Revisiting 70:20:10 — 15 Comments

  1. Thank you for the excellent overview of the model/framework, Tom.

    I agree that 70:20:10 is neither a fad nor a brand. Neither is it a ‘learning theory’ (as I have heard a couple of academics refer to it). In practical terms it is a tool or ‘lens’ that describes how adult learning in work actually happens. As such, it provides a mechanism through which we can re-focus our effort and resources for greater efficiency and effectiveness to help build high performance.

    The strength of the 70:20:10 model lies in its ability to help organisations understand that learning comes via a myriad of daily challenges and interactions in addition to any away-from-work structured training that might be undertaken. Learning in the workflow (the ’70’ and ’20’) tends to be more frequent and have greater impact than the ’10’. All learning is contextual, so the closer to the context of where learning is to be turned into action that the learning occurs, then the more impact it’s likely to have.

    Incidentally, together with Jos Arets and Vivian Heijnen, I’ve just published a new book titled ’70:20:10 Towards 100% Performance’. the book (all 315 pages in large ‘coffee table’ format and weighing 4lb/1.8kg) describes the 70:20:10 model in detail and recommends new roles and tasks that are needed to support its effective use. If of interest, its available from the 70:20:10 Institute

  2. I can feel the conviction you hold for 702010 and I appreciate that you shared your opinion in this conversation that took place on Twitter and on Mark Britz blog ( . I know you are smart and open so I can say things as I think them.

    Let’s stay pragmatic. The debate was not about the validity of 702010 or challenging it, but about why it faces a chasm in its diffusion and that some people don’t get it. For this, we should first apply empathy. I’m with Nick Shackleton-Jones who said on last LearningNowTV ( that Design Thinking has the potential to be the most disruptive opportunity of L&D. I’ve seen in this debate that Jane Hart started or in xAPI the same at play: Taking one’s source always on the convinced, L&D side instead of being open to the world, listening and learning from various sources.

    I know also you have nothing to sell and you are not interested in Marketing, yet it’s a technique that teaches us a lot on listening, understanding groups, educating and convincing them. I have years in practice here.

    An example: When I said the branding is not adequate I used the commonly accepted definition of Branding:

    “A brand (or marque for car model) is a name, term, design, symbol or other feature that distinguishes one seller’s product from those of others”. Source Wikipedia.

    or from Google:

    “Branding is a marketing strategy that involves creating a differentiated name and image — often using a logo and/or tag line — in order to establish a presence in the consumer’s mind and attract and keep customers.”

    In your post, you reused your definition of Branding and build on it a reasoning I can’t take. Search the Wikipedia page: no mention of persona. Nobody will object that “Coca-Cola” is a brand, or Apple. yet neither have a single persona corresponding to it. Persona belongs to segmentation and audience, a different topic. Branding is just putting a flag on a piece of territory, here a concept. To be recognized it must be accepted and understood by everyone.

    Most of your sources come from the Internet Alliance. They are consistent, coherent and it’s good but it doesn’t help to understand why people object. Repeating the same arguments don’t help. It’s necessary for that to take the opposite point of view, welcome others views and a place in their shoes.

    I’ll stop here, will not reply to everything, before I’m filling your blog :)

    #WOL I was precisely here when we resumed our convo on Twitter. Fascinating how we hop from timezones to timezones and media to media.

    • Bruno – I wonder if you’re tilting at a few windmills here. I’ve worked with more than 200 organisations across the world who ‘get it’ with 70:20:10. You mention Coca-Cola. That’s one organisation which uses 70:20:10, as do many, many others – from well-known multi-nationals to smaller national and local orgs.

      I agree with you that there’s a continued need to educate people on what using 70:20:10 as a strategic framework is, and what it isn’t (the recent book is about just that). However, when I encounter challenges about the model, they are principally from [a] academics who think 70:20:10 is some type of ‘learning theory’ and go on to demand peer-reviewed references on ‘the numbers’ – looking for a p-value <0.05 (it isn't a 'theory' it's a reference model); and [b] from L&D/HR professionals who don't accept the relevance or importance of experiential and social learning (the '70' and '20') because they think it degrades the importance of their (principally '10') work (i.e. they're in defense/denial mode). I sometimes wonder whether it's worth spending extra time and effort with either of these groups.

