Dynamic SCORM: How to Update a Course in Every LMS Without Re-uploading
Updated 12 November 2026 8 min read Czytaj po polsku
If your course runs in one LMS, updating it is a chore: export, upload, check. If it runs in fifteen client LMSs, a one-word fix turns into fifteen emails to fifteen administrators, some of whom will upload the new version weeks later and some of whom will reset everyone's attempts by accident. Dynamic SCORM exists to solve that. This article explains how it works, the two designs you'll meet, the problems each one runs into, and when an ordinary package is still the better choice.
The problem it solves
A normal SCORM package contains the whole course. Once it's uploaded, the LMS has its own copy, and that copy never changes. Every update means a new ZIP in every LMS.
That's fine for one LMS and rare updates. It hurts when:
- you sell or license courses to many organizations, each with its own LMS,
- a group of companies runs several LMSs with the same compliance course,
- content changes often: product training, prices, procedures,
- you need to stop a client's access when a contract ends.
How a dynamic package works
A dynamic package looks like any SCORM package to the LMS: a ZIP with imsmanifest.xml, uploaded the usual way. The difference is what happens at launch. Instead of opening course pages stored inside the ZIP, it loads the current version of the course from a server the author controls. Tracking still goes to the LMS through the normal SCORM API, so reports work as usual.
You'll see several names for the same idea: dynamic SCORM, proxy packages, remote SCORM, and "dispatch", the name Rustici Software uses for its product that delivers content into other LMSs.
Two designs
Design A: the course runs in a frame from another domain
The package's launch page contains a frame that loads the whole course from the provider's server. Because browsers stop a page from one domain reaching into a window from another, the course in the frame can't find the LMS's SCORM API itself. So the package includes a small bridge: the remote course sends messages with window.postMessage, and the launch page, which is on the LMS's domain, passes them on to LMSSetValue, LMSCommit and the rest.
Design B: the player stays in the package, only content is loaded
The package contains the course player and all the SCORM code. At launch, the player fetches the course content, such as text, quiz data and images, from the author's server over HTTPS, and renders it itself. Every SCORM call happens inside the package, on the LMS's domain, exactly as in a static package. Only data crosses domains.
| A: course in a frame | B: content loaded as data | |
|---|---|---|
| What updates without re-upload | Everything, including the player | Content; the player updates with a new package |
| Where SCORM calls happen | Relayed from the frame through a bridge | Directly in the package, on the LMS domain |
| Needs cookies across sites | Often, if the remote course keeps a session | No, requests can carry a token instead |
| Fallback when the server is unreachable | Usually none | Possible: a bundled copy of the last version |
| Server requirement | Hosting the whole course | Serving data with CORS headers |
Neither design is wrong. Design A is easier for a provider to build on top of an existing course player. Design B avoids the browser-policy problems below, at the cost of shipping a new package when the player itself changes.
Pitfalls and how to handle them
Corporate firewalls and proxies
The learner's browser must reach the content server. Many companies filter outgoing traffic, and a domain the proxy has never seen may be blocked or held for inspection. The learner sees a blank course. Ask the provider for the list of domains IT must allow, send it to the client before launch, and test from the client's network, not yours. A bundled fallback copy turns a blocked domain into a slightly outdated course instead of a broken one.
Third-party cookies
In design A, the remote course runs in a frame from another domain, so its cookies are third-party cookies. Safari has blocked those by default since version 13.1 in 2020, Firefox isolates them per site, and companies can block them by policy. A course that keeps its session in a cookie then loses it: blank frames, login prompts or lost progress in some browsers only. Design B doesn't need cookies for this.
CORS
In design B, the browser lets the package read data from another domain only if the server answers with a header allowing the LMS's origin:
Access-Control-Allow-Origin: https://lms.client-company.com
A list of allowed LMS origins on the server is advisory, not access control. It stops a package copied to an LMS that isn't on the list from loading content in learners' browsers there. It doesn't stop anyone who has the package from requesting the content outside a browser, where the origin can be anything. And some LMSs run packages in a sandboxed frame that sends no origin at all, so a strict list refuses those launches too. Use the list to keep a course where it belongs, and don't rely on it to protect the content.
Availability and speed
Every launch now depends on the content server. Serve content through a CDN, set a short timeout so a slow server doesn't leave learners staring at a spinner, and decide what happens if the request fails.
Caching
Browsers and CDNs cache files. An update may take a while to reach everyone unless content URLs change with each version. Ask how quickly a published update becomes visible.
