Caption programme maturity model for L&D teams: five levels from reactive accommodation to strategic content quality infrastructure — how to assess your level and what moves you to the next

Why a maturity model is different from a self-assessment or a compliance checklist

The caption programme self-assessment generates a score across eight domains and identifies specific gaps. That is a diagnosis tool. A maturity model is a planning and communication tool — it characterises where a programme is positioned on a recognised developmental scale and specifies what the path forward looks like.

The distinction matters for two reasons. First, a score is hard to communicate to a leadership audience. “We scored 48 out of 80” means nothing to a CLO who has no reference point for what 48 represents. “We are at Level 2 (Developing) and need two specific investments to reach Level 3 (Established), which is the level where a regulatory review would find documentation rather than gaps” — that is a sentence a CLO can act on.

Second, a gap list without a trajectory is demoralising and hard to prioritise. If the self-assessment identifies 23 gaps across eight domains, the L&D team faces 23 parallel action items with no indication of which matter most or in what order. The maturity model collapses the gap list into a sequence: “These five things move you from Level 2 to Level 3. Once those are done, these four things move you from Level 3 to Level 4.” The sequencing is the planning value.

The caption compliance programme build guide covers how to stand up a programme from scratch. This guide assumes a programme already exists in some form and focuses on how to characterise its maturity, plan its development, and communicate that plan to decision-makers. The two guides are complementary — the build guide is for programmes at Level 1 moving to Level 2; this guide is for programmes at Level 2 and above planning their path to Level 3 and beyond.

Level 1: Reactive

A Reactive caption programme treats captioning as an individual accommodation, not a programme. Captions are produced only when a specific employee requests them. Without a request, content is published without captions as a default. There is no written policy requiring captions on training content, no designated caption production workflow, and no vendor relationship maintained outside of individual accommodation fulfilment.

Caption decisions at Level 1 are made by whoever receives the accommodation request — typically the L&D director, the HR manager, or the instructional designer closest to the content — not by a documented process. Budget for captioning is handled as a miscellaneous accommodation expense under HR, not as a caption programme line item. Back-catalogue content is entirely uncaptioned. New content is published without captions as a default.

The LMS auto-caption feature, if enabled, is typically counted as “captioning” at Level 1 — the LMS metadata shows captions as present, and the organisation believes it has addressed the requirement. As the WCAG 2.1 AA captioning requirements make clear, a caption track is not the same as equivalent captions. Auto-generated captions at 80–90% accuracy do not satisfy WCAG 2.1 SC 1.2.2’s equivalence requirement for content with specialised vocabulary. A Level 1 organisation that has enabled LMS auto-captions is not at Level 2; it is at Level 1 with a false indicator of compliance.

Level 1 regulatory position

High exposure. An OCR complaint investigation at a Level 1 organisation finds: no written policy, no systematic coverage, no vendor documentation, no accuracy records, no accommodation log. The response to an investigation is improvised documentation assembled under pressure, which regulatory investigators recognise immediately. Many Reactive organisations are technically in violation of WCAG 2.1 AA for the majority of their content continuously — the only reason this does not produce immediate regulatory findings is that complaint volumes are lower than the non-compliance rate and investigations take time.

Who is at Level 1

Most L&D teams at organisations with fewer than 100 employees and no formal accessibility function. Organisations that have not yet received an accommodation request for captioning. Organisations that believe LMS auto-captions satisfy their accessibility obligation. And, importantly, organisations that have received one accommodation request, handled it case-by-case, and returned to the default of unprompted non-captioning once the specific request was resolved.

Level 2: Developing

A Developing caption programme has moved beyond pure reactive accommodation but has not yet built the documentation layer that converts operational practice into defensible compliance. Some proactive captioning exists: new content may be captioned in some programmes but not others. A vendor relationship or SaaS self-service tool is in use. A caption requirement for new content exists as an informal norm — “we caption everything” is the L&D team’s self-description — but the requirement is not in a written policy that applies to all content producers and all content types.

The critical gap at Level 2 is documentation. The operations are approximately right; the written evidence that the operations are being performed systematically is absent. Accuracy is not measured by a defined methodology — the team relies on the vendor’s accuracy claim or on informal viewing to assess whether captions are acceptable. The vendor contract does not include an accuracy SLA, audit rights, or a reference transcript retention clause. Accommodation requests are handled, but the log is informal, incomplete, or in one person’s files rather than a centralised system. The back-catalogue has not been formally audited, and no project plan for remediation exists.

Live training at Level 2 is handled ad hoc. A CART vendor may be known but not under contract. LMS course listings for live training do not state an accommodation request deadline. Facilitators have no written protocol for what to do if CART is not working. Post-event accessible alternatives are provided inconsistently.

