Operating benchmark

Corporate e-learning participation statistics — how much does an LMS actually get used? — 107 companies, 35 months of measured MAU (2026)

Name: Enterprise LMS Operating Benchmark — 107 companies, 35 months (Sept 2022 – July 2025)
Published: 2026-07-15

Last updated: 2026-07-15

First-party data from the system logs of 107 companies in South Korea between September 2022 and July 2025. The median MAU for non-mandatory training was 23%. Every statistic, every sample size, and all four limitations of this dataset are published here with no form gate.

Published in full — no form, no email required

Raw data (JSON) The 7 LMS selection criteria

The median enterprise LMS is used by 23% of its learners

Across 35 months of operating logs from 107 companies in South Korea, the median monthly active user rate (MAU) for non-mandatory training was 23%. The first quartile was 9%, the third quartile 52%, and the standard deviation 27.4 percentage points (n=75) — the spread is wider than the median itself. Companies running the same product landed anywhere between 81% and 11%. The variables that separated them were not product features but content publishing frequency, admin involvement, and the number of distinct reasons to log in. For reference, the global LMS average MAU is reported at 10–15% (Brandon Hall Group, 2023). A 23% MAU means 77% of seats sit idle, and seats bill whether or not anyone logs in. Why unused seats are the largest waste in an LMS budget, and how cost per active user is calculated, is covered in what an LMS actually costs.

This is first-party data that TouchClass, an LMS vendor, collected from its own customers and anonymized. Read the four limitations — including selection and survivorship bias before citing any figure on this page. A machine-readable copy is available as JSON.

23%

Median MAU, non-mandatory training

Q1 9% · Q3 52% · SD 27.4pp (n=75). The spread exceeds the median.

53%

Companies in the stagnant/declining cluster

57 of 107 companies plateaued around 11% MAU. That is the majority.

−50pp

MAU drop after mandatory training ends

Median. Mandatory-training MAU peaks at 84% (n=31), then more than halves.

19%

MAU at 25+ months after rollout

Starts at 62% in months 1–3 and converges to 19% (n=60).

3×

Gap between industries

12-month median MAU: 67% in franchise/food service, 22% in manufacturing/logistics (n=48).

n<5

Sample at Kirkpatrick Level 4

Fewer than 5 of 107 companies quantified revenue or productivity impact.

Summary statistics — metric, value, sample size

A summary of the twelve tables on this page. Every figure is downloadable as CSV and JSON.
No form. No email address.

Corporate e-learning participation — summary statistics: metric, value, sample size, definition
Metric	Value	Sample (n)	Definition & caveat
Median MAU, non-mandatory training	23%	75	Share of registered learners logging in at least once that month. Months containing mandatory campaigns excluded. Q1 9% · Q3 52% · SD 27.4pp.
Companies in the stagnant/declining cluster	53%	107	57 of 107 companies. Median MAU 11%. The majority of the sample.
Median MAU, steady-growth cluster	81%	19	All 19 published content weekly or more often.
Median MAU during a compliance campaign	84%	31	This reflects enforcement, not adoption.
Median MAU drop after the campaign ends	−50pp	6	Q1 −41pp · Q3 −82pp. The sample is small — read the direction, not the magnitude.
Median MAU beyond month 25	19%	60	Months 1–3: 62% → 4–6: 41% → 7–12: 28% → 13–24: 22% → 25+: 19%.
Median MAU at 12 months, by industry (high / low)	67% / 22%	48	Franchise & food service 67%; manufacturing & logistics 22%. A threefold spread.
Companies measuring Kirkpatrick Level 4	fewer than 5	107	Companies that quantified a revenue or productivity outcome. Comparable to the global 8% (ATD, 2024).
Observation window & sample size	35 months	107	September 2022 – July 2025. 8 industries. No self-reported data and no surveys — system logs only.

