a producer encounters a case not covered by the decision tree, the decision tree should direct them to the accessibility coordinator for a ruling, not to their own judgement.

Pattern 3: "Nobody's checking" (accountability gap)

The producer who assumes nobody is checking is making a rational inference from the absence of visible monitoring. If a producer publishes twenty videos without caption files in the first month of the programme and receives no follow-up, the inference that monitoring is not real is reasonable. The accountability gap is not a character failing; it is a system design failure. Producers who would follow the workflow if they knew that non-compliance was being tracked do not follow it when they have evidence that it is not.

The structural counter is a monitoring mechanism that produces visible evidence of its operation from the first month. The first monthly report should be shared with department leads — not just the accessibility coordinator — so that producers understand that their managers have visibility into submission rates. The first follow-up contact should happen within five business days of the first report, even for producers who have only missed one or two submissions, so that producers learn quickly that the report produces a response. The follow-up does not need to be punitive to be effective; it needs to be timely and consistent. A supportive follow-up — "I noticed a gap and wanted to make sure you have what you need to submit" — is effective precisely because it demonstrates that the monitoring is real and responsive, which is the information producers are acting on when they decide whether to skip the step under time pressure.

Pattern 4: "The vendor is responsible" (vendor blame deflection)

The producer who attributes caption quality problems to the vendor — "the captions are wrong, it's not our fault" — is partially right: vendor accuracy is the vendor's responsibility. But the producer's role in the quality chain is the review step: reviewing the returned caption file before uploading it to the LMS, catching systematic errors, and either correcting them or returning them to the vendor for revision. A producer who uploads a caption file they have not reviewed, and then blames the vendor when the quality is low, is conflating the vendor's production responsibility with the producer's review responsibility.

The structural counter is a clear review protocol that specifies what the producer is responsible for finding and what they are not responsible for fixing. The review step is not a comprehensive accuracy audit — that is the accessibility coordinator's QA function. The review step is a systematic pass through the caption file for the error types that are most likely to appear and most visible to the content audience: wrong product names, incorrect acronyms, speaker identification errors, and formatting failures (all-caps segments, missing punctuation on boundary errors). A producer who can perform this review in 10–15 minutes for a 30-minute video has fulfilled their review responsibility; errors that pass the review are vendor-quality issues that should be addressed through the vendor SLA conversation, not attributed to the producer's failure. Giving producers a clear, bounded review responsibility — one they can fulfill without becoming caption specialists — reduces vendor blame deflection because producers have a clear answer to "what are you responsible for?" that does not require them to be responsible for everything.

LMS-specific rollout notes

Each LMS platform presents a distinct set of producer-facing workflow steps and administrator-facing configuration requirements. The notes below cover the ten platforms where rollout friction is most common, based on the workflow patterns described in prior posts covering LMS migration caption management and async video captioning. These notes are not full integration guides — those are covered in the platform-specific SEO pages — but they identify the rollout-specific issues that arise in producer training and live-period monitoring for each platform.

Kaltura

Kaltura has the most mature captioning vendor integration ecosystem of any L&D video platform. The native REACH integration (Kaltura's caption vendor marketplace) allows producers to submit videos for captioning directly within the MediaSpace or KMS interface without leaving the platform. For rollouts using the REACH integration, the submission step reduces to a single checkbox in the video upload workflow, which eliminates the primary source of workflow friction. The rollout risk in Kaltura environments is on the review side: Kaltura's caption editor is powerful but unfamiliar to producers who are not technically fluent, and the review step requires producers to understand how to open the caption editor, navigate the timeline, and identify errors by scanning the transcript. Producer training in Kaltura environments should include a dedicated 10-minute segment on the caption editor that covers opening the file, using the search function to jump to specific terms, and saving without overwriting the delivery-quality file. LMS administrators configuring Kaltura for the rollout should verify that the REACH integration is configured with the correct output format (SRT vs DFXP vs 608 closed captions, depending on the player configuration) before producer training begins.

