The E-Learning Market Filtration Hypothesis: Evidence Against the "Dead Market" Narrative in Latin America

1. Introduction

1.1 The Prevailing Narrative

A consistent pattern emerges in the public discourse of Spanish-speaking online course creators. In trade publications, podcasts, and social-media posts, established creators describe sales declines of 30% to 50% relative to their 2020–2022 peaks, prospects who decline purchases citing access to ChatGPT, and growing skepticism about whether the online course business model has a viable future. The dominant framing of these observations, repeated across creator communities, is that the market itself is saturated, exhausted, or — in the most frequent formulation — dead.

This framing has acquired the status of common knowledge within the creator economy. It informs strategic decisions: creators reduce prices to compete with free generative-AI alternatives, postpone launches, expand advertising spend in response to declining conversion rates, or exit the category entirely. The narrative is rarely interrogated against aggregate market data; it is treated as self-evident, established by the visible experience of peers.

1.2 The Problem

If the narrative is correct, it justifies disinvestment from the category. If it is incorrect — or, more precisely, if it conflates a specific business-model failure with a broader market collapse — it produces a class of strategic errors with measurable economic consequences. Creators who reduce prices in response to a perceived saturation forfeit margin without addressing the actual mechanism of declining conversion. Creators who increase advertising spend pour additional traffic into funnels whose underlying conversion problem is structural rather than promotional. Creators who exit the category abandon a positioning that, under different operational assumptions, may remain commercially viable.

The cost of misdiagnosis, therefore, is not abstract. It accrues in real time, in the financial statements of working creators, and in the strategic latitude they perceive themselves to have. A clear empirical accounting of the market is a necessary precondition for any rational strategic response.

1.3 Research Question and Thesis

This paper examines whether the “dead market” narrative withstands empirical scrutiny when tested against (i) aggregate global and regional market data published by independent research firms; (ii) platform-level financial disclosures from publicly traded and privately held online-learning companies; and (iii) the cognitive-science literature on perception, salience, and bias.

We propose an alternative framework — the Market Filtration Hypothesis — under which the perception of contraction is the cognitive trace of a structural filtration process. Specifically, we hypothesize that aggregate market growth has continued unabated, but that the growth has been concentrated in segments and platforms that combine content with implementation infrastructure, while segments dependent on unstructured information delivery have stagnated or declined. The remainder of the paper proceeds as follows: Section 2 reviews the relevant literature on cognitive biases, the implementation gap, and historical analogues from other content industries. Section 3 presents aggregate, platform-level, and segment-level evidence. Section 4 concludes with limitations and directions for future research.

---

2. Literature Review

2.0 Methodological Note

This review synthesizes peer-reviewed empirical studies, organizational case reports, audited financial filings, and institutional market data published between 1973 and 2026, sourced from Google Scholar, PubMed, the U.S. Securities and Exchange Commission EDGAR system, and the published reports of established industry-research firms (Polaris Market Research, Precedence Research, Grand View Research, Allied Market Research, IMARC Group, Future Market Insights, and Arizton). Inclusion criteria prioritized studies and reports with measurable outcomes — completion rates, revenue growth, gross merchandise volume — and verifiable methodologies. Because the present-day creator-led online-course segment is underrepresented in peer-reviewed research, grey literature from established industry sources was included where peer-reviewed evidence was limited; such sources are explicitly identified throughout the paper.

2.1 The Availability Heuristic and the Subjective Estimation of Frequency

The cognitive mechanism underlying the “dead market” perception was first formalized by Tversky and Kahneman (1973), who demonstrated that subjects estimate the frequency of events not by reference to base-rate data but by the ease with which exemplars can be retrieved from memory. The subsequent program of research on judgment under uncertainty (Kahneman & Tversky, 1974) established this availability heuristic as a robust feature of human cognition. The mechanism is simple: vivid, recent, and emotionally charged exemplars produce inflated estimates of the underlying phenomenon’s frequency, while absent or unsalient evidence does not enter the calculation.