Privacy
Ask what the content server receives. In some designs, learner names or IDs pass through the provider's servers, which then process personal data and need a data processing agreement. A data-loading design can be built so that learner data never leaves the LMS and the content server sees only anonymous requests.
Access tokens
The package has to identify itself to the server, usually with a token. Anyone with the ZIP can read that token, so it's an identifier, not a secret. An origin list keeps honest copies in place, but only revoking the token really stops it, so make sure that is possible.
Keeping learner progress when content changes
The LMS stores the bookmark and suspend data written by the version a learner used last. After an update, the new version reads that data on the next launch. Whether that works depends on how the course refers to its content.
- Stable IDs, not positions. A bookmark that says
slide-9breaks when you insert a slide before it. A bookmark that names the slide's ID survives reordering. - Removed content must be ignored gracefully. If the bookmark points to a slide that no longer exists, the course should continue from the nearest one, not crash.
- Recorded results stay recorded. The LMS doesn't re-evaluate learners who completed version 1. That's usually what you want, but it means a stricter quiz in version 2 applies only to learners who haven't finished yet.
- Some settings live in the manifest. The SCORM version, the pass mark and completion settings are read by the LMS at upload. Changing them needs a new package.
| Change | Push it dynamically? |
|---|---|
| Fix a typo or a wrong number | Yes |
| Replace an image or a video | Yes |
| Rewrite a slide | Yes |
| Add an optional slide | Usually; check where bookmarks land |
| Reorder slides | Yes with ID-based bookmarks, no with position-based ones |
| Add, remove or change quiz questions | Release as a new version |
| Change the pass mark | No: new package |
| Switch SCORM 1.2 to 2004 | No: new package |
When a static package is better
- The LMS or the learners' network can't reach outside content: offline sites, strict proxies, isolated networks.
- An audit requires proof of exactly which approved version a learner saw on a given date.
- A client's contract requires a self-contained deliverable.
- There is one LMS and updates are rare. Re-uploading is simpler.
- You're not sure the service will be around as long as the course. Ask what happens to packages if you stop paying.
Questions to ask any provider
- Do SCORM calls happen inside the package, or are they relayed from a frame on your domain?
- Does it depend on cookies in a cross-site frame?
- What happens when your server can't be reached? Is there a fallback copy?
- What happens to the packages in client LMSs if we cancel?
- Which domains must the client's IT allow?
- What learner data reaches your servers?
- Can access be limited to specific LMS domains and revoked?
- Which changes require a new package?
Dynamic SCORM in Scormbee
Scormbee offers Dynamic SCORM on the Creator and Studio plans, and it follows the second design: tracking stays inside the package in your LMS, and only the course content comes from Scormbee. Every package loads the latest version you publish; if Scormbee can't be reached, it plays the copy that came inside the package. Learners keep their progress after an update, no cookies are involved, and you can name the LMS addresses that should get updates and switch updates off when a contract ends. The Dynamic SCORM page explains the setup, including the domains to allow, and a regular static SCORM 1.2 or 2004 export of the same course is always available for LMSs that can't reach outside content. Plans are listed on the pricing page, and How to update a Dynamic SCORM course walks through publishing an update. Dynamic SCORM also works in the xAPI and cmi5 packages Scormbee exports. If your LMS supports cmi5, that's another route to the same goal; SCORM, xAPI and cmi5 explained compares them.
Frequently asked questions
What is dynamic SCORM?
A SCORM package that doesn't contain the course content itself. When a learner launches it, it loads the current version of the content from the author's server, while tracking still goes to the LMS through the normal SCORM API. Updating the content on the server updates the course in every LMS.
Does dynamic SCORM work in every LMS?
It uploads like any SCORM package. Whether it runs depends on the learner's browser reaching the content server, which corporate firewalls can block, and, for designs that load the course in a frame from another domain, on cookies that some browsers block.
Do learners lose their progress when I update a dynamic course?
Not if the course identifies slides and questions by stable IDs rather than positions. Completions already recorded stay in the LMS. Bigger changes, such as new quiz questions, are best released as a new version.
Is dynamic SCORM the same as cmi5?
No. cmi5 allows content to be hosted on any domain by design, but your LMS must support cmi5. Dynamic SCORM achieves a similar result with ordinary SCORM packages, which almost every LMS accepts.
Can I change the pass mark of a dynamic course without re-uploading?
Usually not. The pass mark and the SCORM version are part of the package's manifest, which the LMS reads at upload. Changing them needs a new package.