Whatfix captions: digital adoption platform training video exports, vocabulary failure modes, and WCAG 2.1 AA compliance

How Whatfix generates training video and why captions require a separate step

The Whatfix content creation workflow

Whatfix operates as a browser-based or desktop overlay on top of enterprise applications. An L&D professional or application trainer uses the Whatfix Studio to record a guided walkthrough of a software workflow — clicking through screens, completing form fields, triggering system actions — while adding annotations, tooltips (SmartTips), step narration, and guidance overlays. This walkthrough, when published, runs as an in-application interactive guide that employees can follow inside the live application.

Whatfix provides three primary output formats for training content:

• Live interactive guidance — the SmartWalkThru runs inside the target application, step by step, as the employee performs the actual workflow. This is the primary DAP use case and is not directly relevant to video captioning.
• Self Help widget — contextual in-app help panel providing access to flows, task lists, and search. The widget surfaces training content but does not itself generate video.
• Video export (MP4) — Whatfix can render a completed flow as an MP4 video, creating a screen-capture training video of the software workflow with narration. This is the output format that creates the captioning requirement. L&D teams upload these MP4 exports to their LMS — Cornerstone OnDemand, Workday Learning, SAP SuccessFactors Learning, TalentLMS, Docebo, or any SCORM-compatible LMS — as standard training videos for employees who cannot run the live interactive guide or who need an asynchronous video reference.

The MP4 video export does not include a caption track. Whatfix Studio does not generate subtitles, SRT files, or VTT files as part of the export workflow. The exported MP4 contains the screen recording with synchronized narration audio, but there is no accompanying text representation of that audio for hearing-impaired learners.

Where captions must be added in the Whatfix-to-LMS pipeline

The captioning workflow for Whatfix-generated training video is a post-export, pre-upload step. After the L&D team exports the Whatfix flow as MP4, and before uploading to the LMS, a caption file must be prepared and attached. The specific mechanism for attaching the caption file depends on the target LMS:

• SCORM packaging path — for LMSes that accept SCORM packages (Cornerstone, Docebo, SAP SuccessFactors Learning, Absorb), the MP4 video can be packaged with a VTT or SRT sidecar in a SCORM manifest. The SCORM player in the LMS then associates the caption file with the video during playback. This requires including the VTT file in the SCORM package and referencing it in the imsmanifest.xml or equivalent packaging metadata.
• Direct video upload with sidecar path — for LMSes that accept direct video uploads with VTT/SRT sidecar files (TalentLMS, LearnUpon, Workday Learning, most modern LMS platforms), the L&D admin uploads the MP4 and the VTT file together, and the LMS associates them at the video level. See TalentLMS captions, Cornerstone OnDemand captions, and Workday Learning captions for platform-specific upload workflows.
• YouTube/Vimeo intermediate hosting path — some LMSes embed externally hosted video rather than hosting video natively. If the Whatfix MP4 is uploaded to YouTube (unlisted) or Vimeo and then embedded in the LMS, the caption file must be uploaded to the video host (YouTube Studio or Vimeo) rather than to the LMS directly. See Vimeo captions for training videos for the Vimeo workflow.

In all cases, the caption file must exist as a separate VTT or SRT file before the LMS upload. There is no post-upload captioning path within Whatfix itself.

Why the Whatfix video export specifically needs high-accuracy captioning

Whatfix MP4 exports are not general-purpose instructional videos. They are screen-capture walkthroughs of enterprise software workflows, narrated to accompany specific screen states. The audio track of a Whatfix export has characteristics that substantially increase the vocabulary error rate for generic STT:

• Dense proper-noun vocabulary — every enterprise application that Whatfix is deployed on has its own proprietary vocabulary. A Whatfix walkthrough of a Salesforce CPQ configuration workflow will narrate Salesforce-specific terms (opportunity, quote template, approval process, CPQ product configuration, price rule, discount schedule) at high density. A walkthrough of an SAP Fiori workflow will narrate SAP T-codes, module abbreviations, and configuration path names. These terms are out-of-vocabulary for generic STT models.
• UI element narration — the narrator describes on-screen elements by their exact names: "click the 'Manage Subscriptions' button," "navigate to the 'Contract Line Items' related list," "enter the GL account code in the 'Cost Center' field." UI element names are the most proprietary and most variable vocabulary in enterprise software, and they change with every application version upgrade.
• Configuration and parameter names — enterprise software walkthroughs routinely narrate configuration parameter names, field labels, dropdown option values, and system codes that are entirely specific to that implementation. A SAP implementation may have custom Z-tables, custom T-codes, and organization-specific field labels that differ from the standard SAP vocabulary. These are zero-corpus items for any STT model.
• Procedural step narration — Whatfix narration follows a procedural structure with precise action steps: "click Save and Submit," "confirm the workflow triggers," "verify the integration sync completes." The precision of procedural language means errors on technical terms are particularly consequential — a mis-transcribed system name or workflow step in a compliance training walkthrough gives the learner incorrect instruction.

Enterprise application vocabulary failure modes in Whatfix training exports

SAP implementations: T-codes, module abbreviations, and Fiori app names

SAP is one of Whatfix's largest deployment contexts. Enterprise companies running SAP S/4HANA or SAP ECC use Whatfix to create guided walkthroughs for finance, procurement, supply chain, HR, and other SAP processes. The vocabulary failure surface in SAP-based Whatfix exports is among the most severe of any enterprise application:

• T-codes — SAP transaction codes (T-codes) are alphanumeric identifiers like FB01 (post document), ME21N (create purchase order), MIGO (goods movement), FBL3N (G/L account line items), VA01 (create sales order), MM01 (create material), and hundreds more. When a trainer says "open T-code FB01 to post the journal entry," generic STT produces "open T-code FB 01" or "open T-code FB zero one" — technically readable but inconsistent, and completely wrong when the narrator uses the standard SAP pronunciation of running the letters and numbers together. More critically, lesser-known T-codes are often mis-transcribed as common words: "MD04" becomes "MD for," "ME53N" becomes "ME 53 N," "SU53" becomes "SU 53" or "sue 53."
• SAP module abbreviations — FI (Financial Accounting), CO (Controlling), SD (Sales & Distribution), MM (Materials Management), PP (Production Planning), HCM (Human Capital Management), PM (Plant Maintenance), QM (Quality Management), PS (Project System) — all of these two-letter module abbreviations appear constantly in SAP training narration. "FI-CO integration" becomes "FICO integration" or "fee co integration" depending on the narrator's pacing.
• SAP product names from the Oracle Health transition — SAP S/4HANA (often pronounced "S-4-HANA"), SAP Fiori ("fee-OR-ee"), SAP SuccessFactors, SAP Ariba, SAP Concur, SAP BTP (Business Technology Platform), SAP Business Connector, SAP GUI — each has pronunciation patterns that generic STT handles inconsistently.
• Custom implementation vocabulary — most SAP implementations include customer-specific Z-tables, custom T-codes, custom Fiori tile labels, and organization-specific process names. These are entirely proprietary and will never appear in any STT training corpus. A company with a custom T-code "ZHRP001" for their position management process will see it transcribed as "Z-HRP 001," "Z H R P 0 0 1," or similar non-standard forms.

For companies using both Whatfix and SAP, the captioning challenge compounds with the SAP Enable Now authoring tool, which is discussed in detail on the SAP Enable Now captions page. Whatfix and SAP Enable Now serve similar DAP/authoring purposes and are often deployed together or considered as alternatives; the vocabulary requirements are parallel.

Salesforce implementations: Cloud names, Agentforce, and org-specific vocabulary

Salesforce is the other dominant Whatfix deployment context. Companies use Whatfix to guide sales teams, service teams, and administrators through Salesforce workflows — CRM data entry, approval processes, CPQ configuration, Salesforce Flow automation. The Salesforce vocabulary failure modes in Whatfix exports include:

• Salesforce AI and product names — Agentforce (2024-era AI platform, previously Einstein), Revenue Cloud, Sales Cloud, Service Cloud, Marketing Cloud, Data Cloud, Commerce Cloud, MuleSoft, Tableau, Slack (Salesforce-owned). Each of these multi-word compound names is a high-frequency error surface. "Agentforce" is particularly new (coined 2024) with near-zero STT training data exposure.
• Salesforce developer vocabulary — Apex (Salesforce's proprietary programming language), SOQL (Salesforce Object Query Language), SOSL, LWC (Lightning Web Components), Aura Components, Flow Builder, Process Builder (deprecated), Visualforce, SFDC (Salesforce.com abbreviation), ISV (independent software vendor). Developer onboarding walkthroughs that narrate code in Apex or query construction in SOQL produce high error densities.
• Org-specific custom objects and fields — every Salesforce org has custom objects (typically with __c suffix), custom fields, and custom picklist values that are entirely proprietary. A Whatfix walkthrough that guides a sales rep through a custom object workflow ("navigate to the 'Renewal Opportunity Detail' custom object") will have those object names misrecognized.

For Salesforce Trailhead training specifically, the Salesforce Trailhead captions page covers the Trailhead-specific surface in more detail. Whatfix generates the internal employee walkthrough layer; Trailhead provides the official certification training layer.

ServiceNow implementations: Now Platform vocabulary and CMDB terminology

ServiceNow is deployed at 85% of the Fortune 500 and across 70+ US federal agencies. Whatfix is commonly used to guide employees and administrators through ServiceNow Now Platform workflows — incident management, change management, service catalog requests, CMDB updates, and HR case management. The ServiceNow vocabulary failure surface includes:

• ITSM vocabulary — CMDB (Configuration Management Database), CI (Configuration Item), CI class, change request, change advisory board (CAB), service level agreement (SLA), operational level agreement (OLA), underpinning contract (UPC), problem management, root cause analysis (RCA), known error database (KEDB). Generic STT handles these inconsistently across speakers.
• Now Platform product vocabulary — ITOM (IT Operations Management), HRSD (HR Service Delivery), IRM (Integrated Risk Management), SecOps, CSM (Customer Service Management), FSM (Field Service Management), TSOM (Telecom Service Operations Management), Flow Designer, IntegrationHub, GlideRecord (scripting API), Update Sets, Service Portal. The acronym density in ServiceNow training is among the highest of any enterprise platform.
• Federal agency-specific vocabulary — for federal agencies using ServiceNow, the Whatfix walkthroughs may include agency-specific configurations, federal record-keeping vocabulary, and GSA/OMB process terminology that intersects with government employee training captioning requirements.

The ServiceNow Learning captions page covers the Now Learning training platform specifically. Whatfix walkthroughs generate the custom workflow training layer that sits alongside Now Learning's official product education.

Workday implementations: business process and HCM terminology

Workday is widely deployed for HR, finance, and planning. Whatfix walkthroughs for Workday guide employees through self-service HR transactions, manager workflows, and finance processes. Workday vocabulary failures include Workday-specific terminology (business process, worklet, integration, report writer, calculated field, supervisory organization, position management, headcount plan) and the transition from legacy systems (SAP HCM, Oracle HRMS) that creates dual-naming contexts in training narration.

ADA Title I, Section 508, and EAA for Whatfix customers

ADA Title I: mandatory software onboarding and training

ADA Title I (42 U.S.C. § 12112) requires employers with 15 or more employees to provide reasonable accommodations for employees with disabilities, including hearing impairments. Software onboarding training delivered through Whatfix-generated videos is mandatory training — employees are required to learn the systems they use to do their jobs. A hearing-impaired employee assigned to complete a Whatfix-generated SAP onboarding training module has the same right to an accurate, accessible training experience as any hearing employee.

The ADA Title I accommodation standard requires functional equivalence: the hearing-impaired employee should be able to obtain the same information and comprehension from the training as a hearing employee watching without captions. A Whatfix video export captioned with generic STT at 80-90% accuracy — with SAP T-codes, Salesforce API names, and ServiceNow CMDB vocabulary systematically misrecognized — does not provide functional equivalence. The employee cannot follow the precise procedural steps described in the narration if the technical terms are garbled.

The ADA obligation attaches to the mandatory nature of the training assignment, not to the software platform. Using Whatfix as the content creation tool and Cornerstone or Workday Learning as the LMS delivery platform does not modify the employer's accommodation obligation. The obligation to provide accessible training exists because the training is mandatory, regardless of which tools create and deliver it.