Level 2 regulatory position

Moderate exposure. A Developing organisation has some captions, which improves its position relative to Reactive — but the absence of a written policy, accuracy measurement, and vendor accountability means it cannot demonstrate systematic compliance. An OCR complaint investigation at a Level 2 organisation will find coverage gaps, inconsistent accuracy, missing vendor documentation, and an accommodation log that is incomplete or not retrievable in the required format. The response is improvised in the places where documentation should be but is not.

Who is at Level 2

The majority of mid-market L&D teams that have been captioning for one to three years. Teams that started captioning after a first accommodation request or after ADA Title II deadline awareness. Teams that have a vendor relationship but negotiated the contract on standard vendor terms without adding accuracy SLA or audit rights. Teams that self-assess as “compliant” because they produce captions, without evaluating whether those captions are documented, measured, and defensible.

Level 2 is the most common state for L&D caption programmes in the 50–500-employee organisation segment. The caption compliance reporting to leadership guide notes that most L&D teams at this level cannot answer the CLO’s questions — “what is our coverage rate?” “what accuracy are we getting?” “can we pass an audit?” — because the measurement infrastructure to answer those questions is not in place.

Level 3: Established

An Established caption programme has the documentation layer that Level 2 lacks. A written, organisation-wide caption policy is in force and applies to all content producers — including instructional designers, subject-matter experts recording screencasts, external vendors producing content for delivery in the LMS, and any AI video production in use. The policy is not a draft; it is signed by a responsible leader and distributed to all content producers.

Accuracy is measured using a documented methodology: reference transcript comparison, with a 99%+ accuracy target written into the policy. This target is not just in the vendor contract — it is in the internal policy, so the standard applies regardless of whether a vendor or an internal team produces the captions. The vendor contract includes a 99%+ accuracy commitment with a defined methodology, a reference transcript retention clause of at least 12 months, and a basic remediation clause that specifies the vendor’s obligation when output falls below the accuracy SLA.

The accommodation request process is documented: a designated intake contact or monitored inbox, a log with required fields (date received, employee and department, accommodation type, confirmation date, arrangement made), and a written confirmation process for every request. The log is in a centralised system, not in a single person’s files. A back-catalogue audit has been completed and the results are documented — not an estimate, but an inventory with caption status for each item. A project plan for remediating the highest-risk uncaptioned content exists, with a named owner and a target completion date.

Live training at Level 3 has minimum infrastructure: at least one CART vendor on the approved vendor list, and a published accommodation request deadline in LMS course listings for live training courses. The CART vendor may not be under a full SLA contract, and the live training protocol may not cover the full feed-failure decision tree — those are Level 4 requirements. But the deadline is published and the vendor is identified.

Level 3 regulatory position

Low exposure. An Established organisation can demonstrate systematic compliance: written policy, new-content coverage, accuracy measurement methodology and target, vendor commitment, accommodation log, and a remediation project plan. An OCR complaint investigation will find documentation. Gaps at Level 3 are typically in depth rather than absence — vendor contract terms may be thin (accuracy commitment present but audit rights absent), live training protocols may be minimal (vendor identified but no SLA, no feed-failure protocol), and back-catalogue remediation may be in progress but not complete. These gaps are defensible in context; the absence of any documentation is not.

Who is at Level 3

L&D teams that have built their programme deliberately using the compliance programme build guide or equivalent structured approach. Organisations with a dedicated accessibility coordinator who has had the time and authority to build the documentation layer. L&D teams that have been through one OCR complaint or formal accessibility audit and used the experience to build structure. Teams that have adopted a written caption programme governance policy.

Level 3 is the realistic first-year target for a mid-market L&D team that is intentionally building its caption programme. It requires no additional budget beyond internal time — the policy writing, vendor contract amendment, accommodation log setup, and back-catalogue audit are documentation investments. The back-catalogue remediation plan may require future budget, but writing the plan requires only the audit output.

Level 4: Proficient

A Proficient caption programme has full operational and accountability infrastructure across all eight programme domains. All Level 3 requirements are met and the documentation depth at each domain is complete. The vendor contract includes the full accountability stack: a 99%+ accuracy SLA with a defined measurement methodology (not a vague accuracy commitment), explicit audit rights giving the organisation the right to request vendor processing logs and reference transcripts on reasonable notice, a reference transcript retention clause of at least three years (consistent with EEOC and OCR investigation timelines), and a remediation clause with a defined SLA and formal notice process.

Vendor accuracy has been independently verified against the organisation’s own content — not by reviewing the vendor’s internal QA report, but by applying reference transcript comparison to a sample of content the vendor has processed for this organisation. The verification is documented and on file. This is the step that most Level 3 organisations have not taken, and it is the step that closes the gap between vendor accuracy claims and actual accuracy on domain-specific content.