Docebo

Docebo's captioning workflow is file-based rather than integration-based: producers upload videos to Docebo, then upload the caption file as a separate asset attached to the video. Docebo does not have a native vendor integration that automates the submission and delivery loop. For rollouts using Docebo, the submission workflow requires producers to submit the video file to the vendor through the vendor's portal (a separate browser action from the Docebo upload), wait for delivery, download the SRT, and return to Docebo to attach it. This multi-step workflow is a significant source of workflow friction and is the primary driver of submission rate decay in Docebo environments. The rollout mitigation for Docebo is to make the vendor portal submission as simple as possible — a URL pre-populated with the organisation's account and a drag-and-drop upload — and to set up a Docebo custom field that tracks caption status (submitted / in review / delivered / uploaded) so that the accessibility coordinator can monitor the pipeline without requiring producers to report status separately. Producer training in Docebo environments should spend disproportionate time on the submission and re-upload steps because these are the steps where producers most commonly introduce errors.

Cornerstone OnDemand

Cornerstone OnDemand supports external caption file attachment at the course material level, with SRT and VTT as the accepted formats for the integrated video player. The rollout challenge in Cornerstone environments is that the caption file attachment interface is not discoverable from the standard course-creation workflow — it is a secondary panel that producers must know to open after uploading the video file. Training in Cornerstone environments should explicitly demonstrate the caption attachment step with screen-recording instructions because producers who miss it during training consistently report that they could not find where to attach the file when they returned to it independently. Cornerstone LMS administrators should verify that the video player configuration is set to display captions by default (caption display is a player-level setting that overrides video-level settings in some Cornerstone configurations) before producer training begins.

TalentLMS

TalentLMS supports VTT caption file upload for videos hosted within the platform. The rollout workflow for TalentLMS is straightforward: producers upload video files, then use the "Add Subtitles" option in the video unit editor to attach the VTT file. The primary rollout friction in TalentLMS environments is format: some captioning vendors deliver SRT files by default, and producers may not know to request VTT format or to convert the SRT file before upload. TalentLMS does not accept SRT files natively in its video unit editor. The rollout training should address the format requirement explicitly — "request VTT format from the vendor, not SRT" — and provide a simple one-line conversion command or free browser-based converter link for producers who receive SRT files and need to convert them. Failure to address the format issue in training produces a predictable failure mode: the producer downloads the SRT file, attempts to upload it to TalentLMS, receives a format error, and abandons the upload, creating an uncaptioned published video and a support request to the accessibility coordinator.

Panopto

Panopto is widely used in university and corporate L&D environments for lecture capture and recording. Panopto has native ASR captioning built into the platform, which creates a rollout complexity: producers who have been using Panopto's built-in ASR captions (which are typically 85–92% accurate for non-technical content) need to understand why the captioning programme replaces them with vendor-supplied captions rather than relying on the native ASR. The training in Panopto environments should address this transition explicitly: built-in ASR is being replaced by vendor ASR with a domain-specific glossary, which produces significantly higher accuracy on domain vocabulary, and the caption file workflow is: request vendor caption, receive SRT or VTT, use the Panopto caption import function (Settings → Captions → Import Captions) to import the vendor file, which replaces the built-in ASR transcript. Producers who have been relying on built-in ASR may resist the new workflow on the grounds that it is extra work for content that already has captions; the training should address this resistance directly with a before/after accuracy comparison on domain vocabulary.

Workday Learning

Workday Learning supports caption files through the video content hosting integration (Workday does not natively host video; it integrates with Kaltura, Vimeo, or other video hosting platforms). The rollout workflow for Workday Learning depends entirely on which video hosting platform is integrated. If Workday is integrated with Kaltura, the Kaltura REACH workflow applies and caption submission is handled at the Kaltura level before the content is published to Workday. If Workday is integrated with Vimeo or another platform, the caption file must be uploaded to that platform first and then surfaced through the Workday integration. Producer training in Workday Learning environments should clarify the caption workflow at the video hosting platform level, not at the Workday level, and ensure that producers understand that Workday is the delivery surface, not the captioning workflow platform.