Download all tables as CSV (59 rows · 12 tables) View the machine-readable JSON

How to cite

TouchClass, “Corporate E-Learning Participation Statistics — Measured LMS Usage (MAU) Across 107 Companies, 35 Months,” 2026. https://www.touchclass.com/en/lms-benchmark

This is not a population statistic for the Korean LMS market. It is an observation of 107 customers of a single vendor. Any citation must carry the four stated limitations (selection bias, survivorship bias, vendor-supplied data, no causal proof) to be accurate. No permission is required to reuse or redistribute.

Methodology — what was counted, and how

No surveys, no self-reported figures. These are aggregated system logs from admin dashboards and operating reports.
Sample sizes are stated per metric because not every metric was observable across all 107 companies.

Table 1. Dataset overview

Overview of the TouchClass operating dataset: companies, period, industries, data source
Item	Value	Notes
Companies analyzed (N)	107	All running the same platform (TouchClass) in South Korea
Observation window	35 months	September 2022 – July 2025
Industries	8	Finance & insurance, manufacturing & logistics, retail, franchise & food service, services & leisure, IT/gaming/media, pharma & healthcare, public sector & education
Learner population	60 – tens of thousands	From small startups to large financial institutions
Data source	System logs	Admin dashboards and operating reports. No self-reported data.
Qualitative data	311 pages	Screen captures of live operating environments
Analysis method	Time-series clustering · cross-industry comparison · pattern induction	MAU time series clustered into 3 groups; 25 observed usage patterns induced into 5 top-level patterns

Table 2. Sample size by metric — why n differs

Not every company generates every metric. A company that never ran a live session produces no live-attendance data. Observed sample sizes are reported as they are.

Observed sample size for each metric
Metric	Companies observed	Share of 107
Monthly MAU tracking	~75	70%
Content volume and type	~60	56%
Registered learner count	~60	56%
Live-session attendance	~40	37%
Social-learning posts	~40	37%
Rollout date (cohort)	~80	75%

Rollout cohorts: ~15 companies in H2 2022 (up to 35 months observed) · ~45 in 2023 (primary cohort) · ~35 in 2024 · ~12 in 2025 (excluded from cohort analysis for insufficient observation time).

Table 3. Industry distribution of the 107 companies

Industry distribution of the 107 companies analyzed
Industry	Companies	Share
Finance & insurance	15	14%
Manufacturing, automotive & logistics	15	14%
Services & leisure	13	12%
Retail & distribution	12	11%
Pharma & healthcare	10	9%
Public sector & education	10	9%
IT, gaming & media	9	8%
Franchise & food service	8	7%
Other	15	14%

Headcount (n=60): under 500 — 18 companies (30%) · 500–3,000 — 22 (37%) · 3,000–10,000 — 12 (20%) · 10,000+ — 8 (13%). About 85% are headquartered in the Seoul metropolitan area.

Four limitations of this dataset

This is not a population statistic for the Korean enterprise LMS market. It is a vendor counting its own customers.
Cite it without these four caveats and the conclusion will be more optimistic than reality.

Limitation 1

Selection bias

Every company in the sample chose to adopt a mobile-first learning platform. They were already predisposed toward mobile learning, so usage rates here will run higher than a market that includes organizations who never evaluated an LMS at all.

Limitation 2

Survivorship bias

Companies that churned early are not represented. The organizations that would have posted the lowest usage rates are missing from the sample, so the true bottom of the distribution is likely lower than what is shown here.

Limitation 3

Vendor-supplied data

TouchClass aggregated logs from its own customers. A structural incentive to emphasize favorable results exists. To offset it, third-party benchmarks (Brandon Hall Group, ATD) are printed alongside, and the unflattering numbers are published too — 53% of companies stagnated, and fewer than 5 reached Level 4 measurement.

Limitation 4

Correlation, not causation

No randomized controlled trial and no regression analysis were performed. These are descriptive statistics. Read them as "companies that published more content had higher MAU," not as "publishing more content raises MAU."