For online course creators, the relevant exemplars are the visible failures of peers, the testimonials of declining sales shared in trade communities, and the personal experience of underperforming launches. The relevant absent evidence is the aggregate financial performance of the platforms on which creators operate — which, as Section 3 will show, tells a different story.

2.2 The Negativity Bias

Compounding the availability heuristic is the negativity bias documented by Baumeister et al. (2001) in the canonical review Bad is Stronger than Good, published in the Review of General Psychology. The authors synthesize evidence across attention, memory, learning, emotion, and social judgment to argue that negative information is processed more thoroughly, retained more durably, and weighted more heavily than equivalent positive information. The asymmetry is robust across domains and populations.

In media-consumption contexts, Trussler and Soroka (2014) show empirically that consumers reveal a preference for negatively framed news content in behavioral tests, even when they explicitly deny that preference in survey self-reports. Soroka, Fournier, and Nir (2019), in a 17-country psychophysiological study published in the Proceedings of the National Academy of Sciences, demonstrate that the negativity bias in news reactions is cross-nationally robust, observed in arousal and valence responses even where survey self-reports vary. Together, these findings imply that negative narratives about market conditions enjoy structural amplification in media ecosystems independent of the underlying empirical state.

2.3 The Knowing-Doing Gap

A separate strand of literature, originating with Pfeffer and Sutton’s (2000) study of organizational performance, identifies a persistent gap between what individuals and organizations know and what they reliably do. The authors argue that improvements in performance depend less on the acquisition of new knowledge than on the implementation of knowledge that is already available. This finding generalizes from organizational behavior to consumer education contexts: the limiting factor in transformative outcomes is not access to information but the structural support required to translate information into action.

The knowing-doing gap finds its sharpest empirical expression in the literature on Massive Open Online Courses (MOOCs). Reich and Ruipérez-Valiente (2019), analyzing six years of data from the MITx and HarvardX programs in Science, report that MOOC completion rates declined from approximately 6% in 2014–2015 to 3.13% in 2017–2018 even as enrollments increased, and that a majority of enrollees never began the courses they registered for. The authors describe a “MOOC pivot” away from open access toward credential-bearing, fee-based, structured offerings, framed explicitly as a response to the failure of the original information-delivery model. By contrast, cohort-based course platforms report completion rates an order of magnitude higher; Maven (2021), the platform founded by Wes Kao and Gagan Biyani, reports completion rates of 75–96% across its cohort programs at launch. The empirical contrast — between an unstructured information-delivery format and a structured cohort format — is the clearest evidence in the literature that completion is a function of structure, not of the content being delivered.

2.4 Historical Analogues: Industries Misdiagnosed as Dead

The proposition that an industry can be widely declared dead while undergoing structural reorganization rather than collapse is supported by recent histories of three adjacent content industries.

The recorded music industry was repeatedly described as dying between 2000 and 2014, during which physical-album revenue collapsed under the joint pressures of digital piracy and unbundled single sales. The International Federation of the Phonographic Industry (IFPI, 2025) reports that global recorded-music revenue reached USD 29.6 billion in 2024 — the tenth consecutive year of growth — with streaming representing 69% of revenue and Latin America showing year-on-year growth of 22.5%. The industry was not dead; a specific business model (the packaged physical album) was dead, while the streaming and live-performance segments grew.

The newspaper industry presents a parallel pattern. Pew Research Center (2024) reports that U.S. weekday newspaper print circulation declined from 55.8 million in 2000 to 24.3 million in 2020. However, leading publications that successfully transitioned to paid digital subscription models — the New York Times and the Wall Street Journal among them — report combined digital and print circulation that has reversed the long-running decline since 2020. As with music, the contraction was concentrated in a specific business model (advertising-supported general-interest print) rather than in journalism as a category.