Live training at Level 4 has complete operational infrastructure. A CART vendor is under contract with a written SLA covering lead time, captioner qualification standards, accuracy commitment for live delivery, and post-event transcript availability. All LMS course listings for live training include the accommodation request deadline, calculated accurately against the CART vendor’s actual lead time plus the organisation’s processing buffer. Facilitators receive a written feed-failure decision tree before any session with CART deployed — a document that covers at minimum display-side configuration failure, feed-stopped failure, audio quality degradation, and provider unreachable. Post-event accessible alternatives (captioned recording or verbatim CART transcript) are available within a written, communicated delivery timeline for all live session participants, not just those who requested an accommodation.

Back-catalogue remediation at Level 4 is funded and actively in progress. Budget has been allocated at a volume and per-minute cost that reflects the actual audited content volume (not an estimate) and the actual vendor price for reviewed, 99%+ accuracy captioning (not the price of auto-captioning with no review). The remediation update-lifecycle process is in place: when a captioned video is updated, the caption file is updated or reprocessed rather than the old file applied to new audio without review.

Caption compliance is reported to a leadership audience on a regular cadence. The leadership reporting guide covers how to structure this report. At Level 4, the report includes: coverage rate for new content, accuracy measurement results and trend, vendor SLA performance, accommodation request volume and resolution time, and back-catalogue remediation project status. The report exists as a document, not just as an informal update at a team meeting.

Level 4 regulatory position

Very low exposure. A Proficient organisation can produce documentation on request for any domain of caption compliance. Coverage policy, vendor accuracy records with independent verification, audit rights documentation, reference transcript retention records, accommodation log with all required fields, live training operational procedures, and back-catalogue remediation project status — all are on file and retrievable. An OCR complaint investigation at a Level 4 organisation is a documentation production exercise, not an investigation into programme gaps. The residual gap is the back-catalogue remediation in progress, which is being addressed with a funded, documented project plan.

Who is at Level 4

Large L&D teams with three or more staff members and a dedicated accessibility lead who has time to manage programme depth rather than just programme breadth. Organisations that have been through a formal Section 504 review or institutional accessibility audit. Public universities post-ADA Title II enforcement (April 2026) that have used the deadline as a forcing function to build full programme infrastructure. Organisations in regulated industries — healthcare, financial services, legal — where the compliance training captioning stakes are high enough that legal has driven vendor contract quality.

Level 5: Strategic

A Strategic caption programme is not primarily defined by what it does differently from a Proficient programme on any individual domain — a Proficient programme already has excellent coverage, accuracy, vendor accountability, and live training infrastructure. Strategic is defined by how the programme is integrated into governance, succession, and organisational content acquisition.

At Level 5, captioning is integrated into the content production workflow as a standard quality requirement alongside production timelines, learner completion rates, and content review cadences. Caption accuracy is tracked as a content quality metric and reported in the same dashboard that tracks course completion and learner satisfaction — not as an accessibility sidebar but as a core content quality indicator. The caption glossary is maintained as part of the product or clinical terminology update process: when a product is renamed, a drug name changes, or a regulatory acronym is updated, the glossary update is scheduled as part of the content change workflow, not discovered after the fact during a QA review.

Vendor performance is reviewed semi-annually with documented output: performance against each SLA element (accuracy, turnaround time, reference transcript availability), any accuracy drift identified by independent verification, and a contract renewal readiness assessment. The vendor portfolio is actively managed rather than assumed to be stable.

Caption compliance is integrated into new content acquisition contracts. Any externally produced content — purchased eLearning modules, vendor-produced compliance training, commissioned instructional video — must be delivered with WCAG 2.1 AA captions as a contract deliverable, not a post-delivery add-on. The organisation’s legal team has reviewed the standard caption deliverable clause and approved it for use in all content acquisition agreements.

The programme has a succession plan. All processes are documented to the degree that a new hire could maintain the programme from the documentation alone — vendor contact information, contract terms and renewal dates, accommodation log field requirements, the CART vendor lead time and operational decision tree, the glossary maintenance schedule, the vendor verification methodology and sampling protocol, the leadership reporting template, and the back-catalogue priority framework. The programme does not depend on any individual who could leave, be promoted, or be out on leave.

Caption programme data feeds into board-level or governance-level accessibility reporting. For organisations with a board accessibility committee, a public accessibility statement, or an institutional DEI reporting obligation, the caption programme metrics — coverage rate, accuracy trend, accommodation request volume, remediation progress — appear in that report as measurable programme outcomes. This is the level at which captioning becomes a demonstrable institutional commitment rather than an operational fact.