Absorb LMS

Absorb LMS supports SRT caption file attachment for video lessons through the lesson editor. The rollout workflow is similar to TalentLMS: producers upload video, then attach the SRT file via the subtitle option in the lesson editor. Absorb does not have a native captioning vendor integration, so the multi-step submission workflow applies. The primary rollout friction in Absorb environments is the lesson vs. course structure: in Absorb, captions are attached at the lesson level (the individual video), not at the course level. Producers who create multi-video courses may attempt to attach captions at the course level and be confused when the captions do not appear. Training in Absorb environments should demonstrate the lesson-level caption attachment process explicitly and confirm that producers understand which level of the content hierarchy the caption file belongs to.

WorkRamp

WorkRamp is primarily a sales enablement platform with L&D functionality, and its caption support is more limited than dedicated L&D platforms. WorkRamp supports closed captions for videos hosted externally (Loom, Vidyard, YouTube, Vimeo) through those platforms' native caption support; videos uploaded directly to WorkRamp may have limited native caption display options. The rollout for WorkRamp environments typically requires a decision about video hosting: should videos be hosted in WorkRamp natively, or hosted in an external platform (Vimeo, Wistia) and embedded in WorkRamp? The caption workflow is more reliable and LMS-independent if videos are hosted in the external platform, which handles the caption file and display, and embedded in WorkRamp for the learning experience. The rollout training for WorkRamp environments should clarify this hosting decision and ensure producers understand the caption workflow for their specific configuration.

Brightspace (D2L)

Brightspace by D2L supports caption files through its media management tool and through integration with Kaltura or Ensemble Video. For organisations using Brightspace with integrated video hosting, the caption workflow follows the video hosting platform's process. For organisations using Brightspace's native media tool, captions can be uploaded via the Media Library as SRT files associated with specific video assets. The rollout friction in Brightspace environments tends to appear in organisations using multiple content authoring tools (SCORM packages built in Articulate or Lectora, plus native Brightspace content, plus video assets in the Media Library), where producers may be confused about which content types are covered by the captioning programme and where the caption file is managed for each type. Training in Brightspace environments should map the captioning workflow specifically for each content type the organisation uses and clarify which types are in scope for the captioning programme and which caption workflow applies to each.

Canvas LMS

Canvas LMS supports caption files through its native media recorder and through integration with Kaltura, Arc, or Studio. Canvas's native HTML5 video player displays SRT and VTT caption files attached to video assets in the Rich Content Editor. The rollout challenge in Canvas environments is that caption file attachment varies significantly depending on whether the video is uploaded natively to Canvas, hosted in Studio/Arc, or hosted in Kaltura and embedded in Canvas. Producers who use multiple video pathways in Canvas may have inconsistent workflows for each. Training in Canvas environments should map each video pathway explicitly: native Canvas video → attach caption in RCE; Kaltura/Studio embed → handle caption at the Kaltura/Studio level; external embed (YouTube, Vimeo) → handle caption at the external platform level. The accessibility coordinator should confirm with the LMS administrator which video pathways are in use before designing the training for Canvas environments.

The eight change management metrics

The live-period monitoring produces eight metrics that together indicate whether the rollout is succeeding. The metrics fall into two categories: leading indicators (which predict future compliance before it shows up in the caption library) and lagging indicators (which measure the state of the caption library at a point in time). Leading indicators should be reviewed monthly; lagging indicators are more appropriate for quarterly reporting to leadership.

Leading indicators

1. Submission rate. The percentage of new videos published that are submitted for captioning within five business days of publication. Target: 95%+ by month 3, 98%+ by month 6. This is the primary leading indicator of programme health. A submission rate below 90% that persists for two consecutive months indicates a structural workflow or training problem, not individual producer non-compliance. Investigate at the department level before escalating to individual follow-up.