Section 508 for federal agencies and contractors using Whatfix

Whatfix is deployed at federal agencies and at federal contractors who build and maintain government IT systems. Section 508 (29 U.S.C. § 794d) requires that federal agencies' electronic and information technology — including training content — be accessible to employees with disabilities. The technical standard referenced by Section 508 is WCAG 2.0 Level AA (for web content) and parallel standards for multimedia. For training video derived from Whatfix exports, this means captions that meet the accuracy standard specified in WCAG Success Criterion 1.2.2.

Federal contractors providing Whatfix-generated training to agency employees as part of a contract deliverable are bound by the same Section 508 standards through FAR clause 52.239-2 (or equivalent agency-specific clauses). A system integrator using Whatfix to create SAP S/4HANA onboarding training for a federal agency must deliver captioned training video that meets Section 508 standards — which means corrected caption files, not uncaptioned or auto-captioned exports.

For a comprehensive overview of the federal training video captioning obligation, see Section 508 captions for federal training video.

California FEHA for California technology companies

California FEHA (Gov. Code § 12940(m)) requires employers with five or more employees to provide reasonable accommodations for employees with disabilities. Whatfix's primary market is technology and enterprise software companies — a market heavily concentrated in California. The five-employee threshold makes virtually every California technology company that deploys Whatfix a FEHA-covered employer from its earliest growth stage. A 10-person startup using Whatfix to build product onboarding walkthroughs for its own internal team is a FEHA-covered employer for any hearing-impaired employee it hires.

California also requires specific mandatory training that creates captioning obligations when delivered as video: harassment prevention training under AB 1825/SB 1343 (two hours for supervisors, one hour for non-supervisors), pay equity and FEHA training for HR staff, and workplace safety training under Cal/OSHA standards. If any of these mandatory trainings are delivered as Whatfix-generated video, the FEHA accommodation obligation requires accurate captions.

European Accessibility Act for EU-deployed Whatfix implementations

The European Accessibility Act (EAA), enforceable from June 28, 2025, requires that digital products and services placed on the EU market meet WCAG 2.1 AA accessibility standards. For companies using Whatfix to create training video for employees in EU member states, the EAA applies to training content accessed by those EU-based employees. The practical consequence: Whatfix-generated MP4 exports uploaded to an LMS for EU-employee training must include WCAG 2.1 AA-compliant captions. YouTube auto-captions or uncaptioned exports do not satisfy the EAA standard.

The EAA applies in parallel with EU employment law: Article 5 of the EU Employment Framework Directive (2000/78/EC) requires employers to provide reasonable accommodations for employees with disabilities, including accessible training materials. A hearing-impaired employee in Germany, France, or the Netherlands assigned mandatory SAP onboarding training via Whatfix-generated video has the right to accessible captions under both the EAA framework and national employment accommodation law.

Caption file formats for Whatfix LMS integrations

VTT and SRT: the universal sidecar formats

The two caption formats that cover essentially all LMS upload scenarios for Whatfix-generated MP4 exports are WebVTT (.vtt) and SubRip (.srt). Both formats represent the same information — timestamped text segments synchronized to the video timeline — with minor syntactic differences.

• VTT (WebVTT) — the HTML5 standard caption format, supported natively by all modern web browsers and most LMS video players. VTT supports styling cues (color, position, font weight) that SRT does not, which can be useful for distinguishing different speaker tracks or highlighting key terminology in training content. For LMSes that run video in a browser-native HTML5 player, VTT is the preferred format. See VTT captions for training videos for format details.
• SRT (SubRip) — the most widely supported legacy format, accepted by every LMS platform as a fallback. SRT is plain text with sequential numbered cues, start/end timestamps, and caption text. For maximum LMS compatibility, SRT is the safer default if the LMS documentation does not specify VTT preference. See SRT captions for training videos.

For Whatfix exports used in SCORM packages, the VTT file is included in the SCORM package directory and referenced in the video element's track tag within the HTML content page. The SCORM player in the LMS then serves the caption track alongside the video. For direct video uploads to the LMS, the VTT or SRT file is uploaded as a sidecar alongside the MP4 in the LMS content management interface.