    • As always, your thoughts and comments are welcome, Bruno. I’m beginning to understand how your perspective differs from mine and why. Your comments about branding miss my point that I see no need for it – so the “branding” being inadequate doesn’t bother me at all. By your own statement, branding conveys differentiated name and image, which then allows one to take shortcuts of sorts – useful in advertising when everyone knows Xerox connotes a brand image for professional copiers, or when the television commercial targets the heart and closes with the sports hero drinking a bottle of Coca Cola.

      But I also recognize that the learning and development industry as a whole has a terrible track record using buzz phrases, effectively brands, without defining them or assuming that everyone knows what it means: elearning, blended learning, ADDIE, learner-centered instruction, and so on. Note I clearly defined what I meant by 70:20:10 before I discussed it, because I feel strongly we need to always clearly define what we mean. As an aside, I personally loathe the phrase, “blended learning,” and refuse to use it.

      I may use 70:20:10 when talking with managers or leaders. but I would clearly explain the concept and the basis for it. More likely, however, the conversation would start with some basic agreement that learning a job is better on the job where it can be put into action. As Charles wrote, “Learning in the workflow tends to be more frequent and have greater impact…” I wouldn’t likely bring up “70:20:10” without basic agreement on how learning happens in the workplace and the relatively minor role of formal training.

  3. An interesting conversation, which I join as a learning practitioner working in an organisation which adopted 702010 framework many years ago (incidentally, Coca-Cola Amatil). I joined the Twitter convo in response to a post by Mark Britz that many people don’t get that it’s not about the numbers. This surprised me as the framework is so ingrained in my organisation that no-one talks about the numbers or tries to apply as a formula. A bit like a fish in water I just don’t ‘see’ the numbers. I’ve actually found them a good launching point for a conversation with someone who isn’t familiar with the concept they encapsulate. Having thought about the alternative suggested in the Twitter conversation for a few days (‘The 3 Es – Education, Exposure, Experience), I actually think I’ll stick with 702010 as it’s easier to remember and is a useful conversation starter. I don’t feel like I need to put energy and time into ‘re-branding’ or changing the shorthand I use to refer to the framework. Much better to put this energy into improving organisational and individual performance.

    Another key point is that the framework is indeed about a whole of business approach to learning and performance, not something ‘owned’ by L&D. This is a shift still underway in my business, which is helped in part by the fact that my role and team report to an operational business unit (Supply Chain), not the HR dept. Once we started using performance consulting to reorient discussion of requests for courses/training the shift to broaden buy-in / use of the framework got underway. It continues, helped in part by the fact that many of our ‘Capablity Managers’ (which is what we call our local L&D roles) also have responsibility for ‘Operational Excellence’, our continuous improvement process.

    I’ll stop there as I think I have a blog post in the making myself. Bottom line – as a practitioner my best play is to take action, get on and do useful things to improve organisational performance adopting the 702010 concept as an overarching guide to the wide range of approaches that can be used to make a difference.

    • Well-said, Michelle. The field of L+D has a somewhat tarnished reputation because of (in my opinion) poor communication, and may “hide behind” our buzz phrases. Perhaps I was tainted early by a contract instructional designer who held up his ADDIE model and insisted we follow it, but as I spent time and began to understand what he was actually talking about, I realized ADDIE was a process of sorts, similar to the systems engineering process my company already used. We need to be better communicators, bridge the gap between different disciplines, and using 702010 as a conversation starter is a smart use. The key is the conversation, not the label.

  4. While I believe in the concept of 70-20-10, the term itself makes no sense. The Center for Creative Leadership basically calls it a predictive model, “A research-based, time-tested guideline for developing managers says that you need to have three types of experience, using a 70-20-10 ratio: challenging assignments (70 percent), developmental relationships (20 percent) and coursework and training (10 percent).”

    However, now it is being rebranded as a reference model. In addition, it is being changed from 70-20-10 to 70:20-10 to make it look more like a ratio. Thus, we now present a ratio, 70:20:10, but tell people don’t pay any attention to the ratio, as it is only a “reference model.”

