Making the caption ROI argument to your VP of Finance: exposure math, vendor cost comparison, and the half-FTE payback model

Why the compliance-only frame fails with Finance

The first mistake most L&D leads make when requesting caption software budget is leading with the legal requirement. "We need to comply with ADA" produces one of two responses from a finance partner: either "we're already compliant" (which is often wrong but hard to refute without a detailed technical audit in the room) or "put it in Legal's project list" (which moves it out of the L&D budget conversation and into a legal queue that may never convert). Neither response moves the budget conversation forward.

The second mistake is leading with volume. "We have 200 uncaptioned videos" is a project management metric, not a finance metric. Finance partners do not manage to-do lists. They manage risk, labour, and capital allocation.

The third mistake is leading with vendor comparison. Walking in with a feature matrix for GlossCap vs Rev vs 3Play before the budget is approved signals that you have already decided to buy and are asking for rubber-stamp approval. Finance partners resent that framing and are more likely to slow the process down, not speed it up.

The correct sequence is the sequence that every strong business case follows regardless of the spend category: downside risk first, then operating cost, then capital allocation. In captioning terms, that translates to: (1) here is the liability we are currently carrying and what it costs to resolve if triggered; (2) here is the labour we are currently absorbing and what it costs per year; (3) here is the vendor model that closes both and what it costs per month. The capital allocation conversation — "approve $99/month" — is the last item, not the first.

This post builds the three components in that order. Read it before your budget meeting, not during it.

Component 1: The exposure math — quantifying the ADA and WCAG liability

Finance partners evaluate risk in three variables: probability of event, cost of event if triggered, and cost of mitigation. The captioning liability calculation runs on all three. Before walking through the numbers, it is worth being precise about which legal frameworks actually apply to a given org — because the exposure profile differs materially across employer type, sector, and geography.

Which frameworks apply to your org

Note what is not on this table: ADA Title I (the employment discrimination framework). Title I is about reasonable accommodation for specific employees — captioning an otherwise-inaccessible training video on request. The frameworks above are about the organisation's obligation to provide accessible content regardless of whether any individual has filed a request. The distinction matters because the "we'll accommodate on request" approach, which many L&D teams currently rely on, does not satisfy Title II, Title III, Section 508, or WCAG 2.1 AA as a systemic compliance strategy for back-catalogue content.

What enforcement actually costs

The finance question is never "what is the statutory penalty" — it is "what does it cost us when this is triggered?" Those numbers differ by framework and by the maturity of the enforcement action.

OCR complaint (ADA Title II / Section 504). The Department of Education's Office for Civil Rights handles complaints against higher-education institutions and public-sector employers. OCR complaints do not produce immediate fines; they produce resolution agreements that require the institution to remediate identified non-compliance within a negotiated timeline, typically 12–24 months. The direct cost of an OCR complaint resolution has three components: outside counsel for the resolution agreement negotiation ($15,000–$45,000 for a straightforward complaint; $40,000–$120,000 for a complaint that escalates to a detailed investigation), remediation labour (captioning the back-catalogue identified in the complaint — typically 50–300 hours of video at 4× correction time, priced at market rates for outsourced captioning), and internal coordination time (accessibility officer time, IT time for LMS remediation, legal review of the final agreement). A representative mid-sized public university that received an OCR complaint in 2023 described their total resolution cost — outside counsel plus remediation plus internal time — as "just over $90,000." The complaint was filed by a single deaf student over three courses in one semester. The university had 40,000+ uncaptioned videos in its LMS.

ADA Title III civil suit. Title III suits against private employers and training providers are filed by individual plaintiffs in federal district court. The economics of these cases are well-documented: plaintiffs' attorneys work on contingency and pursue injunctive relief plus attorney fees, which are recoverable under the ADA even when monetary damages are zero. The real cost to the defendant is almost never the judgment — it is the settlement to avoid a trial and the legal fees to get to settlement. For mid-market companies (50–500 employees, $10M–$100M revenue), Title III captioning settlements have ranged from $10,000 to $50,000 in direct payment plus plaintiffs' attorney fees averaging $75,000–$150,000 per case in 2024–2025 data. Most cases that are not dismissed at the pleadings stage resolve for $85,000–$200,000 total cost to the defendant. Add outside defense counsel ($200–$450/hour × 100–300 hours = $20,000–$135,000) and internal management time (HR, legal, executive, and L&D coordination through a 12–18 month litigation timeline) and a single civil ADA caption case costs a mid-market company $110,000–$340,000 in total economic impact.

