Sunday L&D Myth #4: People Forget 70% of What They Learn Within 24 Hours

This is the fourth in a series examining myths that circulate in the learning and development space. If you’ve got myths you’d like me to tackle, send them my way.

I’ve sat in many meetings and conference talks where someone has confidently stated that “people forget 70% of what they learn within 24 hours.” Each time, this statistic was used to justify expensive reinforcement programmes, elaborate spaced learning systems, or extensive follow-up interventions. Each time, the room nodded along as if this were an established fact rather than a distortion of 140-year-old research on memorising nonsense.

The problem isn’t just that the statistic is wrong; it’s that this myth misshapes how we think about workplace learning. It creates false urgency around memory retention when we should be focused on performance improvement, drives organisations to invest in reinforcement for content that never needed memorising in the first place, and perpetuates the confusion between “can’t remember” and “can’t do the job.”

Understanding where this claim actually comes from, what the science tells us about forgetting, and when memory retention actually matters can save organisations substantial money whilst improving learning outcomes. That seems worth 15 minutes of your Sunday morning.

The appeal of a simple answer

I understand why this myth persists. It sounds scientific, offers a memorable number for presentations, and provides a compelling explanation for why training doesn’t seem to stick. When we’re asked to justify follow-up programmes or reinforcement budgets, having a statistic that suggests catastrophic knowledge loss within a day makes our case considerably easier. It transforms “I think we should do more training” into “the science says we must do more training.”

The myth also taps into something we all recognise: we do forget things. You’ve experienced sitting through training and struggling to recall details days or even hours later. The 70% figure feels intuitively plausible because forgetting is real, common, and frustrating. A simple statistic that explains this experience, whilst appearing to be backed by research, is tremendously appealing, even when it’s wrong.

The claim has been repeated so often, in so many L&D books, conference presentations, and blog posts, that it’s achieved the status of conventional wisdom. Once something becomes “what everyone knows,” questioning it feels pedantic. But sometimes what everyone knows is precisely what needs examining most carefully.

What Ebbinghaus actually studied

Hermann Ebbinghaus conducted his memory experiments between 1879 and 1885, but what he studied bears almost no resemblance to workplace learning. He deliberately created 2,300 nonsense syllables, three-letter combinations like “WID,” “ZOF,” and “BOK,” specifically to eliminate any influence from prior knowledge, meaning, or associations (Murre and Dros, 2015). For seven months, he tested only himself, memorising lists of these syllables until he could recite them perfectly, then testing his retention at intervals from 20 minutes to 31 days.

The methodology reveals immediately why extrapolating to training is flawed. Ebbinghaus wasn’t measuring whether people could recall or recognise information; he was measuring what he called “savings,” which is how many fewer repetitions were needed to relearn material (Roediger, 1985). If initial learning required 20 repetitions and relearning needed 15, that’s 25% savings. This is categorically different from testing whether someone can recall information, and you cannot legitimately convert savings scores into “percentage forgotten.”

Even more telling, Ebbinghaus himself acknowledged his findings might not generalise. He noted that meaningful material like poetry required “one-tenth the effort” of memorising nonsense syllables, and his equations, he wrote, “have here no other value than that of a shorthand statement of the above results which have been found but once and under the circumstances described” (Roediger, 1985). He was notably cautious about the scope of his findings; it’s subsequent generations who have been reckless with them.

The 2015 replication study by Murre and Dros successfully confirmed the curve’s shape for nonsense syllables but found evidence of a memory boost at 24 hours, possibly due to sleep consolidation, rather than catastrophic 70% loss (Murre and Dros, 2015). Modern research consistently shows that Ebbinghaus’s curve accurately describes forgetting of meaningless material under artificial laboratory conditions; it does not describe how people forget meaningful information in more natural contexts.

The distortion cascade

I’ve not been able to find a primary source that directly links Ebbinghaus’s data to the specific claim that “people forget 70% within 24 hours.” What appears to have happened is a gradual distortion over decades: Ebbinghaus’s exponential decline curve was simplified in early 20th-century psychology textbooks, morphed into “people forget X% in Y time” claims by mid-century, and became corporate training conventional wisdom by the 1980s and 2000s (Thalheimer, 2010). Today, L&D blogs, eLearning vendors, and training magazines repeat variations without citations, some claiming 70% in 24 hours, others 90% in a week, still others 50% in an hour, each seemingly more alarming than the last.