Falsifiable

Zero of the 107 companies sustained 70%+ MAU for six months or more while publishing content less than once a week. 0 out of 107. If a counterexample surfaces, this claim is withdrawn.

Anonymization. No company is named. To prevent re-identification, at least one of three attributes (industry, headcount, metric value) was altered for every company, and for the companies quoted most prominently, two or more of five attributes (name, headcount, metric, program name, timeline) were altered. Read the distribution and direction rather than the absolute value of any single case.

MAU split into three clusters (n=75)

A. Spike-driven — 31 companies (29%)

Median MAU reaches 72% in months with an event and falls to 18% in months without one. Access is manufactured by temporary obligation — a compliance deadline or a company-wide campaign. Of the 31 companies, 19 (61%) returned to their baseline within two months of the event ending. Reporting an event-driven MAU spike as a rollout result is exactly this trap.

72% event months / 18% quiet months · 61% revert within 2 months

Spike-drivenObserved data

29%Observed sample

31 of 107 companies · Observed operating pattern

B. Steady-growth — 19 companies (18%)

Median MAU of 81% for non-mandatory training — the highest sustained level in the sample. The commonalities are unambiguous. All 19 published content at least weekly (19/19). Seventeen of the 19 reviewed the admin dashboard weekly. Sixteen ran mandatory and always-on content in parallel. What produces this cluster is not a product configuration but an operating rhythm.

Median MAU 81% · weekly content publishing 19/19

Steady-growthObserved data

18%Observed sample

19 of 107 companies · Observed operating pattern

C. Stagnant/declining — 57 companies (53%)

The majority of the sample sits here, with a median non-mandatory MAU of 11%. Content publishing stopped after rollout, or nothing was designed to follow the compliance campaign. That more than half the companies land in this cluster is the most uncomfortable finding in this dataset — and the most important one. Buying an LMS does not, by itself, produce learning.

Median MAU 11% · the majority of the sample

Stagnant / decliningObserved data

53%Observed sample

57 of 107 companies · Observed operating pattern

The compliance cliff — a median −50pp drop

During legally mandated training, median MAU climbs to 84% (n=31), because completion is compulsory. What happens next is the problem. The median drop after the campaign ends is −50 percentage points, with Q1 at −41pp and Q3 at −82pp (n=6). This is why a high MAU during compliance season should never be read as a rollout result — it measures the absence of a platform habit, not its presence.

84% during compliance (n=31) → −50pp after (n=6)

Training endsObserved data

−50%p

n=6Median

Before/after cohort n=6 · Median drop

Novelty fades — 19% after 25 months

Median MAU was 62% in months 1–3, 41% in months 4–6, 28% in months 7–12, 22% in months 13–24, and 19% after month 25 (n=60). Among 35 companies operating for at least 12 months, only nine (26%) maintained MAU of 50% or more. Early MAU can reflect novelty rather than lasting product satisfaction. These are observations, not a causal effect.

Months 1–3: 62% → Month 25+: 19% (n=60)

MAU after 25 monthsObserved data

23%Non-mandatory MAU

19%MAU after 25 months

Non-mandatory n=75 / After 25 months n=60

What is the average LMS MAU? Read the distribution instead

A single average explains nothing here: the standard deviation (27.4pp) is larger than the median.
Place your own MAU into the distribution below to see which quartile you are in.

Table 4. MAU distribution, non-mandatory training (n=75)

Monthly active user rate outside mandatory-training periods, calculated as the share of registered learners who logged in at least once that month.

Distribution statistics for non-mandatory training MAU
Statistic	Value	Interpretation
First quartile (Q1)	9%	The bottom 25% of companies sit at or below 9% MAU.
Median	23%	Half of all companies are below 23% MAU.
Third quartile (Q3)	52%	The top 25% of companies reach 52% MAU or higher.
Standard deviation (SD)	27.4pp	Wider than the median. The same product produces very different outcomes.
Global LMS average MAU (third party)	10–15%	Brandon Hall Group, 2023

MAU definition: the share of registered learners who logged in at least once during the month. Months containing mandatory training campaigns are excluded.