Platform-specific caption upload paths for Whatfix LMS targets

The following LMS platforms are common targets for Whatfix MP4 export uploads, with different caption upload mechanisms:

• Cornerstone OnDemand — Cornerstone accepts VTT caption files uploaded as a sidecar to video content in the Vignette content type. The admin uploads the MP4 and the VTT file together through the content upload interface. See Cornerstone OnDemand captions.
• SAP SuccessFactors Learning — SuccessFactors Learning supports caption file upload for video content through the Learning Module content management interface. VTT is preferred. The SAP vocabulary in the Whatfix export is the same vocabulary that appears in SuccessFactors Learning training content more broadly — see the internal link there for the full compliance analysis.
• Workday Learning — Workday Learning accepts VTT files as a sidecar to video content in the Learning Content administration. See Workday Learning captions.
• TalentLMS — TalentLMS accepts SRT and VTT files through the course builder's video content settings. Direct upload with caption sidecar. See TalentLMS captions.
• Docebo — Docebo accepts VTT and SRT files through the course asset management system. SCORM packages with embedded VTT tracks are also supported. See Docebo captions integration.

Building the glossary for Whatfix training video captioning

Two-layer glossary for DAP-generated content

Captioning Whatfix-generated training video requires a two-layer vocabulary approach because the audio contains two distinct vocabulary domains simultaneously:

Layer 1: the target application vocabulary — the enterprise software being demonstrated. This includes product names, module names, T-codes or equivalent identifiers, UI element labels, field names, dropdown values, and workflow step names. For SAP implementations, this means the SAP standard vocabulary plus the organization's custom Z-objects and custom T-codes. For Salesforce orgs, this means Salesforce product names plus org-specific custom object names and picklist values. This layer is the larger and more variable of the two.

Layer 2: the Whatfix platform vocabulary — terms specific to Whatfix itself that appear when the walkthrough narrates Whatfix UI elements or when the training content explains how to use the Whatfix Self Help widget. Whatfix-specific terms include: SmartTip, Flow, Task list, Beacon, Pop-up, Slideout, Guided Tour, Survey, NPS, Self Help widget, ContextBot, Whatfix Guidance, Whatfix Analytics, Task Manager, Flows Editor, Whatfix Studio. For internal training teams explaining how to use Whatfix itself, or for organizations training Whatfix administrators, the Layer 2 vocabulary is directly present in the audio.

GlossCap's glossary input accepts both layers as a combined glossary. The captioning engine applies glossary terms preferentially during the STT decoding process, recovering correct terminology for both the application layer and the DAP layer simultaneously. For organizations using Whatfix across multiple enterprise applications, a consolidated master glossary covering all deployed applications plus the Whatfix platform terms is more efficient than per-video glossaries.

Iterative glossary refinement for multi-application DAP deployments

Large enterprises that deploy Whatfix across five or more enterprise applications — SAP, Salesforce, Workday, ServiceNow, and one or more proprietary internal applications — accumulate vocabulary that spans different domains and update cycles. Application upgrades introduce new vocabulary (Salesforce Spring/Summer/Winter releases, SAP enhancement packages, ServiceNow Now Platform releases) that requires glossary updates. The recommended approach is a living glossary that is updated with each application release cycle and validated against new Whatfix content before LMS upload.

For regulated industries (healthcare, financial services, government), the glossary validation step also serves a compliance documentation function: demonstrating that the captioning workflow included a vocabulary-specific quality control step is evidence of due diligence in meeting the applicable accessibility standard.

FAQ — Whatfix captions

Further reading

• SAP Enable Now captions: authoring tool for SAP implementation training
• Salesforce Trailhead captions: Agentforce, Revenue Cloud, and Salesforce vocabulary
• ServiceNow Learning captions: Now Platform vocabulary and federal agency compliance
• Workday Learning captions: HCM and finance training video
• Cornerstone OnDemand captions: enterprise LMS caption upload workflow
• WCAG 2.1 AA captions: the 99% accuracy standard for training video
• Section 508 captions: federal agency and contractor training video requirements
• ADA Title I captions: employer accommodation obligations for training video
• VTT captions for training videos: format details and LMS upload workflows
• SRT captions for training videos: format details and maximum compatibility
• Engineering onboarding captions: SDK names, kubectl, and technical vocabulary density
• Glossary-biased captioning: how proprietary technical vocabulary is recovered from ambiguous audio