    It reminds me of The Wizard of Oz, “don’t look behind the curtain!” Thus, as a reference model it fails the first test – encourage clear communication. Yes, we can tell everyone personally that it is a reference model, but the model should stand on it’s own, which it doesn’t; because we have to consistently remind people that we want 70:20:10, but we don’t really, really, want 70:20:10.

    It is also not a reference model because it does not provide both entities and relationships. The entity is 70:20:10, which is a ratio, however, we are told not to worry about the ratio. Huh? We are presented with an entity and told not to worry about it?

    The relationship is missing because the model fails to describe how the three numbers connect, interact with one another, and exhibit joint properties — it only tells us percentages.

    Perhaps more importantly, it fails to provide meaningful information about the learning environment because the numbers are wrong. The original research asked leaders how they thought they learned. However, new research discovered a 55:25:20 ratio ratio of actual learning time spent and that this number varies very little by leadership level.

    I found this statement from DDI rather thought-provoking, “Even more problematic, the 70:20:10 ratio—in fact, any ratio—emphasizes the separation of learning methods rather than their integration. Allowing learning methods to compete rather than integrating them so they can build on one another undermines their impact and their value.”

    • So I agree with both comments from wiredjava and (of course) Charles. How so? Wiredjava points out it’s difficult to separate formal, social, and experiential. Indeed, I’ve facilitated many face-to-face “training” sessions in highly specialized technical disciplines. The benefit to this format in our workplace is the gathering of intelligent people committed to a single purpose who want to learn something about tangential disciplines. It’s a formal setting, but the informal/social aspects are invaluable and the reason that format continues to be preferred among our technical staff. How does one “count” this in a study? And to Charles’ point, the numbers would come out very differently for my organization – as Lisa Minogue-White calls “complex” – than it would for a corporation with similar roles filled in multiple locations such as tellers at a bank or baristas at espresso shops.

      Thank you both for a lively exchange, I look forward to more comments and responses.

  5. @wiredjava (apologies, I can’t identify you from your pseudonym here)

    The 70:20:10 framework/ model is more akin a business reference model than systems or software engineering model. I find that most people understand that point.

    As a business reference model, it is being used increasingly to help ‘re-think’ the way we support learning and development – through a combination of experiences, exposure to others, and structured education/training.

    Various studies have been published similar to DDI’s – but with different ratios. Here’s my response to one that ‘found’ the ratio to be 50:26:24.
    Without access to their sampling approach, my initial question would be whether the DDI data is likely to be skewed towards the ‘formal’ simply due to that organisation’s focus (i.e. DDI primarily offers courses – the ’10’) and the fact that their sampling of ‘13,000 leaders’ is likely to be drawn from people who have used, or are likely to use, DDI’s services. I’m very happy to be disabused if someone can point me to DDI’s actual sampling methodology.

    Just a minute’s thinking would lead anyone involved with human performance and development to realise that every study will produce different numbers. There is no ‘right’ ratio, just a reference model. For now, the ’70:20:10’ label seems to work for most people. This isn’t some ‘golden mean’ we should aspire to, it’s just a label to remind people that most learning occurs in the ‘70’ and ‘20’ domains. I’m sorry you have a problem with the label, although appreciate that you say you believe in the 70-20-10 concept.

    Where I do agree with DDI’s sentiment, and the point you’ve highlighted, is around the fact that a ratio has the potential to emphasise the separation of learning methods rather than their integration (which is why I regularly write about their integration when using the 70:20:10 approach – through adding, embedding, extracting and sharing learning). We need to be wary of that – ‘10’ type learning opportunities, for instance, will always benefit from some ‘20’ (coaching) and ‘70’ (experience). As will ‘70’ with some ‘20’. Many organisations (DDI included) have been separating the ‘10’ from the ‘20’ and ‘70’ for years.

    Basically, the 70:20:10 approach is not about the numbers, it’s about change. Do we expect to find every application of the Pareto Principle to deliver exactly an 80:20 ratio? We’d be very stupid if we did.

    If people find the concept of 70:20:10 useful, use it.