Level 5 regulatory position

Minimal exposure. A Strategic organisation can demonstrate not just compliance but a programme infrastructure designed to maintain compliance through staff turnover, vendor transitions, technology changes, and regulatory updates. Regulators who have reviewed many organisations’ caption programmes recognise the difference between a Level 4 programme that is compliant today and a Level 5 programme that has built the institutional infrastructure to remain compliant over time. A Strategic programme with a documented succession plan, board-level reporting, and external contract language is a model programme, not just a compliant one.

Who is at Level 5

Public universities with a dedicated accessibility office and full compliance team. Large healthcare systems with formal compliance team oversight of training content accessibility. Enterprise L&D organisations with ten or more team members and a formal DEI and accessibility governance infrastructure. Organisations where accessibility is a stated institutional value with executive sponsorship and board-level accountability.

Eight-domain maturity level assessment table

Use this table to cross-reference your programme’s state across the eight domains from the scored self-assessment with the five maturity levels. Your overall level is determined by the lowest-scoring domain — a programme that is at Level 4 in seven domains and Level 1 in Domain 4 (live training) is at Level 1 for live training purposes and cannot be described as a Level 4 programme without qualification.

Transition investments: what moves you from each level to the next

The following transition guides identify the specific investments required at each level change. Investments are ordered by priority within each transition: the first investment on each list has the highest leverage for regulatory exposure reduction. The caption programme budget planning guide covers cost estimation for each category of investment.

Reactive → Developing: minimum first investments

The three investments that move a programme from Level 1 to Level 2 are the minimum viable compliance infrastructure. None of them is a large budget commitment; all three are primarily time investments.

1. Identify a captioning tool or vendor and begin captioning new content proactively. This does not require a vendor RFP or a long procurement process. Even a SaaS self-service captioning tool that produces reviewed captions at 99%+ accuracy is sufficient to move from Reactive to Developing. The key threshold is systematic production of captions on new content without waiting for an accommodation request. Note: enabling LMS auto-captions does not satisfy this investment — auto-captions with no accuracy review are not equivalent to WCAG-compliant captions and should not be counted as proactive captioning infrastructure.
2. Designate a captioning intake point and create a basic accommodation request log. The intake point can be a monitored email inbox with a clear label. The log can be a shared spreadsheet. The minimum fields are: date received, employee name and team, accommodation type requested, date confirmed to employee, arrangement made. This log is the evidence that accommodation requests are being handled and that the programme is not purely reactive.
3. Write a one-page caption requirement for new content and distribute it to all content producers. This does not need to be a full caption policy (that is a Level 2-to-3 requirement). A one-page document that states “all new training video content must be captioned to WCAG 2.1 AA standards before publication in the LMS” with the name of the tool or vendor to use and the name of the person to contact with questions is sufficient to establish an organisational norm. Distribute it to all instructional designers, SMEs who record screencasts, and external contractors who produce video for the LMS.

What does NOT constitute a transition to Developing: Captioning one high-profile course while leaving the rest of the library uncaptioned. Getting a single accommodation request and handling it ad hoc, then returning to the default of no proactive captioning. Enabling LMS auto-captions and updating the LMS metadata to show “captions available.”

Developing → Established: the documentation inflection point

The transition from Level 2 to Level 3 is the most important transition in the model. It is the point where operational practice becomes defensible compliance. It is also the transition that most mid-market L&D teams delay longest — because the documentation work feels less urgent than content production, and because it requires structured time rather than budget. The five investments that make this transition:

1. Write and publish a formal caption policy that covers all content types. This is the single highest-ROI documentation investment in the caption programme. The caption programme governance policy template provides a starting structure. The policy must: apply to all content types (video, audio-only, screen recording, AI-generated video, externally produced content); name the owner; state the accuracy standard (99%+); specify the enforcement mechanism; and be signed by a responsible leader. Distribute it to all content producers and document the distribution. An undistributed policy is not a policy for compliance purposes.
2. Establish a reference transcript comparison accuracy methodology. The caption QA methodology guide covers how to implement this. At minimum: define how accuracy is measured (errors per 100 words using a reference transcript comparison), specify the sampling protocol (what percentage of vendor output is verified and how), state where results are filed, and document the first verification run as the baseline for future comparisons. This can be done in one to two weeks for a team that is already producing captions — the tool is already in use; the verification methodology is the missing documentation layer.
3. Amend the vendor contract to include a 99%+ accuracy SLA, a reference transcript retention clause, and a basic remediation clause. The vendor SLA and contract review checklist specifies what each clause should contain. Most vendors will accept a contract amendment adding these terms without a price increase — the accuracy commitment reflects what the vendor is already trying to deliver, and the retention clause does not require the vendor to do anything it is not already doing if it maintains reference transcripts as part of its QA process. The conversation is a letter or email from the L&D team to the vendor account manager; it rarely requires legal escalation to close.
4. Conduct a formal back-catalogue audit and write a project plan. The audit produces a per-item inventory of the video library with caption status for each item: captioned with reviewed accuracy, captioned with auto-captions only (does not count as WCAG-compliant), captioned with reviewed accuracy but file may be outdated, or not captioned. The project plan identifies the highest-risk uncaptioned items (compliance-mandatory content first, then high-traffic recent content, then older content), assigns a named owner, states a realistic target completion date, and includes a cost estimate at the vendor’s per-minute rate for reviewed captioning. The cost estimate is often a surprise; that is the point — discovering what the remediation costs is better than leaving it undiscovered until a regulatory review.
5. Publish a CART vendor lead time and accommodation request deadline in live course LMS listings. This one change addresses one of the most common accommodation failure modes for L&D teams. Even if only one or two live courses are delivered per quarter, listing the accommodation deadline and intake contact converts the ad hoc process into a documented one. The live course accessibility statement guide covers the specific language for LMS listings.