The U.S. cable television industry follows the same pattern. Multichannel pay-TV penetration declined from over 80% of U.S. households in 2011 to 34.4% by year-end 2024 (Adwave, 2025). What replaced cable was not the absence of paid video consumption but a filtered set of streaming subscriptions. The total addressable market for paid video persisted; the delivery mechanism filtered.

2.5 Research Gap

The literature documents three relevant phenomena: the cognitive mechanisms by which negative narratives outcompete base-rate data in subjective frequency estimates (§2.1, §2.2); the structural failure of unstructured information-delivery formats to produce completion or transformation (§2.3); and the historical pattern of industry filtration being misperceived as industry collapse (§2.4). What the literature does not yet address is whether the present-day Spanish-speaking online course market — a specific, bounded, and economically significant segment — exhibits the same filtration pattern. The remainder of this paper undertakes that empirical examination.

---

3. Analysis and Discussion

3.1 Aggregate Market Growth: The Base-Rate Evidence

The first test of the “dead market” narrative is whether independent industry research firms project contraction or growth for the global e-learning market. They do not.

Polaris Market Research (2025) values the global e-learning market at USD 439.92 billion in 2025 and projects USD 2.34 trillion by 2034, implying a compound annual growth rate of 20.4% (Polaris Market Research, 2025). Precedence Research (2025) values the e-learning services market at USD 313.65 billion in 2024 with a forecast of USD 2.04 trillion by 2034 at a CAGR of 20.6%. Grand View Research (2024) projects the market reaching USD 842.6 billion by 2030 at 19.0% CAGR. Spherical Insights (2024) forecasts USD 1.52 trillion by 2033 at 17.35% CAGR. Facts and Factors (2024) projects USD 1.08 trillion by 2032 at 17.58% CAGR. Allied Market Research (2024) projects USD 933.5 billion by 2032 at 14.8% CAGR. The most conservative estimate identified, from IMARC Group (2025), still implies a 7.65% CAGR through 2034.

The dispersion among these forecasts reflects methodological differences in segmentation — services-only versus content-plus-services, corporate versus consumer scope, regional inclusion thresholds — but no major industry-research firm projects contraction. The lower bound of the consensus is mid-single-digit growth; the upper bound exceeds 20%. By any reasonable interpretation, the global e-learning market is not contracting.

For Latin America specifically, Future Market Insights (2025) projects 14.5% CAGR through 2035 for the regional EdTech industry, with the highest national CAGR in Argentina at 16.2%. Arizton (2025) values the Latin American e-learning market at USD 30.51 billion in 2024, forecasting USD 52.10 billion by 2030 at a 9.33% CAGR. IMARC Group (2024) projects the Latin American online education market growing at 20.68% CAGR through 2033. Whichever segmentation is used, Latin American e-learning growth runs ahead of regional nominal GDP growth and well above the global average reported by Grand View Research.

The base-rate evidence is unambiguous: the market is not dead. The first proposition of the popular narrative is empirically false at the aggregate level.

3.2 Platform-Level Bifurcation: Where the Growth Concentrates

If the aggregate market is growing while creators perceive contraction, the data must show that the growth is unevenly distributed. It does.

In March 2024, Hotmart Company — the parent organization of the Hotmart and Teachable platforms, headquartered in Amsterdam — announced that creators on its platforms had collectively generated USD 10 billion in cumulative gross merchandise volume since 2011 (Hotmart Company, 2024). The same announcement reported more than 200,000 active creators globally and 21 million unique consumers in the trailing twelve months. Hotmart’s GMV from Latin American creators excluding Brazil grew 50% from 2022 to 2023. In Brazil, Hotmart estimates that one in five economically active Brazilians purchased a creator’s product through the platform in 2023, and a study with the Fundação Getúlio Vargas attributes more than 300,000 direct and indirect jobs to creator activity on the platform.

In August 2025, Kajabi announced that its creators had collectively earned USD 10 billion in revenue through the platform, having crossed the USD 9 billion threshold only months earlier (Kajabi, 2025). The company reports more than 100,000 active creators with average annual earnings of approximately USD 190,000 per active creator, approximately 1,800 creators with cumulative earnings exceeding USD 1 million, and more than seventy creators with cumulative earnings exceeding USD 10 million. These figures describe a creator-platform segment expanding rapidly.