2. On-time submission rate. The percentage of submissions that arrive at the vendor within the submission window (typically five business days of video publication). Late submissions indicate that producers are submitting but late — often because they have been reminded by a deadline, not because the caption step is integrated into their normal publishing workflow. On-time rate below 85% indicates that the submission step is being deferred, which is the precursor to non-submission.

3. Review completion rate. The percentage of delivered caption files that are reviewed and marked as approved before being uploaded to the LMS. Target: 100%. A review completion rate below 100% means that some producers are uploading the vendor-delivered file without reviewing it, which means that vendor errors are going directly into the published LMS content. This is a training and accountability issue: the review step must be a required step, not an optional one, and the monitoring mechanism must track review completion, not just submission completion.

4. First-time review pass rate. The percentage of caption files that pass the producer review without being returned to the vendor for revision. Target: 90%+ by month 3. A high revision rate in the first weeks is normal — the glossary is new and domain vocabulary accuracy is still improving. A revision rate that remains above 15% after 3 months indicates a persistent glossary gap or vendor-side systematic error that requires a vendor quality conversation, not a producer training intervention.

Lagging indicators

5. Producer compliance rate. The percentage of individual producers who have maintained a submission rate at or above the threshold for the review period. Report by department and by individual. Producer compliance rate is a lagging indicator of the training and monitoring effectiveness. A high organisation-wide submission rate with a low producer compliance rate (many producers submitting zero or one video but a few heavy producers submitting many captioned videos) masks a programme that is compliant in aggregate but non-compliant in practice for a large percentage of the producer population.

6. Exception rate. The percentage of submitted videos for which a formal exception has been approved. Target: below 5%. An exception rate above 10% indicates either that the content gate is too broad (capturing too much content that genuinely should be exempt) or that the exception procedure is being used as a workaround by producers who want to avoid the captioning step. Review exception log distributions by category and by producer to distinguish the two causes.

7. Correction volume per producer per quarter. The average number of caption corrections (errors caught in producer review or in the accessibility coordinator's spot-check) attributable to each producer's submitted videos. High correction volume from a specific producer's content indicates a systematic domain vocabulary gap in that producer's content type — a signal to update the glossary with terms from that domain. High correction volume across all producers indicates a vendor accuracy problem or a systematic glossary gap. The caption QA methodology covers the error taxonomy and correction classification that makes this metric meaningful; tracking correction volume without error classification produces counts that are hard to act on.

8. Glossary growth rate. New glossary terms added per quarter. Target: varies by content domain, but programmes that are actively maintaining the glossary add 20–50 terms per quarter in their first year as producers encounter and report domain vocabulary errors. A glossary growth rate of zero in a quarter where correction volume is high indicates that the producer-to-glossary feedback loop is not functioning — corrections are being made but not fed back into the glossary, so the same errors will recur. The caption feedback loop post covers the glossary maintenance architecture that makes this feedback loop automatic rather than manual.

Eight failure modes

The failure modes below are the patterns that cause captioning programme rollouts to stall after adequate technical preparation. They are structural failures, not individual failures — their root cause is in the programme design or rollout sequencing, not in producer non-compliance or vendor underperformance.

Failure mode 1: Launching without a finalised governance policy. Going into producer training before the governance policy is approved and published means that producers are trained on a workflow that references a document that does not yet exist in its final form. When the policy is later finalised and differs from what producers were trained on — even in minor ways — producers experience the inconsistency as evidence that the programme is not well-organised, which reduces their confidence in it. The policy must be approved, signed, and published before the first producer training session. See the governance policy template post for the document structure.

Failure mode 2: Not completing a baseline audit before launch. Without a baseline audit, the rollout has no starting point for measuring progress and no producer-level compliance heatmap to sequence the training. More practically, without a baseline audit, the first monthly submission report will be hard to interpret: is the 85% submission rate a good result relative to the pre-programme baseline, or is it a decline from a historical rate of 92%? The audit also surfaces the existing uncaptioned content backlog, which must be addressed in the training — producers who ask about the backlog and receive a vague answer lose confidence in the programme's seriousness. See the enterprise LMS caption audit methodology post.