EAA non-compliance. The European Accessibility Act penalty structure is set by each member state. Germany, France, and the Netherlands have published fine schedules; Germany's current maximum per violation is €100,000, with ongoing violations treated as continuous events. The more immediate EAA exposure for most affected companies is not the fine but the market-access suspension mechanism: non-compliant companies can be barred from selling covered digital services in the enforcing member state while non-compliance persists. For a €5M ARR SaaS company with 20% of revenue in Germany, a six-week market-access suspension is a €115,000 revenue event. This is the number that belongs on the finance slide for EAA-exposed companies.

The liability estimate for a representative mid-market org

A 150-employee SaaS company with a US customer base and no current WCAG-compliant caption programme might have the following exposure profile:

Using the low ends of both probability and cost, the 5-year expected value of the captioning liability exposure for this org is approximately $14,000. Using midpoints it is approximately $56,000. Using high ends it is approximately $86,500. Even the low-end number is 14 years of GlossCap Team-plan subscription cost. The expected-value frame is not the only valid one — many finance partners prefer to evaluate the single worst case rather than the probability-weighted average, in which case the number is "one Title III case costs as much as 139 years of the software."

A brief note on framing: do not present this table as "we will definitely get sued." Finance partners who feel they are being manipulated by catastrophising will disengage. Present it as "this is the actuarial framing — what is the expected cost of carrying this risk versus mitigating it?" The expected-value comparison between software cost and expected liability cost is the argument that works, and it works because it is rigorous rather than alarmist. For more on the specific compliance obligations that create this exposure, the ADA Title II sprint plan post and the ADA Title II captions requirements page have the regulatory detail.

Component 2: The labour line — the half-FTE already embedded in your calendar

The second component of the finance case is the one most finance partners find the most surprising, because it requires showing them a cost they are already paying — just not in a form that appears on any budget report. This is the absorbed caption correction labour described in detail in the hidden half-FTE analysis; the following is a summary of the key numbers at three org configurations, designed to give you the table for a finance presentation.

The 4× real-time correction rate

Correcting auto-caption output (from YouTube, Whisper baseline, or any LLM-based captioning service) to a WCAG 2.1 AA 99% accuracy bar runs at approximately 4× the runtime of the source video as the working median, based on DCMP training data and our own minute-by-minute audit of three L&D operations teams. A 15-minute training video takes one hour to correct. A 60-minute lecture capture takes four hours. The 4× multiplier is conservative; teams using non-specialist tooling (a text editor and a video player side-by-side, which is the most common setup we observe) typically run 5–6×.

The reason this work is invisible is structural. It distributes across calendars: an instructional designer spends Monday mornings on correction, a video coordinator handles re-uploads on Fridays, a contractor bills two hours on each quarterly compliance module. Nobody owns the total; nobody reports it; it never rolls up into a line item that a CFO would recognise as "caption correction." It is simply absorbed as overhead in the operational rhythm of the L&D function.

The three-org table

The $50/hour loaded labour rate is the 2026 loaded median for a US-based L&D operations specialist (cash compensation ~$85,000, loaded at 1.3× for benefits, payroll taxes, equipment, and overhead = ~$110,000 against 2,080 working hours). Public-sector orgs and universities run lower (~$38/hour); enterprise tech orgs run higher ($65–$80/hour for a credentialled instructional designer). Adjust the table by ±25% and the conclusion does not change, because the software cost is two orders of magnitude smaller than the labour cost in every scenario.

The "% of one FTE" column is what you bring to a finance conversation. A VP of Finance understands "38% of one full-time employee is spent on a task that a $99/month subscription eliminates." They do not need to understand caption correction to evaluate that number. If your organisation is already past the point where the l&d headcount is insufficient to meet the training production demand — which is true of most orgs past 150 employees — the conversation is even simpler: "we are currently allocating 38% of our highest-leverage training capacity to a mechanical task that software can handle, and that allocation is costing us more in opportunity cost than in direct labour."