Coursera, the publicly traded MOOC platform, reported full-year 2024 revenue of USD 694.7 million, with Consumer-segment revenue of USD 397 million (a 5% year-on-year increase) and Enterprise revenue of USD 234 million (a 7% year-on-year increase). The company reported 168 million registered learners at year-end 2024, up from 142 million at year-end 2023 (Coursera, 2025). The Coursera 2025 Global Skills Report identifies 27.8 million learners in Latin America, with cybersecurity enrollments in the region growing 106% year-on-year and Brazilian Professional Certificate enrollments growing 231% year-on-year.

The contrast with Udemy’s Consumer segment is sharp. Udemy reported full-year 2024 Consumer revenue of USD 292.1 million, a 5% year-on-year decline, against full-year 2024 Enterprise revenue of USD 494.5 million, an 18% increase (Udemy, 2025a). Quarterly Consumer revenue declined further in 2025, reaching USD 62.9 million in the third quarter — described by Class Central (2025) as Udemy’s lowest Consumer-revenue quarter since 2019. The company’s Q3 2025 earnings call disclosed that management was intentionally reducing single-course Consumer sales to migrate users toward a subscription model. Subscription revenue grew 43% year-on-year to USD 11.7 million in Q3 2025, with 294,000 paid Consumer subscribers compared with 156,000 in Q3 2024. Full-year 2025 Consumer-segment revenue was USD 265.8 million, a 9% decline (Udemy, 2026). The firm has, in effect, conceded that the unstructured single-course Consumer model is no longer commercially viable and is reorganizing its Consumer segment around recurring, structured access.

The platform-level data describe a clear bifurcation. Platforms whose creators bundle content with structured implementation — community access, scheduled cohorts, coaching, follow-up — report growth. The platform whose Consumer segment is most closely aligned with the unstructured pre-recorded-course model is contracting and explicitly migrating away from it. This is the Filtration Hypothesis at the platform level.

3.3 Pre-Pandemic Baseline and the Artificial 2020–2022 Spike

A subordinate claim within the “dead market” narrative is that current creator sales are catastrophically below the recent norm. The platform-level data and creator-side accounting suggest a different interpretation: current sales are at or near the pre-pandemic baseline, and what declined was the pandemic-era demand spike.

Marketing consultant Rachel Mazza (2024), reviewing the financial records of multiple online-course clients, reports that after detailed accounting, sales for the majority of her client base are “on par with pre-shutdown numbers” once the 2020–2022 spike is excluded from the baseline. Her observation is consistent with the Hotmart finding that COVID-19-era new creator sign-ups grew approximately 160% before normalizing (Hotmart Company, 2024).

A complementary signal comes from venture-capital flows. HolonIQ (2024) reports that global EdTech venture funding fell to approximately USD 2.4 billion in 2024 — the lowest level in a decade and an 89% decline from the 2021 peak of USD 20.8 billion. This contraction reflects a recalibration of speculative capital expectations following the pandemic-era surge, not a contraction in end-user demand. The two phenomena — venture-capital normalization and end-user market growth — are easily conflated, particularly by creators whose operating environment was shaped by pandemic-era marketing benchmarks.

The relevant comparison, then, is not 2024 versus 2021 but 2024 versus 2019. By that comparison, creator-economy platforms have grown substantially, and creator-side accounting suggests that individual creators’ sales have largely returned to the pre-pandemic baseline.

3.4 The Proposed Framework: The Market Filtration Hypothesis

The aggregate, platform-level, and segment-level evidence converge on a single empirical structure. Aggregate demand for online learning has continued to grow at double-digit compound rates. That growth has flowed disproportionately to platforms and creators who combine content delivery with implementation infrastructure: scheduled cohorts, coaching, structured progress tracking, personalized support. Conversely, segments built on unstructured pre-recorded content — exemplified by the Udemy Consumer segment and by individual creators distributing video libraries without implementation scaffolding — have stagnated or declined.