Will Thalheimer, a learning researcher who conducted the most comprehensive review of actual forgetting research, analysed 69 separate experimental conditions representing over 1,000 learners. His finding demolishes the myth entirely: people forgot anywhere from 0% to 94% depending on context (Thalheimer, 2010). In the restricted timeframe of one to two days alone, forgetting ranged from 0% to 73%. Even within single experiments studying the same materials, variation was enormous. Rules of thumb showing people forgetting at some predefined rate are, as Thalheimer concludes bluntly, “just plain false.”

Forgetting rates depend dramatically on how meaningful the material is, how much prior knowledge someone has, how well it connects to existing schemas, how the learning occurred, and whether the information gets used. A universal percentage cannot possibly capture this complexity.

How meaningful learning defies simple curves

Modern cognitive science reveals why Ebbinghaus’s curve doesn’t describe real learning: memory isn’t mechanical reproduction, it’s semantic reconstruction. Frederic Bartlett’s 1932 research demonstrated that we don’t store disconnected facts like files in a cabinet; instead, we integrate information into webs of understanding, what cognitive psychologists call schemas, and reconstruct memories using existing knowledge frameworks (Baddeley et al., 2022).

David Ausubel’s meaningful learning theory established that when new information connects to prior knowledge, it becomes “anchored” in cognitive structures and doesn’t require constant revision to be retained (Ausubel, 1968). The duration of retention vastly exceeds rote-learned information. Daniel Willingham’s principle captures this elegantly: “Memory is the residue of thought” (Willingham, 2003). What people think about during learning determines what they remember, not just exposure to material. Understanding, as he notes, is remembering in disguise.

Schema theory research reveals the mechanism. When appropriate schemas are available, information can be consolidated within 24 hours, dramatically faster than for unrelated material (van Kesteren et al., 2022). Schemas reduce cognitive load, provide organisational structure, and enable faster integration. Information that fits existing schemas undergoes what researchers call accretion, easy incorporation into existing knowledge; information that partially fits requires tuning, adjustment of schemas; and fundamentally challenging information triggers restructuring, conceptual change (Brod et al., 2021). Each of these processes creates more durable memories than attempting to memorise isolated facts.

The role of prior knowledge cannot be overstated. Research in STEM education demonstrates that prior knowledge is essential for processing and retention of new information; learning is constructive, with much of what people learn built on prior knowledge in long-term memory (Brod et al., 2021). This creates dramatic individual variation: experienced employees retain far more from the same training than novices precisely because they possess relevant schemas for integration. The “70% forgotten” myth ignores this entirely, assuming uniform decay rates regardless of learner background.

The real culprit: interference and retrieval failure

John McGeoch’s 1932 research challenged simple time-based decay theories, establishing that time itself doesn’t cause forgetting; interference does (Wixted, 2004). New learning interferes with old learning, making some memories harder to retrieve. Robert Bjork’s New Theory of Disuse distinguishes storage strength, how well embedded knowledge is, from retrieval strength, how accessible it currently is (Bjork and Bjork, 1992). Storage strength can remain intact even when retrieval strength weakens dramatically. Forgetting, in this framework, is often temporary inaccessibility, not permanent loss.

This explains why Ebbinghaus-inspired “reinforcement” often fails. Simply re-exposing people to material doesn’t address interference or strengthen retrieval pathways. Robert and Elizabeth Bjork’s desirable difficulties research reveals a counterintuitive principle: conditions that make performance improve rapidly during training often fail to support long-term retention and transfer, whereas conditions that create appropriate challenges often optimise long-term retention (Bjork and Bjork, 2011). This performance-learning distinction is crucial. What appears to be excellent learning during training, high test scores and confident learners, may indicate only temporary retrieval strength, not durable storage strength.

Retrieval practice, actively recalling information rather than restudying, is far more powerful than the repetition strategies the forgetting curve myth encourages. Henry Roediger’s extensive research shows that testing enhances learning more than additional study, even when retrieval attempts are unsuccessful (Roediger and Karpicke, 2006). Each retrieval episode strengthens and can modify the memory trace, integrating it with related knowledge. Through the act of retrieval, our memory for that information is strengthened and forgetting becomes less likely to occur.

The spacing effect, one of the most robust findings in learning science, demonstrates optimal timing for these retrieval opportunities. Thalheimer’s comprehensive review of over 100 studies confirms that spacing repetitions over time facilitates long-term retention far more than massed practice (Thalheimer, 2010). The ideal spacing interval should roughly equal the retention interval, the time until information is needed on the job. Foreign language research found that 13 repetitions spaced at 56-day intervals yielded the same retention as 26 repetitions spaced at 14 days, cutting required practice in half through optimal timing (Bahrick and Hall, 2005).

Interleaving, mixing practice of different topics rather than blocking them, creates another desirable difficulty. Research on learning artistic styles found interleaving improved identification ability despite learners believing blocked practice was more effective (Kornell and Bjork, 2008). This metacognitive illusion is dangerous: what feels like effective learning, easy and fluent processing, often produces fragile knowledge, while approaches that feel challenging produce durable learning. Mathematics studies found interleaved practice produced 63% correct on delayed tests versus 20% for blocked practice (Rohrer and Taylor, 2007).

The economic cost of myth-driven practice

The “70% forgotten in 24 hours” myth drives massive resource waste across our industry. When we believe that without intervention, 70% of training content will vanish within 24 hours, we implement multiple touch points, reminder emails, follow-up modules, and spaced reinforcement schedules, often for content that doesn’t require memorisation at all. A one-size-fits-all approach emerges, ignoring that forgetting rates vary wildly based on meaningfulness, prior knowledge, and whether the task is performed regularly.

More insidiously, the myth perpetuates focus on memory retention rather than workplace performance. Organisations measure knowledge retention, can employees recite information, rather than performance impact, can employees do their jobs better. This confusion between memory and performance leads to training designed to help employees “remember,” when the actual need is to help them “perform.” For many workplace tasks, performance support tools, job aids, checklists, electronic performance support systems, are vastly more effective than memorisation-based training.

Companies that replaced week-long training courses with one day of training plus job aids saw proficiency increase 50% whilst slashing training time (Rossett and Schafer, 2007). Job aids can be deployed 75% faster than traditional training at reduced expense. For infrequent tasks, those performed monthly, quarterly, or less, performance support consistently outperforms training. Yet the forgetting curve myth drives organisations toward memorisation-based solutions even when on-demand reference materials would be superior.

When forgetting actually matters

Not all forgetting is created equal, and effective training requires discernment about when memorisation is critical. Memorisation matters when speed is critical and referencing performance support would take too long. A physician facing a medical emergency needs instant recall of procedures. A sales professional in client conversation needs key product features in working memory. Emergency responders require immediate access to safety protocols without consulting references.

Memorisation is also necessary when referencing resources would undermine credibility or be physically impossible. You cannot consult notes during high-stakes presentations or negotiations. Equipment operators in hazardous environments may not have hands free to check procedures. Foundation knowledge, core concepts needed to understand new learning, must reside in long-term memory to enable schema formation.

But vast swaths of corporate training content don’t meet these criteria. For infrequent tasks, complex multi-step procedures, reference information, and regularly changing protocols, performance support is more effective and economical than memorisation. Equipment repair procedures for occasional repairs, software procedures for monthly tasks, compliance checklists, detailed specifications; these benefit far more from well-designed job aids than from training aimed at memorisation.

The critical distinction is between “can’t remember” and “can’t perform.” These represent different problems requiring different solutions. An employee who can’t remember rarely used procedure steps isn’t experiencing a training failure requiring remediation; they’re experiencing a normal human limitation that performance support solves elegantly. Conversely, an employee who can’t perform a skill they should have mastered needs deliberate practice with feedback, not another review session.

Where we go from here

The “70% forgotten in 24 hours” myth persists because it feels intuitively true, we do forget things, and it provides a simple explanation for training failure. It offers a memorable statistic for presentations, a compelling rationale for reinforcement budgets, and apparent scientific legitimacy from Ebbinghaus’s name recognition. These factors create what educational researchers call a zombie theory: thoroughly debunked but refusing to die.