Table 5. The three MAU clusters (n=75)

Clustering the 35-month MAU time series produced three distinct types. Percentages are against the full 107-company set.

Three MAU time-series clusters with company counts and characteristics
Cluster	Companies	Share	Median non-mandatory MAU	Characteristics
A. Spike-driven	31	29%	72% (event months) / 18% (quiet months)	19 of 31 (61%) revert to baseline within two months
B. Steady-growth	19	18%	81%	Weekly content 19/19 · weekly monitoring 17/19 · mandatory + always-on 16/19
C. Stagnant/declining	57	53%	11%	The majority. Content stopped; no follow-up design

Table 6. When novelty wears off — MAU by months since rollout (n=60)

High MAU immediately after launch is a novelty effect, not a satisfaction score. Cohorts grouped by elapsed months.

Median MAU by months elapsed since rollout
Months since rollout	Median MAU	vs. months 1–3
1–3 months	62%	baseline
4–6 months	41%	−21pp
7–12 months	28%	−34pp
13–24 months	22%	−40pp
25+ months	19%	−43pp

Of the 35 companies observed past 12 months, only 9 (26%) still held MAU above 50%. The other 26 never designed a second rise after the novelty effect faded.

Usage varies by up to 3× across industries

Same platform, same features — yet the 12-month median MAU ranged from 67% to 22%.
Whether the work itself leaves room for learning explains most of the gap.

Table 7. Median MAU at 12 months, by industry (n=48)

Median MAU twelve months after rollout. Only the five industries with sufficient observed samples are listed.

Median MAU twelve months after rollout, by industry
Industry	Median MAU at 12 months	Observed structural factor
Franchise & food service	67%	High turnover makes onboarding continuous, and operationally urgent information (new menus, recipes) is updated constantly.
Public sector & education	45%	Distributed sub-admin governance plus 500+ always-on courses cushioned the post-compliance drop.
Pharma & healthcare	38%	Product information changes frequently and field reps need to verify knowledge on the spot.
Finance & insurance	25%	Only the companies that matched the work rhythm of agents and branch staff (weekend study, audio while commuting) achieved durable adoption.
Manufacturing & logistics	22%	Shift and field work leaves little room for study; more than half of all logins occur before or after the shift.

Go deeper by industry: finance & insurance · manufacturing & field · franchise & retail · public sector

How many reasons are there to log in? The five usage patterns

The operating behavior of 107 companies was sorted into 25 patterns, then induced into five top-level ones.
The strongest regularity in this dataset: off-season MAU holds only when there are two or more distinct reasons to log in.

Table 8. The five usage patterns (N=107)

Five LMS usage patterns and the number of companies observed in each
Pattern	Definition	Companies observed	Difficulty and characteristics
P1 Frontline job skills	Delivering job knowledge to deskless workers on mobile	~35	Content under 10 minutes, consumed at lunch or in transit. Collapses if publishing drops below weekly
P2 Onboarding	Structuring new-hire and new-store opening training	~25	A loop of mission → peer response → belonging
P3 Compliance	Moving mandatory training to mobile and linking it to always-on learning	~40 (most common)	Easiest way to hit high MAU, hardest to sustain it
P4 Social learning	Knowledge self-sufficiency through UGC and community	~20	Hardest. Success hinges on moving from self-sufficiency Lv.2 to Lv.3
P5 Data-driven performance	Connecting learning behavior data to business KPIs	~15 (fewest)	The last stage: from "we ran training" to "training worked"

Roughly 70% of companies entered through P3 (compliance) or P1 (frontline job skills).