We summarize this structure as the Market Filtration Hypothesis:

The perception of contraction in the consumer-led online course market is the cognitive trace of a structural filtration process. The market did not contract; it filtered. Information-only delivery models lose viability. Implementation-supported delivery models continue to grow.

The hypothesis has three component claims, each separately testable:

1. The aggregate market continues to grow. Tested in §3.1: confirmed.
2. Growth is unevenly distributed across business models, with implementation-supported models growing and information-only models declining. Tested in §3.2: confirmed by platform-level financial disclosures.
3. The perception of contraction among creators reflects the salience of visible information-only failures rather than aggregate market behavior, amplified by availability and negativity biases. Tested in §2.1–§2.2 against the cognitive-science literature: consistent with the established mechanisms.

The hypothesis is structurally analogous to filtration patterns documented in recorded music (§2.4), newspapers (§2.4), and pay television (§2.4). In each case, an industry was widely declared dead during a transition that, in retrospect, proved to be a reorganization rather than a collapse. The salient business model died; the underlying demand did not.

3.5 Practical Implications and the CursoVivo Implementation Model

The Filtration Hypothesis, if accepted, reorganizes the strategic question facing creators. The relevant question is no longer whether the market has space — it manifestly does — but whether a given creator’s offering sits on the growing or the declining side of the filter. Operationally, the question becomes: does the creator’s product deliver content alone, or does it deliver content paired with the implementation infrastructure required to translate content into outcomes?

The empirical anchor for this question is the creator’s own completion rate. Industry benchmarks for unstructured pre-recorded courses cluster between 3% and 15% (Reich & Ruipérez-Valiente, 2019; industry surveys). Cohort-based and structured-implementation programs report completion rates of 75% to 96% (Maven, 2021). The operational gap is an order of magnitude wide and is the single most actionable diagnostic available to a creator.

The CursoVivo implementation model addresses this gap by embedding AI-driven implementation infrastructure within an existing course’s content. Rather than proposing that creators rebuild their offerings around new platforms or new content, the model treats the creator’s existing method, sequence, and materials as the asset to be implemented and adds a structured layer of weekly planning, memory-bearing follow-up, and per-learner deliverable production. The model is anchored in the empirical finding (§2.3) that completion is a function of structure rather than of content. Within the framework of the Filtration Hypothesis, the model’s purpose is to reposition a creator’s offering from the information-only side of the filter to the implementation-supported side, without requiring abandonment of the creator’s existing business architecture.

The intended object of CursoVivo, in the framing of this paper, is not to expand the volume of content available to learners — which generative AI now produces in functionally unlimited quantity — but to expand the structural support that translates available content into completion. The proposition rests on the prior literature: information abundance has shifted the binding constraint in adult learning from access to implementation, and the platforms that capture the next decade of e-learning growth will be those whose architecture acknowledges that shift.

---

4. Conclusions

4.1 Summary of Findings

The “dead online course market” narrative is empirically inconsistent with the available aggregate data, platform-level financial disclosures, and the cognitive-science literature on perception. Aggregate market data project compound annual growth in the global e-learning market of between 7.6% and 20.6% through the early 2030s. Latin American EdTech is projected to grow at 14.5% CAGR through 2035. Hotmart Company and Kajabi creators have each crossed the USD 10 billion cumulative-revenue threshold. Coursera’s learner base reached 168 million by year-end 2024. The platform whose Consumer segment most closely matches the unstructured information-delivery model — Udemy — has acknowledged the collapse of that segment by intentionally reducing single-course Consumer sales and migrating to subscription. The pattern is filtration, not contraction.