Timeline: Three to six months for a team with two to four L&D staff if the back-catalogue audit is scoped appropriately. The policy and contract amendment can be completed in four to eight weeks. The audit timeline depends on library size — a 50-video library can be audited in one to two weeks; a 500-video library requires a structured project.

Established → Proficient: operationalisation depth

The transition from Level 3 to Level 4 builds the operational depth that Level 3 has in outline. Each investment at this level adds robustness to an existing foundation rather than building a new structure.

1. Upgrade vendor contract to full accountability terms. The Level 3 vendor contract has an accuracy commitment. The Level 4 contract adds audit rights, extends the retention period to three years, and adds a remediation SLA with a formal notice process. The vendor audit rights and examination evidence guide specifies exactly what audit rights language should include. This is the contract term that matters most when a regulatory examination asks for vendor processing records.
2. Conduct independent accuracy verification on vendor output against organisation content. Take the reference transcript comparison methodology established at Level 3 and apply it to a sample of content the vendor has produced for this organisation — not the vendor’s test material, but the organisation’s actual training content. If the organisation produces medical training, verify accuracy on a sample of medical content. If it produces engineering onboarding, verify on engineering vocabulary content. Document the result. If accuracy falls below 99%+ on domain-specific content, the next step is to implement a glossary at the transcription stage to close the gap.
3. Build complete live training operational infrastructure. Three components missing at Level 3 that Level 4 adds: (a) a CART vendor under contract with a full SLA covering captioner qualifications and accuracy commitment, not just a vendor on an approved list; (b) a written feed-failure decision tree for facilitators, distributed before any session with CART deployed; (c) post-event accessible alternatives with a written, communicated delivery timeline for all live session participants. The live training audit in the real-time CART captioning guide covers each component in detail.
4. Fund and launch the back-catalogue remediation project. The Level 3 project plan exists with a named owner and target date. Level 4 adds budget allocation and active project progress. This is typically the largest discrete investment in the transition to Level 4 — remediation at 99%+ reviewed accuracy costs significantly more per minute than auto-captioning, and the library audit at Level 3 has now quantified the full scope. The ROI framing for this investment is covered in the caption ROI framing guide for finance audiences.
5. Establish a quarterly caption compliance report to leadership. Not a verbal update at a team meeting, but a document with standard sections: coverage rate, accuracy measurement results, vendor SLA performance, accommodation request volume, and back-catalogue remediation project status. The leadership reporting guide provides a reporting template. This report does two things: it makes the programme visible to leadership, and it creates a paper trail of programme status that is valuable if a regulatory investigation begins.

Proficient → Strategic: governance, succession, and integration

The transition from Level 4 to Level 5 is qualitatively different from the previous transitions. Levels 1–4 are about building the programme’s operational and documentation infrastructure. Level 5 is about integrating that infrastructure into the organisation’s governance, succession planning, and external contracting. The investments at this level are not primarily accessibility investments; they are organisational maturity investments that happen to apply to the caption programme.

