Almost every SCORM-versus-xAPI article asks which standard is better and then answers with some version of “it depends.” The framing is the problem. For most learning teams the choice of standard is not the decision that matters, and treating it as one is how organizations end up migrating to a richer data format they never actually use. The useful question sits one level up: what do you want to know that you cannot answer today, and are you set up to act on it once you can?
What each standard actually is
SCORM is the incumbent, and it earns the position. In its 1.2 and 2004 versions it packages a course into a zip file that any conformant LMS can import and run, and it reports back a small, fixed set of things: whether the learner completed, their score, pass or fail, time spent. That is a narrow data model, and it is also exactly enough for the work most courses do. Every major LMS supports it, and Storyline and Captivate export it directly, so the path from finished course to tracked-in-the-LMS is short and well understood. When people call SCORM dated, this is what they are dismissing, and it is worth being honest that “dated” and “sufficient” are not opposites.
xAPI, still sometimes called Tin Can, works differently. Instead of a package reporting to an LMS, it emits statements — actor, verb, object, roughly “a technician passed the forklift assessment” — and sends them to a Learning Record Store, an LRS. Because the statements can come from anywhere, xAPI can record learning that never touches a course: a mobile job aid, a branching simulation, a task done in the real software. The data model is open rather than fixed, so you can describe almost anything a learner does.
cmi5 is the one the debate usually forgets. It is a profile of xAPI — it runs on xAPI underneath, but it re-adds the launch, completion, and pass-fail rules that LMS administrators relied on with SCORM. In practice it is the closest thing to “SCORM with a better data model,” and for teams that want richer tracking without abandoning the LMS-centric way they already work, it is often the right answer rather than raw xAPI. It still needs an LRS to send statements to, though that store does not have to be a separate system.
What xAPI actually buys you, and what it costs
The upside is real. If you need to see the whole arc of how someone becomes capable — the app they practiced in and the on-the-job task they finally got right — SCORM cannot describe that and xAPI can. Any program where the learning happens across systems, or mostly outside a formal course, is a genuine case for xAPI.
The cost is the part the vendor blog skips. xAPI does not analyze anything. It collects. The statements land in an LRS, and then a person has to ask a question of that data and change something because of the answer. Most learning teams do not do this with the completion data they already have — the dashboard says ninety-one percent complete, and no one has asked what the other nine percent means or whether “complete” predicts anything at all. Collecting ten times more data you will not look at is not a step forward. It is a bigger pile.
xAPI needs a Learning Record Store, and the LRS is where the plan usually stalls. It is a separate piece of infrastructure to buy or host, secure, and maintain, and it is worthless until someone owns the analysis that justifies it. An LRS with no analyst behind it is a data warehouse nobody reads.
Where the standard actually is in 2026
It helps to know where xAPI actually stands, because the version number gets used as a selling point. xAPI 2.0 became an IEEE standard in 2023 (IEEE 9274.1.1), the first time an open-source learning specification was ratified that way, which matters for long-term stability. Adoption is still early. As of 2026 most platforms treat 2.0 as a passthrough — they accept the older 1.0.3 statements with a 2.0 header — and most authoring tools still emit 1.0.3. So “we support xAPI 2.0” is, for now, more a direction than a working difference, and not a reason to move on its own.
So which one should you use
The honest answer is that a real portfolio uses more than one, chosen content stream by content stream, rather than a single standard rolled out everywhere. SCORM is still correct in 2026 for LMS-delivered compliance and onboarding, where you need to know who finished and every LMS already speaks it. xAPI earns its place when learning genuinely happens across systems or outside a course, and a specific question and the capacity to act on it already exist. cmi5 is the pragmatic middle for teams that want the richer model without leaving the LMS-centric workflow behind, and it is the option the “SCORM is dead” posts tend to skip.
None of that requires taking a side in a debate that was mostly manufactured by the people selling the newer thing. The standard is the last decision, not the first. Start from the question you cannot answer today and the action you would take if you could. If you do not have that question yet, SCORM is not what is holding you back. The data was never the constraint. The question always was.