The persistence of the contraction narrative is consistent with two well-established cognitive mechanisms: the availability heuristic, under which subjective estimates of frequency track ease of retrieval rather than base rates (Tversky & Kahneman, 1973), and the negativity bias, under which negative information is processed and retained more durably than positive information (Baumeister et al., 2001) and is structurally amplified in media ecosystems (Soroka et al., 2019). The narrative’s persistence is therefore a feature of how creators perceive their environment, not of the environment itself.

What declined was not the market but a specific business model — the package-and-broadcast model — premised on the assumption that information delivery alone produces transformation. The model was made obsolete in part by generative AI’s commoditization of information access and in part by the structural completion failure documented in the MOOC literature. The growing segments of the market are those that supply what information alone cannot: implementation infrastructure that addresses the knowing-doing gap (Pfeffer & Sutton, 2000) at the level of the individual learner.

4.2 Limitations

This paper is a narrative literature review and a synthesis of secondary data; it does not present novel primary data. The aggregate market projections are produced by industry-research firms whose methodologies vary in segmentation and assumptions; the dispersion across firms is reported transparently in §3.1 but should be understood as a feature of the underlying methodological landscape. Platform-level financial figures from privately held companies (Hotmart Company, Kajabi) are self-reported and have not been independently audited in the public record; figures from publicly traded companies (Udemy, Coursera) are drawn from filings with the U.S. Securities and Exchange Commission and are subject to independent audit. The pre-pandemic-baseline claim relies in part on practitioner accounting (Mazza, 2024) and on the triangulation of platform data; an aggregate study comparing 2024 creator-market end-user revenue against 2019 baselines was not located in the present search and represents a clear opportunity for future quantitative work. The CursoVivo implementation model is presented as a logical extension of the literature on completion rates and the knowing-doing gap; controlled experimental validation of its specific mechanisms remains a direction for future research.

4.3 Future Research Directions

Three lines of further inquiry follow from the findings of this paper. First, longitudinal quantitative analysis of creator-segment revenue, controlled for the 2020–2022 demand spike, would allow direct testing of the pre-pandemic-baseline hypothesis at scale. Second, randomized comparisons of completion rates and learner outcomes between unstructured and AI-implementation-augmented variants of the same course content would establish the magnitude of the structure effect identified in §3.5. Third, cross-cultural extensions of the present analysis to the Lusophone (Brazilian Portuguese) and Spanish-speaking markets — given the disproportionate representation of these markets in Hotmart’s data — would refine the regional component of the Filtration Hypothesis. The intersection of these lines with the broader research program of the Department of Research, Arquitectos de la Felicidad represents the immediate empirical agenda following from this paper.

---

References

Adwave. (2025). How many U.S. households have cable TV? (Q3 2025). https://adwave.com/resources/us-cable-tv-households

Allied Market Research. (2024). E-learning market size, share, trends & forecast – 2032. https://www.alliedmarketresearch.com/e-learning-market-A06253

Arizton. (2025). Latin America e-learning market – Industry outlook & forecast 2025–2030. https://www.arizton.com/market-reports/latin-america-e-learning-market-size-share-analysis

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5(4), 323–370. https://doi.org/10.1037/1089-2680.5.4.323

Class Central. (2025). Udemy is “intentionally reducing” the business that made it famous. It’s not working. https://www.classcentral.com/report/udemy-subscription-pivot/

Coursera. (2025). Coursera reports fourth quarter and full year 2024 financial results. Coursera Investor Relations. https://investor.coursera.com/news/news-details/2025/Coursera-Reports-Fourth-Quarter-and-Full-Year-2024-Financial-Results/default.aspx

Facts and Factors. (2024). Global e-learning market size to hit USD 1,080.32 billion by 2032 at 17.58% CAGR. https://www.fnfresearch.com/news/global-e-learning-market

Future Market Insights. (2025). EduTech industry analysis in Latin America: Global market analysis report 2025–2035. https://www.futuremarketinsights.com/reports/edutech-industry-analysis-in-latin-america

Grand View Research. (2024). E-learning services market size, share & trends industry report 2024–2030. https://www.grandviewresearch.com/industry-analysis/e-learning-services-market

HolonIQ. (2024). EdTech VC reached ~USD 2.4B for 2024, representing the lowest level of investment in a decade. https://www.holoniq.com/notes/edtech-vc-reached-2-4b-for-2024-representing-the-lowest-level-of-investment-in-a-decade

Hotmart Company. (2024, March 21). Hotmart Company, home to Teachable, announces record-breaking USD 10 billion in global creator earnings [Press release]. PR Newswire. https://www.prnewswire.com/news-releases/hotmart-company-home-to-teachable-announces-record-breaking-10-billion-in-global-creator-earnings-302094911.html

IFPI. (2025). Global music report 2025: Amidst highly competitive market, global recorded music revenues grew 4.8% in 2024. International Federation of the Phonographic Industry. https://www.ifpi.org/ifpi-amidst-highly-competitive-market-global-recorded-music-revenues-grew-4-8-in-2024/

IMARC Group. (2024). Latin America online education market – Industry trends, share, size, growth, opportunity and forecast 2025–2033. https://www.imarcgroup.com/latin-america-online-education-market

IMARC Group. (2025). E-learning market size, share & industry analysis report 2025–2034. https://www.imarcgroup.com/e-learning-market

Kahneman, D., & Tversky, A. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131. https://doi.org/10.1126/science.185.4157.1124

Kajabi. (2025, August 6). USD 10B in creator revenue and climbing: Kajabi creators are achieving long-term financial success [Press release]. BusinessWire. https://www.businesswire.com/news/home/20250806424975/en/

Maven. (2021, May 26). Maven raises USD 20 million in Series A funding led by Andreessen Horowitz [Press release]. PR Newswire. https://www.prnewswire.com/news-releases/maven-raises-20-million-in-series-a-funding-led-by-andreessen-horowitz-301295905.html

Mazza, R. (2024). Are online courses dead? https://rachelmazza.com/are-online-courses-dead/

Pew Research Center. (2024). Newspapers fact sheet: State of the news media. https://www.pewresearch.org/journalism/fact-sheet/newspapers/

Pfeffer, J., & Sutton, R. I. (2000). The knowing-doing gap: How smart companies turn knowledge into action. Harvard Business School Press.

Polaris Market Research. (2025). E-learning market size, share & forecast 2034. https://www.polarismarketresearch.com/industry-analysis/e-learning-market

Precedence Research. (2025). E-learning services market size and forecast 2025 to 2034. https://www.precedenceresearch.com/e-learning-services-market

Reich, J., & Ruipérez-Valiente, J. A. (2019). The MOOC pivot. Science, 363(6423), 130–131. https://doi.org/10.1126/science.aav7958

Soroka, S., Fournier, P., & Nir, L. (2019). Cross-national evidence of a negativity bias in psychophysiological reactions to news. Proceedings of the National Academy of Sciences, 116(38), 18888–18892. https://doi.org/10.1073/pnas.1908369116

Spherical Insights. (2024). Global e-learning market size to worth USD 1,515.97 billion by 2033, CAGR of 17.35%. Yahoo Finance. https://finance.yahoo.com/news/global-e-learning-market-size-230000207.html

Trussler, M., & Soroka, S. (2014). Consumer demand for cynical and negative news frames. The International Journal of Press/Politics, 19(3), 360–379. https://doi.org/10.1177/1940161214524832

Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207–232. https://doi.org/10.1016/0010-0285(73)90033-9

Udemy. (2025a). Udemy reports fourth quarter and full year 2024 results. Udemy Investor Relations. https://investors.udemy.com/node/10421/pdf

Udemy. (2025b). Q3 2025 earnings press release [SEC Form 8-K]. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/0001607939/000160793925000138/q32025pressrelease.htm

Udemy. (2026). Udemy reports fourth quarter and full year 2025 results. Udemy Investor Relations. https://investors.udemy.com/news-releases/news-release-details/udemy-reports-fourth-quarter-and-full-year-2025-results