1. Write programme succession documentation. This is the highest-leverage structural investment at Level 5. Documentation should reach the point where a new hire, with no prior knowledge of the caption programme, could maintain the programme from the documentation alone. Required elements: vendor contact information and contract summary with renewal dates; accommodation log system access and field guide; CART vendor lead time and operational decision tree; glossary maintenance schedule and term update process; vendor verification methodology and sampling protocol; leadership reporting template and distribution list; back-catalogue priority framework and current project status; caption policy and its version history. This documentation should be in a location that is accessible to the full L&D team, not in a single person’s files.
2. Integrate caption deliverables into new content acquisition contracts. Any externally produced content — purchased eLearning modules, vendor-produced compliance training, commissioned video, AI-generated content from a third-party platform — must be delivered with WCAG 2.1 AA captions as a contract deliverable. This typically means adding one clause to the standard content acquisition agreement specifying that captions must meet the 99%+ accuracy standard, be delivered in the specified format (SRT or VTT), and be reviewed against the organisation’s glossary before delivery. Legal should review the clause. Once in the standard contract, the organisation stops receiving uncaptioned external content without having to negotiate it separately for each acquisition.
3. Establish semi-annual vendor performance reviews. A formal review cadence — not just the annual review at contract renewal, but a mid-year performance check — catches accuracy drift, staffing changes at the vendor, and SLA compliance before they become programme problems. The review should produce a one-page performance summary: accuracy against SLA (by content type), turnaround time against SLA, reference transcript availability on request, and any issues raised and resolved in the period. This document is also useful input for the contract renewal conversation.
4. Integrate caption programme metrics into board-level or governance-level accessibility reporting. For organisations with a board diversity, equity, and inclusion committee; an institutional accessibility compliance report; or a mandatory public accessibility statement, caption programme metrics should appear as measurable commitments with tracked outcomes. Coverage rate, accuracy trend, accommodation request volume and resolution rate, and back-catalogue remediation progress are the minimum metrics to include. This integration requires coordination with the legal, DEI, or compliance teams that produce governance-level reports — it is not an L&D team unilateral decision.

Budget language for each level: how to talk to a CLO

The caption programme change management guide covers how to build organisational support for the caption programme. This section is specifically about how to frame the investment ask at each level in language that a CLO or VP People can evaluate and approve without needing to understand WCAG technical requirements.

Level 1 → 2 ask

“We currently caption training content only when an individual employee requests it. We have no systematic coverage, no vendor relationship, and no documentation of our accommodation responses. We are asking for permission to select a captioning tool at [estimated annual cost] and four to six weeks of L&D team time to write a one-page captioning requirement and set up an accommodation log. Without this, we have no documented response to our next accommodation request and no record of our compliance state if we receive a regulatory inquiry. This is the minimum investment to be non-reactive.”

Level 2 → 3 ask

“We caption new training content regularly and have a vendor relationship. What we do not have is documentation: a written policy, a vendor contract with accuracy terms, an accommodation log that can be produced on request, and a back-catalogue inventory. These are internal time investments — approximately six to eight weeks of structured work. The back-catalogue audit will reveal the volume of uncaptioned legacy content and what remediation would cost, but the audit itself does not commit us to remediation budget. What it does is tell us what we owe and put us in a position to plan the investment. The risk of not doing this documentation work is that if we receive an OCR complaint, we cannot demonstrate systematic compliance — we can only demonstrate that we have some captions. That is not enough to close an investigation without findings.”

Level 3 → 4 ask

“We have a documented caption programme: policy, accuracy standard, vendor contract, accommodation log, and a remediation project plan. We are asking for three investments to reach the level where a regulatory review would find a complete programme, not just a documented one. First, a vendor contract amendment to add audit rights and extended transcript retention — no additional annual cost. Second, budget for back-catalogue remediation: the audit quantified [volume] hours of uncaptioned content at [vendor price per minute], totalling approximately [$X]. This is a one-time investment with a completion date of [date]. Third, a funded CART vendor contract for live training at approximately [$Y per year]. The ROI case is insurance: a single OCR investigation with a remediation finding typically costs $30,000–$100,000 in legal and remediation fees. The prevention cost is less.”

Level 4 → 5 ask

“We have a complete, documented, defensible caption programme. This investment is about durability, not compliance catch-up. We are asking for: legal review of our standard content acquisition contract to add caption deliverable language (one-time cost); succession documentation time (four to six weeks of L&D team time); and integration of caption programme metrics into the institutional accessibility report (coordination time across teams). This converts our programme from one that is compliant because of the people currently in the roles to one that is compliant because of the documented systems. It means the programme holds through staff transitions, technology changes, and regulatory evolution without requiring reconstruction.”

Using the maturity model in your annual programme review

The caption programme annual review guide covers the full review process. The maturity model adds a strategic layer to the annual review that the operational review alone does not provide.

At the annual review, run the scored self-assessment to get current domain scores. Then cross-reference those scores against the domain table in this guide to determine the current maturity level per domain. The overall programme level is the lowest domain level. Report that level in the annual review document alongside the self-assessment score.

The year-over-year comparison becomes the strategic planning input. If the programme was at Level 2 overall last year and is at Level 3 overall this year, the transition to Level 4 is the next planning horizon. If the programme has been at Level 3 for two years and has not progressed, the annual review is the moment to diagnose why — is it budget (back-catalogue remediation not funded), time (vendor contract amendment not completed), or authority (policy not enforced for external contractors)?

The maturity level also becomes the headline of the leadership report. “We progressed from Level 2 (Developing) to Level 3 (Established) in 2026, meaning our caption programme is now documentable in an external review. Our goal for 2027 is Level 4 (Proficient), which requires [$X] in back-catalogue remediation and [$Y] in CART vendor infrastructure. Here is a two-year investment plan.” That framing gives the CLO a trajectory, not just a status update.