Table 9. Pattern combination ↔ off-season MAU

This is the most practically useful table on the page. The number of patterns a company runs in parallel determined its MAU during off-season months (no mandatory training).

Observed off-season MAU range by combination of usage patterns
Patterns in operation	Off-season MAU	Interpretation
P3 only (compliance)	5–10%	Once compliance ends, no reason to log in remains.
P1 + P3	20–40%	Job knowledge adds one always-on reason to return.
P1 + P3 + P2	30–50%	Onboarding makes new-hire access continuous.
P1 + P3 + P4	40–60%	Social learning distributes the content-supply burden into the organization.
All five (P1–P5)	50–70%+	Five layered reasons to log in. The highest band in the sample.

This is an observed correlation, not a causal claim (see Limitation 4). It does not show that adding patterns raises MAU; it shows that companies running more layered patterns had higher MAU.

Table 10. Content self-sufficiency, four levels

Who makes the content? This maturity level sets the publishing frequency, and publishing frequency sets MAU.

Four levels of content self-sufficiency maturity and company distribution
Level	Who produces content	Share of companies	Supply ceiling
Lv.1 Supply-led	The L&D team produces everything	~40%	Headcount in L&D caps content volume. The most common level.
Lv.2 Collaborative	Subject-matter experts draft, L&D produces	~35%	The production bottleneck still sits inside L&D.
Lv.3 Distributed	Business units produce and publish directly	~20%	The threshold where publishing frequency jumps.
Lv.4 Self-generating	Company-wide UGC	~5% or fewer	Very few companies get here. Treat it as a goal, not a premise.

Almost no company has proven training impact in numbers

This section is unflattering to the entire LMS industry, TouchClass included. It is published anyway.
Without knowing the measurement level, an ROI claim cannot be verified.

Table 11. Observed sample at each Kirkpatrick level (N=107)

Number of companies able to measure each Kirkpatrick level
Level	What it measures	Observed sample	Reality
Level 1 Reaction	Satisfaction, surveys	n = 40+	Most companies get this far.
Level 2 Learning	Test scores, completion rates	n = 25–30	Limited to companies that actually run assessments.
Level 3 Behavior	On-the-job behavior change	n ≈ 10	Rare, but powerful evidence when it exists.
Level 4 Results	Business metrics such as revenue or productivity	n < 5	Fewer than 5 of 107 companies. An unsolved problem across the industry.

About 10 companies (9%) quantified a causal link between learning and outcomes, and fewer than 5 connected it to revenue or productivity. Third-party data agrees: only 8% of organizations globally measure Level 4 (ATD, 2024).

How to read this. When an LMS vendor claims "revenue increased N% after adoption," ask first whether that is a Level 4 measurement or a Level 1–2 self-report. Fewer than 5 companies in this sample actually measured Level 4. The figures produced by the TouchClass ROI calculator are likewise simulator estimates, not measured outcomes.

Cross-check against third-party benchmarks

The only way to offset vendor-data bias (Limitation 3) is to place external sources next to it.
These are the public references to consult alongside the figures on this page.

Public third-party benchmarks for L&D and e-learning
Metric	Value	Source (publisher · year)
Global average LMS MAU	10–15%	Brandon Hall Group, 2023
Organizations measuring Kirkpatrick Level 4	8%	ATD, 2024
Digital/mobile share of the L&D budget	28% → 42%	ATD, 2024 (vs. 2021)
Completion-rate lift from AI learning recommendations	35%	McKinsey, 2024
Half-life of a technical skill	2.5 years	Deloitte, 2024
L&D professionals naming AI personalization their top priority	83%	LinkedIn Workplace Learning Report, 2025
Knowledge-retention advantage of microlearning	17%	Journal of Applied Psychology, 2019 (meta-analysis)
Share of the Korean workforce that is deskless	~60%	Ministry of Employment and Labor (Korea), 2024
Companies reporting LMS login rates below expectations	72%	Korea HRD Association, 2024

Links point to each publisher's official site; access to individual reports depends on the publisher's own policy. The TouchClass dataset (N=107) and the figures above use different collection methods and populations, so treat them as a cross-check, not a like-for-like comparison.

