Completion rates — the real data

#1. The one study worth actually citing

Katy Jordan's 2015 study, "Massive open online course completion rates revisited: Assessment, length and attrition," published in the peer-reviewed International Review of Research in Open and Distributed Learning (IRRODL 16(3)), analyzed 221 MOOCs with enrollment data (129 of which had usable completion data) and found a median completion rate of 12.6%, ranging from 0.7% to 52.1% across the sample. [Established] — peer-reviewed, disclosed methodology, disclosed sample size; this is source-finder tier 1 for this specific claim, and the single most citable number in this entire course. Jordan's own regression analysis found that as MOOC enrollments fell over the period studied, completion rates rose — consistent with a selection-effect explanation: as the free, low-commitment audience thinned out, the remaining enrollees were more likely to be people who genuinely intended to finish.

The "87% of people who start an online course never finish it" figure that circulates in course-marketing content is simple arithmetic on Jordan's own number (100 − 12.6 = 87.4), not a separate finding — worth knowing so a reader can recognize it as the same underlying data restated, rather than as independent corroboration.

#2. What Jordan's data is not evidence of

Jordan's dataset is MOOCs specifically — overwhelmingly free or near-free, open-enrollment courses from platforms like Coursera and Open2Study during the mid-2010s MOOC boom. It is not a study of paid, standalone course products of the kind this course is actually about, and the mechanism from module 1 gives a real, specific reason to expect the two populations to differ: payment itself is a commitment device, and MOOCs' free, no-stakes enrollment selects for exactly the low-commitment browsing behavior that produces a 12.6% completion median. This course could not locate a peer-reviewed or disclosed-methodology study specifically measuring completion rates for paid, self-hosted or platform-sold standalone courses at anything like Jordan's rigor — which is itself a real, worth-stating gap, not a reason to fill it with an unsourced number instead.

#3. What gets claimed to fill that gap, and why none of it clears the bar

A wide range of secondary and content-marketing sources claim specific completion figures for paid courses — "15–40%," "paid courses achieve 60% completion," even one course-tool marketing page's claim of "5% to 85%" as if that range constitutes useful information. None of the sources checked in this research disclosed a sample, a methodology, or a named study behind any of these figures; several appear on sites that sell course-completion or student-engagement tooling, giving them a direct commercial interest in emphasizing how solvable (and therefore how big) the completion problem is. This course does not repeat a specific paid-course completion percentage as fact, because none was found that met this platform's sourcing bar. What can be said with real confidence: the direction — paid, structured courses complete at meaningfully higher rates than free MOOCs — follows directly and plausibly from the accountability mechanism in module 1, even without a rigorous number to attach to the size of the gap.

#4. The Maven cohort-completion claim, revisited here from the completion-data side

The previous module named Maven's 90–96% cohort-completion claim as a disqualified source on incentive grounds — a company's own metric, reported by its own co-founder, in a form (week-two "retention," not full-course completion) more generous than the label attached to it. Worth restating here specifically because it's the single most-repeated "real number" in cohort-course marketing: the underlying mechanism it's gesturing at (live cohorts with real classmates and a fixed schedule improve completion versus asynchronous self-paced content) is plausible and consistent with the accountability argument this course keeps returning to — but the specific magnitude is not established by a source this research could independently verify, and shouldn't be treated as a benchmark to plan against.

#5. What an honest completion-rate posture looks like for your own course

Given the state of the actual evidence: expect meaningfully more than Jordan's 12.6% MOOC median if your course is paid and has any structural accountability mechanism (a cohort, a deadline, drip-released content rather than an all-at-once dump, progress tracking) — the mechanism for that expectation is solid even where the exact number isn't. Don't adopt a specific industry-benchmark percentage as a target, because no source in this research earned the confidence to supply one. Track your own completion data from day one on whatever platform you're using — nearly every platform in module 2 reports this natively — because your own cohort's actual number is the only completion figure in this entire lesson that's genuinely [Established] for your specific business, and it's the input the kill-switch framework later in this module is actually built around.