Eight failure modes that inflate self-assessed maturity level

1. The “we have captions” Level 2 plateau

The L&D team captions most new content, uses a vendor, and self-describes as “compliant.” The vendor contract has no accuracy SLA. The back-catalogue has never been audited. The accommodation log has three entries from 2024, all on paper. The team self-assesses as Level 3 (Established) because the operational pieces are in place. The documentation layer — the policy, the contract terms, the log fields, the audit — is absent. This is the most common maturity inflation error in mid-market L&D. Having captions is not the same as having a documented, defensible caption programme.

2. The policy-without-enforcement Level 3 illusion

A written caption policy exists. It was written two years ago. Three of the five content producers on the extended L&D team are contractors onboarded six months ago who have never seen it. Two compliance training videos were published last quarter without captions. The organisation self-assesses at Level 3 because “we have a policy” (two points on Q8.1 of the self-assessment). But Q8.3 (distribution) and Q8.4 (enforcement mechanism) score 0. A policy that exists but is not distributed or enforced is not a Level 3 governance infrastructure; it is a Level 2 operation with Level 3 paperwork.

3. The vendor accuracy claim problem

The vendor contract says 99.2% accuracy. The vendor’s QA report shows 99.2% accuracy on their test corpus of general business English. The organisation produces medical training content with clinical drug names, diagnostic criteria, and regulatory acronyms. The actual accuracy on that content is 91.4%, but it has never been independently measured. The programme believes it is at Level 3 (99%+ accuracy target in policy, vendor delivering on that target). It is at Level 2 in Domain 2 because independent verification of vendor accuracy on organisation-specific content has not been conducted. The claim and the reality are different.

4. The Level 3 documentation that lives in one person’s head

The accommodation request process works well. The accessibility coordinator handles everything and knows the CART vendor, the deadline to communicate, and the accommodation log format. The log is in the coordinator’s Google Drive. When the coordinator takes parental leave for three months, two accommodation requests arrive and are handled inconsistently — one incorrectly, one with a two-week delay. The process is at Level 3 operationally when the coordinator is present; it collapses to Level 1 when she is not. The issue is not the process; it is that the process is not documented in the succession-accessible form that Level 3 requires.

5. The Level 3-to-4 jump that skips live training

A programme has excellent coverage, documented accuracy measurement, a vendor with a strong contract, a funded remediation project, and a comprehensive accommodation log. Live training is “out of scope” — the team runs recorded eLearning only. A quarterly all-hands meeting is added to the training calendar as a live component. An accommodation request for CART captioning arrives. No CART vendor is on the approved list. No process exists. No LMS listing language has been written. The rest of the programme is at Level 4; Domain 4 (live training) is at Level 1. The programme cannot be described as Level 4 overall.

6. The annual review that measures coverage without measuring quality

The annual review reports: “94% of new content captioned, up from 89% last year.” Leadership approves continued investment. The accuracy trend is not reported — it has not been measured. The vendor was changed six months ago. No one has run the reference transcript comparison on any of the new vendor’s output. The accommodation log shows 12 requests this year, up from 7 last year, but the review does not evaluate whether they were resolved correctly. The programme looks like it is improving. In Domain 2, it may be regressing. Coverage measurement without quality measurement is not a Level 3 annual review; it is a Level 2 review that appears more thorough than it is.

7. The budget conversation without the maturity vocabulary

“We need more captioning budget” is a Level 1 ask. The CLO responds: “How much do we spend now? What are we getting for it? Are we compliant?” The L&D team cannot answer questions two and three because the measurement infrastructure is absent. The budget request fails. The same ask in maturity model language: “We are at Level 2 (Developing). The five investments that move us to Level 3 (Established) — the level where a regulatory review would find documentation, not gaps — are [list]. The internal time cost is [estimate]. The only new budget required is [vendor contract amendment, if any]. The compliance risk of staying at Level 2 is [specific exposure]. This is a six-week project, not a multi-year programme.” This is an answerable decision for a CLO. The maturity model is the difference between a budget conversation and a budget request.

8. The Level 4 programme with no governance succession plan

A proficient programme — written policy, vendor accountability, accommodation log, live training infrastructure, leadership reporting — is maintained entirely by one accessibility coordinator who has been in the role for four years and has the full programme in her head and files. She moves to a competitor. The incoming coordinator inherits a folder of documents but no guide to what they mean, why decisions were made, or where the active processes live. In six months, the vendor contract is up for renewal and auto-renewed on standard terms — losing the accuracy SLA and audit rights that were negotiated four years ago. The programme is at Level 2 within a year. The Level 4 status was personal, not structural. Level 5 would have prevented it.