The second-order costs that are larger than the labour line

The labour line is the most straightforward component to present to finance, but the full analysis identifies three second-order costs that are, in aggregate, larger than the direct correction labour:

1. Time-to-publish delay. Every video that enters the caption correction queue adds 5–10 business days before it is live in the LMS. For compliance training with a regulatory deadline (annual HIPAA refresher, OSHA required modules), that delay is either a breach of the training SLA or a decision to publish without compliant captions. Both outcomes carry cost: the former delays business operations; the latter increments the compliance exposure from Component 1. At an average video production cost of $2,000–$8,000 per hour of finished training content, publishing delays are expensive.
2. Back-catalogue accumulation. Every uncaptioned video that accumulates in the back-catalogue increments the compliance exposure from Component 1. More precisely: it increments the remediation cost that would be required if a complaint or suit triggers. OCR resolution agreements typically require remediation of the entire relevant back-catalogue, not just the videos that were specifically complained about. An org that has been absorbing rather than captioning for three years may have a 500-video remediation obligation that costs far more to execute under an enforcement timeline than it would have cost to prevent.
3. Retention signal in L&D roles. Caption correction is the task most frequently mentioned in L&D operations exit interviews as "not what I was hired for." In a labour market where the median L&D operations hire takes 11 weeks and costs 1.5× annual salary in recruitment and onboarding, a retention problem driven by mechanical work is a predictable and preventable expense. This cost is difficult to put a precise number on, but the directional signal is consistent: orgs that reduce mechanical work in L&D operations roles have lower voluntary turnover in those roles, and the cost of that turnover reduction is real even at partial attribution.

For a finance presentation, the second-order costs go on the slide as qualitative risk factors, not quantitative line items, unless you have a specific data point (e.g., you had one L&D ops departure last year that you can partially attribute to correction burden). The direct labour line is the number you defend in detail; the second-order factors are why the upside of the investment is larger than the primary model suggests.

Component 3: The vendor cost comparison — why per-minute pricing does not scale

The third component addresses a question finance partners almost always raise: "why can't we just pay Rev $1.50 a minute when we have a video to caption?" The answer is not that Rev is a bad product — it is that the per-minute pricing model has a fundamentally different cost profile at real training-video volumes than the flat-monthly model does, and at any volume above approximately 4 hours per month, the per-minute model costs more. Much more. The detailed breakdown is in the pricing comparison post; the following is the finance-presentation-ready version.

Per-minute pricing: what it actually costs at volume

Per-minute or per-audio-minute pricing is the standard model for asynchronous captioning vendors (Rev, 3Play, Verbit, Otter.ai, and most SMB-targeted services). The published rates as of 2026:

Two things stand out in that table. First, the AI-tier pricing from Rev ($0.25/audio-minute) sounds cheap but delivers 80–85% accuracy — which is the accuracy level that requires the full 4× correction overhead described in Component 2. The per-minute spend is in addition to the correction labour, not instead of it. You cannot use Rev's AI tier as a substitute for the labour line; you can only use the human-review tier as a substitute, and that tier costs $1,350/month at the 15-hour median volume. Second, Verbit's volume pricing requires a minimum annual contract of $24,000; a mid-market L&D team that signs a Verbit contract is committing to a spend floor that is 20× the GlossCap Team annual subscription before they produce a single video.

The crossover point

At what video volume does the flat-monthly model cost the same as the per-minute human-review model? Using Rev's $1.50/audio-minute rate:

• At $99/month flat, the per-minute equivalent rate is $99 ÷ 1,800 minutes (30 hrs × 60 min) = $0.055/minute. Rev human review at $1.50/minute is 27× more expensive per minute at the Team plan's ceiling volume.
• The crossover in absolute terms: $99 ÷ $1.50/min = 66 audio-minutes = 1.1 hours of video per month. If you produce more than 1.1 hours of training video per month, flat-monthly costs less than human-review per-minute. The median ICP produces 15 hours. The cost ratio is 13.6×.