Failure mode 3: Training all producers simultaneously before the workflow is stable. All-hands producer training before the LMS integration is fully tested and the glossary is seeded produces a cohort of trained producers who encounter a workflow that does not work correctly in their first week of operation. The trust deficit from early workflow failures — caption files delivered in the wrong format, submission confirmations that do not arrive, review interfaces that display errors — is very hard to recover from. Producers who experience workflow failures in their first week are significantly less likely to maintain the workflow habit than producers whose first experience is smooth. Train the champion cohort first, verify the end-to-end workflow is stable with their production submissions, then train the full producer population.

Failure mode 4: No champion producers in high-volume departments. Without department-level champions, the accessibility coordinator is the only support resource for all producers in the organisation. High-volume production departments (L&D production team, Sales Enablement, Compliance training) will generate more workflow questions than a single accessibility coordinator can support in the first 8–12 weeks. The absence of champions in these departments creates a support bottleneck that causes producers to resolve workflow uncertainty on their own — typically by skipping the step — rather than waiting for a response. Champions are not a luxury in high-volume departments; they are a capacity requirement for maintaining support responsiveness in the live period.

Failure mode 5: No escalation path for non-compliance. A monitoring mechanism without an escalation path is not a monitoring mechanism — it is a reporting function that has no operational consequence. The monthly submission report must connect to an escalation path: non-compliance below threshold for one month → supportive follow-up from the accessibility coordinator; non-compliance below threshold for two consecutive months → escalation to the producer's manager; continued non-compliance after manager escalation → escalation to the L&D director. The escalation path does not need to be punitive to be effective; it needs to be real and consistently applied. Producers who know that non-compliance produces a predictable escalation sequence will prioritise compliance under time pressure at a much higher rate than producers in programmes where the monitoring report is produced but not acted on.

Failure mode 6: Conflating "reviewed" with "accurate." Producer review of caption files is a quality gate, not an accuracy guarantee. A producer who reviews a 30-minute caption file in 10 minutes will not catch every error; they will catch systematic errors in domain vocabulary, speaker identification mistakes, and obvious formatting failures. Treating producer review as equivalent to full accuracy verification — holding producers responsible for errors that passed their review — creates a review burden that is unsustainable and that producers will eventually stop performing. The division of responsibility is clear: the vendor is responsible for technical accuracy to the contracted SLA, the producer is responsible for domain-vocabulary review, and the accessibility coordinator is responsible for systematic QA spot-checks that verify the combined output. Conflating these three responsibilities produces either an impossible producer review standard or a QA gap where no one is responsible for the errors that fall between them. See the caption QA methodology post for the role-based responsibility division.

Failure mode 7: No monitoring cadence in the first 90 days. The first 90 days of live operation are the highest-risk period for submission rate decay. Programmes that establish a monthly monitoring cadence from day one — running the first report at the end of month one, following up within five business days — experience significantly lower decay than programmes that start monitoring at month three or six. By month three, the submission rate has already established itself as a habit or a non-habit for most producers; the window for easy behavioural intervention is largely closed. Starting the monitoring cadence at day one, even when the programme is small and the data is sparse, is more effective than starting a more sophisticated monitoring programme later.

Failure mode 8: Treating rollout as a one-time event. The rollout is not complete when all producers have been trained. The programme enters steady-state operation when the monitoring cadence is running, the exception log is being maintained, the glossary is being updated from correction signals, and the producer population is stable. In practice, steady-state operation is a continuous process because the producer population changes: new hires join who have not been trained, producers change roles and take on video responsibilities they did not have during the original training, and content types evolve in ways that require governance policy updates. A programme that does not have a new-hire onboarding process for captioning (training within 30 days of hire, included in the L&D onboarding checklist) will see its producer compliance rate decay as the trained cohort turns over. The rollout plan should include a process for new-hire onboarding from launch day, not as an afterthought after the initial training is complete.

Frequently asked questions