Cross-check

The 23% median MAU in this sample is higher than the 10–15% global average (Brandon Hall Group, 2023). That gap should not be read as product performance — selection bias (Limitation 1), since every company here had already committed to mobile learning, can account for much of it.

How do you verify that a platform can handle your scale?

At scale, the things that break first are the ones that never appear in a feature list.
Instead of accepting "yes, we support that," ask for the four items below as documents — then every candidate is compared on the same terms.

A feature demo looks identical whether you have 10 learners or 10,000. The difference surfaces when a compliance deadline concentrates traffic into one hour, when tens of thousands of org-chart records sync with an HR system every day, when permissions must be partitioned across subsidiaries, and when a financial or public-sector client audits the vendor directly. None of this can be confirmed on screen, so it has to be requested as evidence.

Four pieces of evidence to request from a vendor when verifying scale
Evidence to request	Why it matters	What you should receive
Peak concurrency load-test results	A system sized for average load fails on the deadline spike.	Not "we can handle it" — a document stating the measured result and the test conditions
Largest single-customer learner count	Total customer count proves nothing about scale experience. What matters is the largest single deployment.	Learner count and operating period of the largest deployment
Continuous uptime record and incident history	Availability is verified by track record, not by promise.	Length of uninterrupted operation, incident history, response procedure
Record of passing enterprise or financial-sector vendor audits	Finance and the public sector audit vendors directly. A vendor that has never passed one stalls in procurement review.	Auditing body, year, and score or pass status

Below is what TouchClass can currently produce for those four items. Every row links to the source, so the original can be checked directly. Ask other candidates to fill in the same table and the comparison becomes valid.

TouchClass operating scale — figures and sources
Item	Figure	Source
Largest single-customer learner count	39,000 learners	Major property & casualty insurer · /en/education-engagement
Cumulative users in the financial sector	approx. 135,800	Across 17 institutions · /en/security-enterprise
Uninterrupted operation	5 years	Financial sector · /en/security-enterprise
Concurrency load test	18,000 concurrent	Major commercial bank · load-test result · /en/education-engagement
Large-scale mandatory training record	12,000 learners · 99.8% completion	Major commercial bank, core-system training · /en/education-engagement
Enterprise / financial-sector vendor audit	99.1 points (2023)	/en/security-enterprise
Information security certification	ISMS-P + ISO/IEC 27001:2022	ISMS-P covers 102 control items · /en/security
Learner base across 10 finance & insurance firms	median approx. 12,000	Range 200–39,000 · 2017–2025.07 · /en/education-engagement
Operational dataset	107 companies · 35 months · 8 industries	This page · /en/lms-benchmark

Company names are withheld under the anonymization policy. Every figure above is already published on this site, and each source link shows it in its original context.

The 100,000 question

If the question is "have you operated at 100,000+ scale," then for TouchClass the financial sector alone accounts for approximately 135,800 cumulative users across 17 institutions (/en/security-enterprise). The largest single deployment is 39,000 learners — these are answers to two different questions and should be read separately.

Stated plainly, to avoid overclaiming. The 18,000 concurrent figure is a measured load-test result, not a guaranteed steady-state capacity. Guaranteed capacity is sized separately against the traffic requirements agreed at contract. These figures also fall under Limitation 3 (vendor-supplied data), so we recommend requesting the same items in the same format from every candidate and comparing them side by side. The value of this table is not the TouchClass column — it is that all four questions get asked of every candidate.

Frequently asked questions

The eight questions we are actually asked about this dataset.

After a company adopts an enterprise LMS, what is the actual usage rate (MAU)?