This is the number that answers the "why can't we just use Rev?" objection in the finance meeting. "Rev human review at our current volume costs $1,350/month. GlossCap Team costs $99/month, delivers the same accuracy bar with glossary correction for our technical terms, and includes the edit UI so our team can verify rather than correct from scratch. The difference is $1,251/month, or $15,012/year." That is a number a VP of Finance can evaluate without needing to understand the captioning industry.

The glossary factor in the cost comparison

The accuracy comparison between per-minute human review services and glossary-aware AI captioning is not identical across all content types. For training video heavy in proper nouns, product names, SDK identifiers, medical terms, or technical acronyms — the content profile of every GlossCap ICP vertical — generic human review services and AI-only services both fail at those specific terms because the reviewer does not know your vocabulary. A Rev human reviewer correcting an engineering onboarding video will catch most phonetic errors but will not know whether "Kubernetes" or "Cubernetes" is the correct spelling; will not know that your company calls the product "LogStream" not "Logstream"; will not know that the compliance module says "PHI" as a term of art, not "Ph.D." The WCAG 2.1 AA 99% accuracy threshold is a corpus-wide metric; WCAG-compliant captions are evaluated on all words including proper nouns, not on non-technical words only. A service that delivers 99% accuracy on conversational English and 85% accuracy on your product vocabulary is not delivering a WCAG-compliant product for your use case. The glossary layer is not a differentiating feature; it is a requirement for the accuracy guarantee to hold on training content.

Building the one-slide business case

The three components above translate directly into a business case that fits on a single slide. Finance partners who receive multi-deck ROI analysis for a $99/month tool are more likely to be annoyed than persuaded. The one-slide format disciplines you to present only the numbers that move the decision, which is also the format most likely to survive the 8-minute slot you will have in a budget review. Here is the structure:

At the median configuration (15 hrs/month, mid-market SaaS, currently using Rev human review at $1,350/month, 5-year liability EV of $30,000 using midpoint estimates), the slide reads:

• Current liability exposure (5-yr EV): $30,000
• Current correction labour / year: $36,000
• Current vendor captioning spend / year: $16,200
• Total current cost: $82,200
• GlossCap Team annual cost: $1,188
• ROI multiple: 69×
• Payback period: <1 month

If your organisation is not currently using a per-minute vendor (i.e., you are absorbing 100% through internal correction with no captioning spend), the vendor line is $0 and the ROI is still 55× on the labour plus liability components alone. The payback period is under 30 days because month-1 labour savings of $3,000 exceed the first month's subscription cost of $99 on day one.

What to put in the recommendation box

The recommendation section of the business case slide should contain exactly two sentences. The first states the ask: "Approve a $99/month subscription to GlossCap's Team plan for a 90-day pilot, with a checkpoint at 90 days to validate correction-time savings against the model." The second states the downside of not approving: "The labour absorption and liability exposure described above continue at their current rate; if a Title III suit or OCR complaint is filed in this period, the event cost exceeds 139 months of the annual subscription." The 90-day pilot framing is deliberate — it transforms an ongoing commitment into a bounded test, which is almost always easier to approve than an indefinite line item.

Anticipating finance objections

VPs of Finance ask good questions. The following are the seven we hear most often in budget conversations about captioning software, with the response that addresses each without overselling.

"Can't Legal handle the compliance risk? That's their job."

Legal handles compliance exposure reactively — after a complaint is filed or an audit is initiated. The cost of legal response described in Component 1 ($85,000–$200,000 for a Title III suit) is the cost of Legal handling it. What Legal cannot do is prevent the back-catalogue from accumulating uncaptioned content, because that is an L&D operations workflow problem, not a legal problem. The way to reduce the expected cost of the compliance exposure is to reduce the size of the non-compliant back-catalogue, which is an L&D operations solution. The counter is: "Legal's involvement begins after the complaint arrives; this investment prevents the complaint from having merit."

"We're already compliant. We have YouTube auto-captions on everything."

