An elearning video is a module a learner is assigned and finishes alone. Building the first one is a project. Building forty, keeping a second language current and reissuing every module a policy touches is an operation, and that is the job this guide is written for.
Below: the five types of elearning video, what each costs, five routes from a PowerPoint deck, the checks to run before a module is assigned, and where the video stops and your learning platform takes over. Still weighing whether a session should become video at all? That decision has its own article.
Try NoLang for FreeWhat an elearning video is, and what makes it different
An elearning video, also written e-learning, is a module the learner is assigned and watches on their own schedule, delivered through a learning management system (LMS), an intranet or a shared link. What separates it from a training video is delivery and record-keeping, not subject matter.
Training magazine’s 2025 Training Industry Report puts 34% of US training hours on online or computer-based delivery, with 92% of organizations running at least part of their compliance training online and 45% entirely online. Compliance is revised every year and taught to everyone, which is why it moved first.
The first module is affordable almost however you make it. A catalog is a different sum: modules, times languages, times how often each is reissued. Languages belong in that sum for most US employers, since the foreign born made up 19.2% of the civilian labor force in 2024 (Bureau of Labor Statistics).
Maverick-kunPick one real module you owe someone this quarter and keep it in mind. Every table below reads better with a candidate.
The five types of elearning video
Sort elearning video by how the learner is meant to study, because that decides your source material and therefore which production route is open to you.
| Type | What it teaches | Where it is watched | Length and format |
|---|---|---|---|
| ① Lecture-slide | Policies, product knowledge, literacy topics | At a desk, as an assigned course | 5–15 minutes, landscape, narrated slides |
| ② Microlearning | One habit or one rule, nothing else | On a phone, between tasks | Around a minute, vertical, caption-led |
| ③ Avatar-led lecture | Annual mandatory topics that want a presenter | At a desk, on a deadline | 3–10 minutes, presenter on screen |
| ④ Case study | Judgment: what to say and do in a situation | Alone, or before a group discussion | 2–5 minutes, two speakers |
| ⑤ On-the-job skills | Procedures, work standards, site safety | Before a shift, or at the workstation | 2–10 minutes, step order |
Type ⑤ is easy to mistake for a video manual: a module that teaches the job is assigned and studied, while a reference opened mid-task is a different product. A software walkthrough belongs with type ⑤ when it is assigned and has to be completed, with a screen recording as the source instead of a work standard; the training video guide covers producing it.
How each of the five types is made
Settle the type before you touch a tool: it fixes whether your source is a deck, a script or footage, and the source determines your route. The design work at this stage is choosing the type and writing the source, not storyboarding shots.
Lecture-slide modules
Most of a catalog is this, and the content already exists: last year’s deck on policy, product knowledge, security or AI literacy is this year’s script.
- Design and source material: the existing deck, cut to one idea per slide, with the spoken version in the speaker notes.
- Production and finish: narrate the slides or generate the narration from the file, then add captions, section breaks and a figure check.
Shortest route: keep the deck as the master, so next year’s edition is an edit plus a rerun (sample).
Microlearning modules
One point per module and nothing else, vertical for a phone. Value comes from how many you publish and how often, not from polish, and keeping each one short is not a stylistic preference — the viewing data is in the design section below.
- Design and source material: a one-line learning point written as a script, with the takeaway on screen for anyone watching muted.
- Production and finish: generate vertically, burn in captions, and publish ten at a time on a schedule rather than as one-offs (sample).
Avatar-led lecture modules
An avatar presenter — an AI-generated presenter shown on screen delivering the narration — reproduces the lecture format without a studio, and removes the scheduling: a presenter, a room, an operator, a recut whenever a clause changes.
- Design and source material: a script written the way a person speaks, opening with why the topic is being taught this year.
- Production and finish: generate the presenter and the narration, add captions, and keep the script file as the thing you revise (sample).
Case-study modules
Case studies teach judgment by showing a situation handled badly and then handled well, usually as a dialogue. Traditionally the most expensive type, because acting it out convincingly takes performers.
- Design and source material: a two-speaker script, wrong way first, right way second, with one specific difference between them.
- Production and finish: have generated speakers perform the exchange instead of casting it, close with a commentary naming what changed, and spend the saved effort on a situation your staff recognize (sample).
On-the-job skills modules
Built from documents already in an operations folder: work standards, procedures, pre-shift safety rules. Film only the steps where the motion, the force or the sound is the information.
- Design and source material: the work standard in its real sequence, with the hazard and the reason at each step.
- Production and finish: generate the explanatory steps from the document, film the few physical ones, and caption everything for a noisy floor (sample).
Shortest route: generate the module from the standard, then decide which steps need a camera.
What each type costs and how long it takes
Elearning is not quoted the way video is: vendors price per finished hour of learning. Blue Carrot’s 2026 pricing guide publishes roughly $10,000 per finished hour for Level 1 (template-based), $15,000 for Level 2 (tailored) and $20,000 for Level 3 (simulation) — one studio’s asking prices, against market ranges elsewhere of a few thousand to $50,000 and up. The same guide prices media inside a course: synthetic avatar video around $50 per finished minute, AI voice-over $2–$5 against $20–$150 for a human.
| Type | Built for you (published rates) | Built in-house (effort) |
|---|---|---|
| ① Lecture-slide | Level 1 territory, about $10,000 per finished hour | Deck edit plus script: hours to two days |
| ② Microlearning | Cheapest per minute; AI voice-over at $2–$5/min | An afternoon for a batch, once the points are written |
| ③ Avatar-led lecture | Avatar video around $50 per finished minute | Script writing and review; no shoot to schedule |
| ④ Case study | Performers and direction push it toward Level 2 | Writing two versions of the exchange is the job |
| ⑤ On-the-job skills | Level 2 pricing where site filming is involved | The document exists; filming only where needed |
The right column stays in hours because the cost is your subject expert’s time; multiply by your internal rate to compare the halves. The benchmark the field still quotes is the 2010 Chapman Alliance study: 49 to 79 hours per finished hour of slide-driven Level 1 elearning, about 184:1 for Level 2. It predates current tools, so read the shape, not the number. The ratio holds because the binding constraint is the subject expert’s availability, not the production tool. Rates outside elearning are broken down in what video production costs.
The costs that repeat
Two multipliers sit on top of every number above. Each language is another translation, voice-over and caption pass, and every policy update or system rename reopens a finished file.
A module library is priced by module count × languages × revision frequency, so the route with the cheapest first version is not necessarily the cheapest catalog.
Maverick-kunSo the question is not what one module costs. It is what the second language and the April rewrite cost you.
Making an elearning module from a PowerPoint deck
Choose between the five routes from a PowerPoint deck on what a revision costs and on whether learner records have to come out of the file. The five are: recording inside PowerPoint, text-to-speech narration assembled in a video editor, generating the module with an AI video generation service, rebuilding it in a SCORM authoring tool, and commissioning an agency.
| Route | Cost | Time for one 20-slide module | Cost of a revision | Learner records |
|---|---|---|---|---|
| ① Record in PowerPoint | The license you have | A real-time take plus retakes: half a day | Re-record the affected slides, export again | From the host, not the file |
| ② TTS plus an editor | TTS plan and an editor license | A day or more of placing audio | Regenerate lines, re-time, re-export | From the host, not the file |
| ③ Generate from the deck | A subscription | Minutes to a first version | Edit the deck, generate again | From the host, not the file |
| ④ SCORM authoring tool | A per-author license | Days, with question design and a test upload | Re-author and re-package | In the package: completion, time, score, pass/fail |
| ⑤ Commission an agency | Quoted per project | Weeks | Back into the pipeline and billed again | Depends what you commissioned |
Read the last column carefully: routes ①, ② and ③ hand you a video file and leave tracking to whatever hosts it, and only ④ puts reporting inside the package. Reserve route ⑤ for what only a crew delivers — a difficult scene acted out, filming on a live site, a flagship course — since routine material is what an agency’s revision cycle handles worst.
Recording narration in PowerPoint
PowerPoint stores narration, slide timings and pen markup per slide, so a correction re-records one slide rather than the deck (Source: Microsoft Support). The price is your calendar: recording runs in real time and comes back at every revision, which is why twenty modules on an annual cycle rarely get refreshed this way.
TTS narration plus a video editor
This route suits a catalog that needs one narrator across modules written by different people. The work is per slide and does not shrink with practice — generate the line, place it, nudge the slide change, listen back — so it is reasonable for three modules and poor for thirty.
Generating the module from the deck
An AI video generation service collapses four jobs into one generation — writing the narration, speaking it, laying out the visuals, exporting the file — so a first version arrives in minutes and the deck stays the master. The generation itself is rarely what decides whether the route works. Four things around it do: how the deck is written, how you accept the output, which modules you keep human-made, and how the file reaches your platform.
Preparing the deck for a module, not for a talk
A deck built for a live session covers a whole topic, while a module has to cover one learning objective a learner is measured against, so the first edit is where the file gets split.
- One learning objective per module: a deck that teaches four things becomes four files, because that is the unit the LMS assigns, records and reissues.
- Speaker notes as the script: notes are read as the narration source, so the sentence you would actually say belongs there.
- Headings that double as chapter titles and file names: learners search with the words that were on screen, and the catalog shows whatever the file is called, so put the topic and the edition in both.
- Text a reissue can reuse: keep figures, policy names and dates as editable text rather than baked into an image, so another output language does not mean remaking the graphic.
Rewriting the slides for the ear — fragments turned into sentences, a twelve-row table cut to what can be spoken, a pronunciation list — is in the training video guide.
Accepting a generated module before it goes on the LMS
A compliance module states the company’s rules to every employee, so acceptance is a written step with a named owner.
| Check | What to look at | Who signs it off |
|---|---|---|
| Figures a rule depends on | Amounts, effective dates, clause and policy numbers | Legal or the policy owner |
| Caption accuracy | Captions present and correct — WCAG SC 1.2.2 is Level A | Accessibility or LMS admin |
| Length and section breaks | Whether it plays as one block a learner cannot resume | LMS admin |
| Catalog metadata | Title, edition and file name matching the assignment record | LMS admin |
The generic checks — pronunciation, narration against the screen, claims the source does not make — are in the training video guide. The sign-off column is what a module adds: the record has to show who approved the edition each learner was tested on. Corrections go into the source deck, since next year’s edition is generated from it and not from the file you patched.
Which modules to keep human-made
Anything whose value is the back-and-forth — a discussion, a practice round with feedback — is not a module at all, so keep it as a live session and use video for the preparation people do beforehand. Physical technique and messages that carry an executive’s authority are the other two cases to film, argued in the training video guide.
MP4 or SCORM?
Upload the MP4 and let your LMS record completion; you need SCORM only when the module itself has to report scores and pass or fail back to the platform.
Rustici Software, which maintains the SCORM specification, lists what the format adds over a bare video file: consistent packaging across learning platforms, plus reporting of completion status, time spent, score and pass/fail. A video file carries none of that. In return you take on an authoring license, questions to design and defend, and a test upload into your LMS — budget it as building a course, not making a video.
The division of labor is clean: NoLang produces, reissues and translates the video, and the LMS assigns it, tests the learner and keeps the record. When the module itself has to report a score and a pass/fail, that is a SCORM package’s job.
How to make an elearning video with NoLang
Mavericks, Inc. builds NoLang, an AI video generation service that turns material you already own into narrated, subtitled video. Two entry points matter for modules.
- From document: a PDF or PPTX file — the route for types ① and ⑤, where a deck or a work standard already exists.
- From script, or Generate from text in the editor: a script or a two-speaker dialogue — the route for types ②, ③ and ④.
Footage and audio work as sources too, and generating from a web page runs in the Chrome extension.
- Subtitles: captions for modules watched on a shared floor with the sound down.
- 34 output languages: reissue the same source for another site or shift without commissioning a recording.
- Avatar presenter: an AI-generated presenter delivers the lecture on screen, with no room to book.
- Mobile or Desktop: vertical for microlearning, landscape for a course module.
Upload the deck under From document
Start from the PDF or PPTX you already teach from. Speaker notes are picked up as the narration source, so real notes beat a tidied-up summary.
Decide how much of the deck to cover
Choose Explain all pages when every slide carries something the learner is accountable for, or Explain key points only for a refresher.
Set the narration detail and the format
Pick Briefly, Normal or Detailed for how much the narration says per slide, choose Mobile or Desktop, and turn Subtitles on.
Choose the output language and generate
Select one of the 34 output languages and run the generation. Narration, visuals and pacing are produced together, in minutes.
Check it against the source, then download the MP4
Run the acceptance table above, fix facts in the deck and regenerate, adjust pacing in the editor, then download the MP4 for your LMS.
A revised policy means swapping the deck and generating again; a second language means changing one setting. Neither reopens a production.
Turn Your Deck into a ModuleMaverick-kunTry it on whichever module is already overdue for a policy update. That is where a rerun proves its worth immediately.
Seven elearning modules made with NoLang
Seven modules produced with NoLang, ordered by the five types above. Six are narrated and captioned in Japanese because they were made for a Japanese audience; the seventh is in English. Any of them can be reissued from the same source in another of the 34 output languages. More sit in the Video Gallery.
Generative AI literacy as an assigned module
- How it was made: an introductory deck generated as narrated slides, with an on-screen presenter and captions.
- Where it fits: a company-wide literacy module, assigned to everyone at once and reissued as the topic moves.
Security training on an annual reissue cycle
- How it was made: generated from a deck covering three everyday rules — checking email senders, managing passwords, locking an unattended screen.
- Where it fits: mandatory training whose annual edition is an edit of the deck plus a rerun.
Reporting in one minute — a vertical microlearning module
- How it was made: generated from a short script into a vertical (Mobile) module built on one point: conclusion first, then the reason, then the background.
- Where it fits: phone-first reinforcement, published on a schedule so one habit lands at a time.
Personal data protection basics
- How it was made: generated from a script, with an avatar presenter delivering three ground rules for handling personal data.
- Where it fits: new-joiner lectures and the annual refresher, produced without booking a presenter.
Handling a customer complaint: wrong way, right way
- How it was made: a two-speaker script generated as a restaurant scene — a late order handled poorly, then well, with the difference called out.
- Where it fits: judgment training for frontline teams, with nobody cast and nothing filmed.
Drill-press work standard
- How it was made: generated from a written work standard, in sequence: preparation, pre-shift inspection, then the safety point at each step.
- Where it fits: manufacturing induction and standard-work refreshers, with no line stopped to film.
Factory safety orientation (English narration)
- How it was made: the same kind of safety source material, generated with English narration and captions instead of Japanese.
- Where it fits: site safety rules for workers who read English, and proof that a second language is an output setting.
Before you publish
Three things decide whether a module gets watched, uploaded and counted: design, specs, hosting.
Designing a module people finish
- One topic per module: two topics in one file means whoever needs the second sits through the first.
- Keep it short: Guo, Kim and Rubin found median engagement near 100% below six minutes, falling to roughly 50% at 9–12 minutes. MOOC data rather than corporate LMS data, but it argues for splitting.
- State the outcome first: within fifteen seconds, what the learner will be able to do.
- Put the comprehension check in the LMS: the video teaches; the platform tests and records.
One topic, kept short, with the outcome stated up front is what gets a module finished, and polish is not the lever. The same study found desk-recorded video held attention better than studio-produced video at longer lengths, which is a reason not to buy production values before the content is right.
Technical specs to check
| Item | What a module needs | NoLang output |
|---|---|---|
| File format | MP4 with H.264 video and AAC audio | Downloads as an MP4 file |
| Resolution | 720p is enough for slide-based teaching video | Generates at 1280×720 (Desktop) or 720×1280 (Mobile); HD or Full HD selectable on export |
| Frame rate | 24 or 30 fps; nothing higher helps a narrated deck | 24 fps or 30 fps, selectable on export |
| File size | Under the per-file cap your platform enforces | Varies with length and export quality |
| Captions | WCAG SC 1.2.2 makes captions a Level A requirement | Subtitles can be switched on at generation |
| Length | Short enough to finish in one sitting | Seven steps from about 30 seconds to about five minutes when generating from text |
Caps vary widely: Slate’s roundup of published LMS limits runs from roughly 350 MB to several gigabytes, and Moodle has no fixed default. Ask your LMS administrator two questions before you produce twenty modules: can the platform take an MP4 directly, and what is the per-file size cap. If a module is over the line, drop the export resolution or split it by topic.
Where to host it
| Destination | When it fits | What gets recorded |
|---|---|---|
| Unlisted link on a video platform | A pilot, a briefing, an audience without accounts | View counts, with no idea who watched |
| Intranet or employee portal | Reference material people should find unprompted | Page-level analytics at best |
| LMS | Anything assigned, mandatory or audited | Assignment, completion, time, and scores if you add a test |
Keep the roles apart: the video teaches and gets regenerated when the content changes, and the platform decides who has to take it and records who did.
Related articles
- How to make a training video — training video in general, including sessions you would film.
- Which trainings are worth turning into video — the decision layer, before production.
- How to make a video manual — reference how-to for a task in progress.
- What video production costs — rates beyond elearning.
Frequently asked questions
What L&D and instructional-design teams ask when they first move a catalog onto self-paced video: the source material they can start from, the settings that matter for an LMS, and the boundary between the video and the platform.