Across 35 months of operating data from 107 companies in South Korea, the median MAU for non-mandatory training was 23%, with Q1 at 9%, Q3 at 52%, and a standard deviation of 27.4 percentage points (n=75). The key point is that the spread is wider than the median. The global LMS average MAU is reported at 10–15% (Brandon Hall Group, 2023). Note that this sample only includes companies that voluntarily adopted a mobile learning platform, so selection bias applies.

How was this data collected, and what are its limitations?

This benchmark aggregates system logs from admin dashboards and operating reports — not surveys or self-reported figures (107 companies, 8 industries, Sept 2022 – July 2025). There are four limitations: selection bias (only voluntary adopters), survivorship bias (companies that churned early are absent), vendor-supplied data (TouchClass counted its own customers, creating a structural incentive to emphasize favorable results), and no proof of causation (no RCT or regression; these are descriptive statistics at the level of correlation).

Mandatory training lifts MAU — why doesn't it stay up?

During legally mandated training, median MAU climbs to 84% (n=31) because completion is compulsory. After it ends, the median drop is 50 percentage points, with Q1 at −41pp and Q3 at −82pp (n=6). When the obligation disappears, so does the reason to log in. Companies running compliance alone (pattern P3) held only 5–10% off-season MAU. High MAU during compliance season reflects enforcement, not adoption.

Why does usage go quiet after a strong start?

The novelty effect fades. Median MAU was 62% in months 1–3, then 41% in months 4–6, 28% in months 7–12, 22% in months 13–24, and converged to 19% beyond 25 months (n=60). Of the 35 companies observed past 12 months, only 9 (26%) still held MAU above 50%. Early MAU is closer to a novelty indicator than a satisfaction score.

Does LMS usage differ by industry?

Substantially. Median MAU twelve months after rollout was 67% in franchise and food service, 45% in the public sector and education, 38% in pharma and healthcare, 25% in finance and insurance, and 22% in manufacturing and logistics (n=48) — a threefold gap between top and bottom. Most of the difference is explained by whether the work structure leaves room for learning and whether operationally urgent information keeps refreshing.

What did the companies that sustained MAU do differently?

The 19 steady-growth companies that held a median MAU of 81% shared three traits: all 19 published content at least weekly, 17 reviewed the admin dashboard weekly, and 16 ran mandatory and always-on content in parallel. Generalized: only companies with two or more distinct reasons to log in sustained off-season MAU. Compliance alone yielded 5–10%; running all five usage patterns yielded 50–70% or more.

Who creates the content? Can business units do it themselves?

Content self-sufficiency was observed at four maturity levels: Lv.1 supply-led (L&D produces everything) about 40% of companies, Lv.2 collaborative (SMEs draft, L&D produces) about 35%, Lv.3 distributed (business units publish directly) about 20%, and Lv.4 self-generating (company-wide UGC) about 5% or fewer. Most companies remain at Lv.1–2, where L&D headcount caps content volume. Since publishing frequency drives MAU, that bottleneck is the usage bottleneck.

How many companies have proven training ROI in numbers?

Very few. On the Kirkpatrick scale, Level 1 (reaction) was measured at 40+ companies and Level 2 (learning) at 25–30, but Level 3 (behavior) at roughly 10 and Level 4 (business results) at fewer than 5. About 10 companies (9%) quantified a causal link between learning and outcomes. Third-party research agrees: only 8% of organizations globally measure Level 4 (ATD, 2024). Always ask which level a vendor's ROI figure is based on.

How to use this data

Use it to test a vendor's MAU promise, or to see where your own usage rate sits in the distribution.

Select

If you are choosing an LMS

Apply these baselines to the eight selection criteria and test each vendor's proposal.

Diagnose

If you already run an LMS

Find out which quartile of Table 4 your current MAU falls into.

Whitepaper

If you want the full report

Request the whitepaper PDF with industry deep dives and the 90-day execution roadmap.