Frequently asked questions

We started captioning six months ago. Could we already be at Level 3?

Yes. The transition to Level 3 is about documentation, not duration. If in those six months you wrote a caption policy, established a reference transcript accuracy methodology, amended your vendor contract to include a 99%+ SLA and basic retention clause, created an accommodation log with required fields, and conducted a back-catalogue audit with a project plan — you are at Level 3 regardless of how recently you started. The six-month timeline to reach Level 3 from a standing start is achievable for a team that has made the documentation work a deliberate priority. What takes longer is the back-catalogue remediation project, which is a Level 4 investment, not a Level 3 requirement.

Our self-assessment score was 48/80. What level does that map to?

The score alone does not map directly to a level — domain balance matters more than total score. A 48/80 with Domain 5 (vendor accountability) at 2/10 is a different programme than a 48/80 with Domain 5 at 8/10. Use the domain table in this guide to find where each domain score places the programme on the level descriptors. Then determine your overall level as the lowest domain level. It is common to find that six domains are at Level 3 and two domains (often Domain 4: live training and Domain 5: vendor accountability) are at Level 1 or 2. That places the overall programme at Level 1 or 2, not Level 3, regardless of total score.

How do I explain the maturity model to a CLO who does not know WCAG or captioning compliance?

Frame it as a capability maturity model — the format the CLO has likely already encountered in technology or operations contexts. The software CMM has five levels from Initial to Optimising; the ITIL maturity model has five levels from Initial to Optimising; the CMMC cybersecurity framework has five levels from Foundational to Expert. The language is the same: “At Level 1, we respond to requirements when they arise and have no systematic process. At Level 3, we have documented processes that are consistently followed and could pass an external review. Our goal is Level 3 by [date], which requires [specific investments]. Currently we are at Level 2.” Most CLOs understand this framework immediately and can use it to make a funding decision without understanding WCAG 2.1 SC 1.2.2 citation formats.

How long does the Level 2 to Level 3 transition realistically take?

Three to six months for a team with two to four L&D staff if the back-catalogue audit is scoped appropriately and treated as a project rather than an ongoing task. The policy can be drafted and reviewed in one to two weeks. The vendor contract amendment typically takes two to four weeks (L&D team drafts the amendment, legal reviews, vendor responds). The accommodation log setup takes one to two days. The back-catalogue audit timeline depends on library size: a 50-video library takes one to two weeks; a 500-video library requires a dedicated project of four to eight weeks. The realistic bottleneck is the vendor contract amendment, which requires legal review and vendor responsiveness. Start that conversation first.

Can a one-person L&D team realistically reach Level 4 (Proficient)?

Yes, but the investment is front-loaded and the back-catalogue remediation project requires budget that may be difficult to secure on a solo L&D budget. The realistic target for a one-person L&D team is a strong Level 3: written policy, 99%+ accuracy target and measurement, vendor accountability basics, comprehensive accommodation log, back-catalogue audit and project plan. If the solo L&D professional reaches Level 3 and documents everything at the succession-accessible standard described in Level 5, that is a programme a contractor could maintain during a vacancy. Level 4 requires funded back-catalogue remediation and a CART vendor under SLA — achievable but requires leadership budget approval, which requires the Level 3 documentation to make the case.

Our LMS auto-captions are enabled on all content. Does that put us at Level 2?

No. Auto-generated captions with no accuracy review are Level 1 for WCAG compliance purposes. The LMS caption feature is technically compliant in the sense that it produces a toggle-able CC track that is synchronised to the audio. What it does not produce is captions that are “equivalent” to the audio, which is the actual WCAG 2.1 SC 1.2.2 standard. Auto-generated captions achieve 80–90% accuracy on general speech and significantly lower accuracy on content with specialised vocabulary — product names, clinical terms, regulatory acronyms, engineering notation, legal Latin. A programme where all content has auto-captions and no accuracy review is a Level 1 programme with a false coverage indicator. Reaching Level 2 requires systematic production of reviewed, 99%+ accuracy captions on new content, not enabling a default LMS feature.

Is Level 5 (Strategic) realistic for a mid-market organisation with a three-to-five person L&D team?

Yes, but the investments required at Level 5 span outside the L&D team’s direct authority — legal review of content acquisition contracts, integration into board-level governance reporting, and cross-functional succession planning all require coordination with legal, DEI, and executive teams. A mid-market L&D team that has reached Level 4 and wants to reach Level 5 should start with the succession documentation investment (entirely within the L&D team’s authority and highest-leverage for programme durability) and the content acquisition contract language (requires legal collaboration but is a one-time investment that then operates automatically on all new content). Board-level reporting integration is a longer-horizon project that typically requires DEI or compliance team sponsorship. Level 5 is achievable for a three-to-five person team over two to three years of deliberate investment after reaching Level 4.