YouTube auto-captions are not WCAG 2.1 AA compliant. WCAG 2.1 AA Success Criterion 1.2.2 requires captions at 99%+ accuracy with correct synchronisation and meaningful punctuation. YouTube auto-captions deliver 80–90% accuracy on average across all content types, and significantly lower on technical vocabulary. The compliance exposure from Component 1 is based on the gap between YouTube-auto-caption accuracy and WCAG 2.1 AA accuracy on your specific content. An org that has YouTube auto-captions on all its training videos and has not had them reviewed against a technical glossary is, from a compliance audit standpoint, in approximately the same position as an org with no captions — because the auditors who enforce Title II and Title III use the WCAG 2.1 AA standard, not a "captions exist" standard.

"What about the $30/hour transcriptionist we already use?"

At $30/hour, the correction labour cost for a 15-hour/month org drops from $36,000 to approximately $22,000/year (using 4× multiplier × 180 hours × $30/hr). The ROI multiple drops from ~55× to ~18×. The 90-day payback period becomes 90 days. The payback still precedes the first annual renewal by 9 months, and the accuracy outcome for technical content is likely better because the glossary layer corrects the terminology errors that a general transcriptionist does not know to look for. If the current transcriptionist is already correcting to WCAG accuracy for a specific content type, the question becomes "is their correction rate for our technical vocabulary reliable at 99%?" — and the honest answer for most general contractors is no. The counter is not "your current setup is wrong," it is "our glossary layer adds accuracy specifically for the vocabulary that general contractors can't catch."

"Why not just negotiate a volume discount with Rev?"

Rev does not publish a volume discount for the human-review tier at the annual volumes a mid-market L&D team generates (roughly $16,200/year at 15 hrs/month). Volume discounts from Rev's human-review service typically begin at $50,000–$75,000 annual spend. Even if a discount were available, the unit-economics problem remains: per-minute pricing scales linearly with video production, while flat-monthly pricing does not. An org that increases video production by 2× doubles its per-minute spend and does not change its flat-monthly spend. For an organisation with a growing training library — which is true of virtually every company in a growth phase — the flat-monthly model has a structural cost advantage that compounds over time.

"What if our video production drops? We'll be paying for capacity we don't use."

At $99/month, the minimum monthly cost of "unused capacity" is the cost of a catered working lunch. The more important observation is directional: the trend in training video production at mid-market companies is consistently upward, driven by remote and hybrid workforce growth, compliance training volume expansion, and product release cadence. An org producing 15 hours of training video per month today is likely to produce 20–25 hours in two years. The per-minute model gets more expensive as volume grows; the flat-monthly model does not. The hedging argument ("what if we produce less") works against per-minute as well — if production drops to 2 hours/month, the flat-monthly cost is $99 and the per-minute cost is $180. The flat-monthly model wins at every volume above 1.1 hours per month.

"This seems like a nice-to-have. Can it wait until next budget cycle?"

The liability exposure from Component 1 is not deferred when the budget decision is deferred. Every video published without compliant captions increments the back-catalogue remediation obligation that would be required under an enforcement action. If the ADA Title II enforcement window (which opened 2026-04-24 for the relevant entity types) produces a complaint in the next six months, the remediation obligation is based on the full back-catalogue as it exists at the time of the complaint — not as it existed at the start of this budget conversation. The "wait until next budget cycle" decision has a concrete expected-cost consequence that can be stated precisely: "deferring by 6 months increases the expected liability exposure by approximately [number of videos to be produced in 6 months × estimated remediation cost per video] and foregoes [6 months × monthly correction labour savings]." Most finance partners will engage differently with a deferred-cost framing than with a "we should really do this" framing.

"Can you show me the actual time savings on a real video, not just the model?"

Yes — and this is the strongest closing move available to you. The caption-mangle scanner on the GlossCap demo page renders a side-by-side of auto-caption output versus glossary-corrected output on a pasted term list or a short video clip. If your finance partner wants to see the accuracy problem rather than read about it, the demo takes 3 minutes and produces a concrete illustration of the vocabulary-accuracy gap that the model above is based on. If they want to see the time-savings claim validated on a real asset before approving the pilot, the Solo plan at $29/month covers 5 hours of video with a paste-in glossary, costs less than most catered lunches, and produces the before-and-after comparison on your actual content within 24 hours.