  6. I have been in the corporate learning field since the mid 80s and have heard more than my share of models, buzzwords (like today’s “ecosystem”) and fads. Although the ratios may not be exactly 70:20:10, it is descriptive of corporate learning in that there is work-related learning, informal learning and formal training.

    From my perspective, the biggest changes to corporate learning in the last two or three decades is that there are more opportunities (like this online article/discussion, webinars, etc.) for informal learning and the pace and amount of work have increased dramatically.

    I don’t think the formal learning part has increased, as training has always been looked at as an expense for companies and time away from the job for the workers. Of course some companies buy into formal employee learning more than others. So maybe the ratios in the 80s were 85:5:10.

    I do not find much value in a descriptive model. What I find more interesting is what has changed and what is changing. As someone involved in technology-based learning, I do find value in ways to influence learning and learning design. So bringing video-based micro-learning and gamification elements to corporate learning are practical approaches to dealing with lack of time for employees and gaining and keeping their interest. Using social media to connect people and ideas is also powerful and still untapped in many corporate settings.

    So you can argue about ratios if you want, but the implications of the changes is what I find much intriguing.

  7. Hi Charles. I’m sorry for the post that simply referred to me as WiredJava. I’ve never used it for commenting before so I did not realize it would only list the title, but it was the option that this blog reported I could comment as. I’m actually Don Clark.

    As to whether the DDI data is likely skewed towards the ‘formal,’ which would not be surprising, however, if you look at the questions that CCL used in their 70-20-10 findings, they seem to be skewed towards the ‘Experience’ part. In addition, I would note that DDI’s ratio comes closer to numbers in the post on Where did the 80% come from? than CCL’ ratio.

    However, as you note, it’s not really about hitting the numbers. But if this is true, why use numbers? This seems even more important to me because 70-20-10 is not even a Business Reference Model. It is not even a model — but rather a ratio that managers, in hindsight, believed they learned.

    So how do you make it a model? There is one learning model that I know of that comes close to 70-20-10. In fact, it might even be based on it: The Full Spectrum Learning Model.

    It was recently developed by the U.S. Army. Now I’m a retired Sergeant First Class (E-7), so I might be rather partial to it, but I believe the concepts behind it make it a more useful model, than simply a picture of a ratio. Note how the model is constructed:


    It uses the terms Training, Experience, and Education. As Tom noted, the Social aspect is highly present in all three environments, experiential, learning from others, and training; thus having a separate category (or ratio) does not really make sense. It fact, I believe it totally skews the percentages.


    This term, Education, might not make sense for some organizations, but to the U.S. Army, it is important. In Human Resource Development, training is normally associated with learning to perform a present job or task, while education is normally associated with learning to perform a future job or task. Thus, in a rapidly changing world, education through formal and/or nonformal environments is a required component if an organization wants to remain competitive. While VUCA affects all organizations to a certain degree, it has a huge impact on the Army, thus the importance of having it as part of their learning environment.

    For example, during our last recession, companies dumped thousands of people in mass layoffs. Now they are whining that they cannot find people who have the education and training that they require. Good organizations should always be looking towards a path to the future by educating people to build and walk that path.

    While you might want to hang on to the Social environment, or perhaps even another term, Education should at least be considered as a viable alternative.


    Rather than build a learning model that focuses on one fixed point, the U.S. Army created the Full Spectrum Learning model on two continuums based on the degree of Responsibility of the learner and the degree of Ambiguity of the learning environment to give it depth. Thus, the three environments are placed within their respective locations in the model.


    The three environments do not stand on their own, which fixes one of the main criticisms of 70-20-10. For example, when the Responsibility of learning is high for the performer, and there is high Ambiguity in the learning environment, they mostly learn from Experience, but the learning is not isolated from Training and Education.

    Now I’m not certain the Full Spectrum Learning Model is the best model out there or could at least use some improvement, but I do believe that it models a better vision of the learning environment than a simple ratio.