Forgetting isn’t a universal, time-based decay process but a complex interaction of interference, retrieval failure, and context-dependent accessibility. Meaningful learning integrated into existing schemas follows different retention patterns than memorising isolated, meaningless information. When we understand that forgetting varies from 0% to 94% based on meaningfulness, prior knowledge, learning methods, and context alignment, we can design interventions targeting actual mechanisms rather than fighting imaginary universal decay.

So, the question isn’t “How do we prevent 70% forgetting in 24 hours?” because that’s not actually happening. The questions are:

• What knowledge truly requires memorisation versus reference?

• How can we space retrieval practice to optimise long-term retention?

• When should we invest in performance support rather than training?

• How do we measure workplace performance rather than just knowledge retention?

---

References

Ausubel, D.P. (1968) Educational Psychology: A Cognitive View. New York: Holt, Rinehart and Winston.

Baddeley, A., Eysenck, M.W. and Anderson, M.C. (2022) Memory. 3rd edn. London: Routledge.

Bahrick, H.P. and Hall, L.K. (2005) ‘The importance of retrieval failures to long-term retention: A metacognitive explanation of the spacing effect’, Journal of Memory and Language, 52(4), pp. 566-577.

Bjork, R.A. and Bjork, E.L. (1992) ‘A new theory of disuse and an old theory of stimulus fluctuation’, in Healy, A., Kosslyn, S. and Shiffrin, R. (eds.) From Learning Processes to Cognitive Processes: Essays in Honor of William K. Estes. Hillsdale, NJ: Erlbaum, pp. 35-67.

Bjork, E.L. and Bjork, R.A. (2011) ‘Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning’, in Gernsbacher, M.A. and Pomerantz, J. (eds.) Psychology and the Real World: Essays Illustrating Fundamental Contributions to Society. 2nd edn. New York: Worth Publishers, pp. 59-68.

Brod, G., Hasselhorn, M. and Bunge, S.A. (2021) ‘When generating a prediction boosts learning: The element of surprise’, Learning and Instruction, 55, pp. 22-31.

Clark, R.C. (2014) Evidence-Based Training Methods: A Guide for Training Professionals. Alexandria, VA: American Society for Training and Development.

Kornell, N. and Bjork, R.A. (2008) ‘Learning concepts and categories: Is spacing the “enemy of induction”?’, Psychological Science, 19(6), pp. 585-592.

Murre, J.M.J. and Dros, J. (2015) ‘Replication and analysis of Ebbinghaus’ forgetting curve’, PLOS ONE, 10(7), e0120644.

Roediger, H.L. (1985) ‘Remembering Ebbinghaus’, Contemporary Psychology, 30(7), pp. 519-523. Available at: http://psychnet.wustl.edu/memory/wp-content/uploads/2018/04/Roediger-1985_CP.pdf (Accessed: 8 November 2025).

Roediger, H.L. and Karpicke, J.D. (2006) ‘The power of testing memory: Basic research and implications for educational practice’, Perspectives on Psychological Science, 1(3), pp. 181-210.

Rohrer, D. and Taylor, K. (2007) ‘The shuffling of mathematics problems improves learning’, Instructional Science, 35(6), pp. 481-498.

Rossett, A. and Schafer, L. (2007) Job Aids and Performance Support: Moving From Knowledge in the Classroom to Knowledge Everywhere. San Francisco: Pfeiffer.

Thalheimer, W. (2010) ‘How much do people forget?’, Work-Learning Research. Available at: https://www.worklearning.com/2010/12/14/how-much-do-people-forget/ (Accessed: 8 November 2025).

van Kesteren, M.T.R., Krabbendam, L. and Meeter, M. (2022) ‘Schemas provide a scaffold for neocortical integration of new memories over time’, Nature Communications, 13, Article 5795.

Willingham, D.T. (2003) ‘Students remember...what they think about’, American Educator, 27(2), pp. 37-41. Available at: https://www.aft.org/ae/summer2021/willingham (Accessed: 8 November 2025).

Wixted, J.T. (2004) ‘The psychology and neuroscience of forgetting’, Annual Review of Psychology, 55, pp. 235-269.