The ROI calculation — three configurations in full

The following table builds the full business case at three representative org configurations. All numbers use the conservative labour rate ($50/hour loaded) and the conservative liability estimate (EV at probability midpoints, cost low-end). The "aggressive" columns use $65/hour and the high-end probability × cost product.

Note that the ROI multiple for the healthcare/university configuration does not increase proportionally with volume, because the Org-plan cost ($299/month = $3,588/year) is higher than the Team plan. Even so, the absolute net savings are larger and the payback period is still sub-month. The conservative ROI of 32× at 30 hours/month is actually conservative — it does not include the second-order costs (time-to-publish delay, back-catalogue accumulation risk, retention signal) that the full labour analysis identifies as larger than the direct correction line. Including those second-order costs at even half their estimated value brings the effective ROI to 50–80× for the 30-hour configuration.

How to present the ROI number

Do not lead with the ROI multiple in a finance conversation. A 49× ROI sounds like a marketing claim, and finance partners hear it that way. Lead instead with the specific numbers that compose it: "we are currently absorbing approximately $36,000 per year in correction labour, spending $16,200 per year on Rev human review, and carrying a measurable liability exposure. The total is $57,800. The tool costs $1,188. The ROI calculates at 49× — but the number I want you to focus on is the $57,800 we are spending today with no software tool, because that is the baseline we are paying right now regardless of what you decide about the software." That framing anchors the conversation on a cost that is real and current, not on a multiple that sounds promotional.

The further reading below includes the vendor RFP playbook if you are at the stage of evaluating multiple vendors and want a framework for the selection decision rather than the budget decision. The pricing comparison post has the per-minute versus flat-monthly analysis at additional volume points and includes the 3Play and Verbit pricing detail skipped above for brevity.

The approval path after the business case

Once the business case is approved, the procurement path for a $99–$299/month tool should be straightforward — it is below the threshold for a formal RFP at most mid-market organisations ($10,000 annual threshold is common), it does not require an IT security review unless your procurement policy triggers one for SaaS tools that receive video uploads (check your TPRM policy), and it does not require legal review unless the contract exceeds your signature authority threshold. For organisations that do require vendor security review, the relevant materials to request from GlossCap at that stage are: SOC 2 Type II report (on roadmap for 2026), data processing agreement and GDPR/CCPA terms, sub-processor list, and data retention policy. The vendor contract review checklist published last week covers the clause-by-clause SLA terms that matter in a captioning vendor contract and is worth reviewing before you sign, even for a subscription tool at this price point — particularly the glossary ownership and data retention provisions.

For the 90-day pilot structure recommended in the business case slide: pilot success criteria should be defined before the pilot starts. A workable set of criteria for a captioning tool pilot: (1) correction time per video reduced by ≥60% measured on a blinded side-by-side of 5 matched videos; (2) glossary-vocabulary accuracy ≥99% on a 20-term test set drawn from your company glossary; (3) LMS export workflow works end-to-end for your specific LMS (TalentLMS, Docebo, Absorb, Kaltura, etc.) without requiring IT intervention; and (4) no open security findings from your TPRM review at 90 days. If those four criteria are met at 90 days, the business case has been validated on your actual content with your actual team, and the continued subscription is a straightforward renewal rather than a budget conversation.

FAQ

Further reading

• The hidden half-FTE in your L&D budget: video caption correction costs — the full labour-cost analysis with minute-by-minute audit
• Rev vs 3Play vs Verbit vs GlossCap: pricing comparison for L&D teams — per-minute vs flat-monthly at every volume point
• The L&D captioning vendor RFP playbook 2026 — how to run a vendor selection process once the budget is approved
• ADA Title II just became enforceable — what training teams need to fix this week
• Caption vendor contract review checklist — the clause-by-clause terms to require before signing
• WCAG 2.1 AA captions: the exact technical spec your content must meet
• ADA Title II captions: who is covered, what is required, and the enforcement timeline
• Section 508 captions: requirements for federal agencies and contractors
• Rev vs GlossCap — detailed per-dimension comparison
• 3Play vs GlossCap — accuracy tier and pricing breakdown
• Verbit vs GlossCap — when enterprise pricing is and is not justified
• Live demo: caption-mangle scanner — see the vocabulary accuracy gap on your own terms