  8. Don, the Full Spectrum Learning Model you pointed to is interesting and very similar in many respects to the way some organisations are using 70:20:10. Cisco and Citibank are two examples that use the principles of a 70:20:10 approach but call it the ‘3Es’ – Experience; Exposure; Education. This is a piece of Citi’s internal information/marketing collateral

    Although I relate to the responsibility/ambiguity matrix in the FS Learning Model very much (but do wonder whether that applies so much in knowledge work and less structured organisations as much as it does in a structured organisation like the Army) the ‘Education’ component of the US Military model is an interesting one. ‘Education’ in most contexts and as used by most people, refers to ‘formal education’. In the FS Learning Model as you describe it, ‘Education’ seems to cover pretty-well everything except training and experience – including directed learning, self-directed learning, undirected learning, formal, informal etc. While I’m in absolute agreement that all of those are necessary to prepare for future tasks, I do find the terminology slightly confusing.

    Incidentally, in the new book I’ve written with two colleagues, we’ve used a model to show that each of the ’70’, ’20’, and ’10’ parts are recursive and mutually supportive. Learning is a complex, dynamic and cyclical process. So each dimension (70, 20, and 10) is likely to contain elements of the other two – the ’70’ (learning through experience) is likely to have elements of learning with and from others, and even possible some formal guidance in there and so on. Sorry I cant’ point to an illustration showing this, but it’s not on the web and I haven’t found a way to include it here.

    A couple of final points.
    I find that once people get beyond only seeing the numbers in the model they tend be able to focus on the main usefulness of 70:20:10 – extending learning beyond the ’10’ – and focusing on performance (outputs). Secondly, I’ve ceased having discussions about the numbers, or even the name (as I have ceased discussing whether or not it’s a ‘learning theory – which it isn’t) as almost everyone will have their own opinion on why ’70:20:10’ is or isn’t a good name. For now, it is as it is. Angels and pins. Some time other models will no doubt emerge from it and they’ll have different labels. People will then have the opportunity to discuss and argue about them, too ;)

  9. Charles, thank you for the comments and the link to the 3E video. Also, you seem to have some doubt as to whether the responsibility/ambiguity matrix applies to knowledge organizations. It might help to think of it in this manner — the amount of responsibility you want to give a person for learning a task.

    For example, you have a new hire and ask her to create a sales forecast. She might make a mistake or two, but there is no real harm to the organization, so she learns it through experience and others. Next, you ask her to create another report, but she has to get the information by joining several tables from a major database that is used throughout the organization. However, if she joins too many tables together at once it could severely slow down the system and affect hundreds of people throughout the organization.

    In the first case, you give her full responsibility for learning as a mistake only has minor consequences. In the second case a mistake may cause major consequences, thus you want the learning to be more formal in order to eliminate the possibility of that type of mistake, which means you place more of the responsibility on someone who actually understands the system. Thus, responsibility is based a lot on the amount of harm a mistake may cause. This includes such impacts as safety, hits to the organization’s reputation, cost of the mistake, and how it affects others, both inside and outside the organization.

    Coupled with that is the type of experience the person has. For example, if the person were a system engineer, then you would expect her to understand this, thus no training required.

    As far as ambiguity, tasks that have low ambiguity are often shaped that way for a purpose. For example, processes and procedures are often created to ensure compliance or the company has discovered it’s the most effective and efficient way to accomplish a task. This implies a more formal learning approach as you do not want the learners to discover it by chance. However, if the task has low ambiguity simply because that’s the nature of it, then training is not normally required.

    Ambiguity as a whole can be reduced if the organization is structured as the Army is. However, the task a person is learning and performing has about the same level of ambiguity no matter if that particular task is performed in the Army or in a knowledge organization.

    In addition, I can understand your concern with the term ‘education,’ as it is similar to my concern with the term ’70:20:10.’ My educational background is in Human Resource Development, which uses three terms:

    – Training is learning that is provided in order to improve performance on the present job. (full evaluation of the results can be determined when the performer returns to the job)

    – Education is helping people to do a future or different job. (full evaluation of the results can only be performed when the performer starts the new job)

    – Development is helping people to acquire new horizons, technologies, or viewpoints. (full evaluation of results is often difficult to determine because of extensive ambiguity)

    Thus, from my frame of reference the term ‘Education and Development’ would be the best fit — Education covers future jobs, while Development covers the VUCA aspect.

  10. Pingback:Three Reasons We Need 70:20:10 Now

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: