How Much Does Legacy eLearning Conversion Cost? – Where AI Cuts It

Somewhere on a shared drive, there's a folder of courses nobody wants to touch. A few were built in Flash, some in an authoring tool version that no longer opens on current laptops, and a couple are just narrated slide decks. They still hold valuable content, though, and your learners still need it. So how much does it cost to convert legacy eLearning, and where can AI bring that cost down?

eLearning conversion cost depends on which of three paths you choose: republishing as-is, rebuilding the same design in a modern tool, or redesigning the course. Republishing is the cheapest option, but it only works when you have editable source files. Rebuilding usually costs a fraction of building new because the analysis and content already exist, while redesigning costs close to a new build. AI tools now cut real hours from the slowest conversion tasks, like pulling text and audio out of old courses, drafting new storyboards, regenerating narration and translating, as long as a person checks the output.

What Are the Three Ways to Convert a Legacy Course?

The three ways to convert a legacy course are republish, rebuild and redesign. Picking the right one for each course is the biggest cost decision you'll make, and different courses in the same library often need different paths.

Many organizations started thinking about conversion when Adobe ended support for Flash Player. Adobe's Flash Player end-of-life notice confirms that support ended on December 31, 2020, and courses that depended on Flash simply stopped running in modern browsers. Those courses can't be republished, so they must be rebuilt or redesigned.

What Drives eLearning Conversion Cost?

The biggest driver of eLearning conversion cost is whether you still have editable source files. After that come the number of screens and interactions, the audio, and how much the content has aged.

• Source files. If you only have the published output, someone has to rebuild every screen by hand from what they can see and hear.
• Screen and interaction count. A 40-screen linear course converts quickly. A course with custom drag-and-drops, branching and timeline animations takes much longer.
• Audio. Old narration may not match new scripts. Re-recording with voice talent is expensive, and partial re-records rarely match the original voice.
• Content accuracy. If policies, products or procedures changed, you're not converting anymore. You're updating, and that needs a subject matter expert (SME).
• Tracking. Moving from an old learning management system (LMS) setup to a current SCORM (Sharable Content Object Reference Model) or xAPI (Experience API) package needs testing in the target LMS.
• Accessibility. Older courses were often built without keyboard access, captions or alt text. Conversion is the right time to fix that, and it adds hours.

Where Does AI Cut Conversion Hours?

AI cuts conversion hours most in the tedious extraction and drafting work that used to eat the first third of a project. That's good news, because those tasks were never the part that needed an expert's judgment.

Pulling Content Out of Old Courses

When source files are gone, the first job is getting the words back. Speech-to-text tools can transcribe old narration in minutes. Large language models such as ChatGPT or Claude can then clean up the transcript, combine it with on-screen text and organize it into a draft content outline. What used to take a developer days of clicking through a course and typing can now take hours, followed by a careful read.

Drafting New Storyboards and Scripts

With the content extracted, AI can draft a storyboard for the new build: screen-by-screen text, suggested interactions and draft knowledge checks. An instructional designer reviews the draft, keeps what works and corrects what doesn't. For a rebuild, the old course serves as the specification, so the AI draft is usually close.

Narration, Translation and Code

Text-to-speech platforms such as ElevenLabs can produce a full narration track from the new script, so you're no longer stuck matching a voice actor from years ago. Machine translation gives translators a strong first pass. And for courses with custom code, coding assistants like GitHub Copilot can help developers rewrite old scripts into modern JavaScript faster, with testing still required.

The human review step isn't overhead. It's the part of the project that makes the savings safe to keep. Budget for it on every course.

What Does the Math Look Like?

The math looks like any custom eLearning estimate: hours per finished hour, times a rate, times the course length. The difference is that conversion ratios are lower than new-build ratios, and AI-assisted steps lower them further.

The chart below is an illustrative example, not research data. It shows example hours to convert one finished hour of content on each path, with and without AI-assisted extraction, drafting and narration. The assumptions are mine, so use them as a shape to compare, not a quote.

Here's how that turns into dollars. Say you have a library of ten legacy courses, each about 30 minutes long, so 5 finished hours in total. Assume four need republishing, four need rebuilding and two need a redesign. At an example rate of $90 per hour:

• Republish: 2 finished hours x 12 hours = 24 hours, or about $2,160.
• Rebuild: 2 finished hours x 42 hours = 84 hours, or about $7,560.
• Redesign: 1 finished hour x 120 hours = 120 hours, or about $10,800.

That's about $20,500 for the library using the AI-assisted assumptions, compared with about $27,000 under the traditional ones (30 + 120 + 150 = 300 hours). Your numbers will differ, but the value of the exercise is seeing which bucket holds most of the cost, which here is the redesigns.

How Should You Estimate Your Own Conversion Project?

Estimate your own conversion project by sorting the library first, then pricing each group. Treat any course on the redesign path like a new build, and use the estimator below to size those.

1. List every course with its length, tool, source-file status and last update date.
2. Ask the content owner whether each course is still accurate. Mark it "accurate," "minor updates" or "outdated."
3. Assign a path: republish, rebuild or redesign.
4. Retire courses nobody needs. The cheapest conversion is the one you skip.
5. Pilot one course from each path to measure real hours, including the AI-assisted steps and human review.
6. Price the rest using the pilot pace, not a guess.

The pilot matters more than any spreadsheet. A single course run start to finish tells you how much time AI actually saved your team, where review took longer than expected and what the LMS testing involved.

Which Hidden Costs Show Up After Conversion?

The hidden costs that show up after conversion are mostly about the LMS and the people around it, not the courses themselves. They rarely appear in a vendor quote, so put them in your own budget.

First, consider completion history, because when you replace an old course with a new version, the LMS may treat it as a brand-new item. Decide early whether learners who finished the old version get credit, and who will map those records. Second, budget for internal testing time. Someone in your organization has to launch every converted course in the real LMS, on the devices learners actually use, and confirm it completes and reports a score correctly. For a big library, that's days of work.

Third, learners and managers need clear communication about which courses changed, which retired and where the new versions live. Fourth, plan for maintenance. A converted course in a modern tool is easier to update, which is part of the payoff, but only if you keep the new source files organized and named sensibly. Losing them again puts you right back where you started.

None of these is large on its own, but together they can add a meaningful slice to the project, and they all land on your internal team's calendar.

What Should You Ask a Vendor About a Conversion Quote?

Ask a vendor how they'll handle missing source files, which steps use AI tools, who reviews AI output, and how they'll test tracking in your LMS. Clear answers to those four questions tell you whether the quote is realistic.

Be wary of a vendor who promises a flat percentage discount for "using AI" without showing which tasks it applies to. Savings are real but uneven: they're large on transcription and narration and small on redesign decisions. For a fuller checklist, see our list of questions to ask an eLearning vendor about AI, which covers data handling, quality control and ownership of outputs.

Convert the Library, Not Just the Courses

eLearning conversion cost comes down to the path each course takes. Republishing is inexpensive when you have source files, while rebuilding is moderate and gets cheaper with AI-assisted extraction and drafting. Redesigning costs close to a new build, and it's worth it only when the old course wasn't working anyway.

Sort your library before you price it, retire what nobody uses, and pilot one course per path. Then let the measured hours, including review, set your budget. That approach turns a scary, open-ended project into a list of decisions you can defend. If you'd like help sorting the library or pricing that pilot, I'm happy to look at a sample of your old courses and give you an honest read on each path.

← All articles