The 70-20-10 Learning Model, Explained (and Fact-Checked)


Create engaging training videos in 160+ languages.
The 70-20-10 learning model proposes that 70% of leadership development happens informally through on-the-job experience, 20% through relationships, and 10% through formal training and coursework. It is widely cited as guidance in learning and development, but it isn't backed by research.
The model also assumes that experiential learning automatically leads to capability development, and that formal learning leads to transfer without reinforcement or practice. Neither assumption holds up under scrutiny.
What L&D practitioners should take away from this model is that successful leaders often develop key skills on the job, and that any formal development should complement these experiences.
The 70-20-10 origin story
In 1988, the Center for Creative Leadership published Lessons of Experience: How Successful Executives Develop on the Job by Morgan McCall Jr., Michael Lombardo, and Ann Morrison. Notably, the book focuses on experiential learning.Β
Over five years, the authors interviewed 191 high-performing executives across six major corporations about key developmental events in their careers.
They identified 616 experiences and 1,547 lessons from the self-reported data, which they coded into three broad categories: assignments, bosses, and hardships. (I promise, this is important.)
In each chapter, they provide anecdotes from the interviews about what their executives learned, and how. For instance, in the chapter, βTrial by Fire: Learning from Job Assignments,β they explain how the "driving force for learning [in these scenarios] was job challenge.βΒ
Where the ratio came from
Nearly a decade later, in 1996, one of the authors, Michael Lombardo, released the Career Architect Development Planner, with Robert Eichinger. Their Planner is a system of 86 βresearch-based and experience-tested development plansβ.Β
In the preface, Lombardo and Eichinger explain the philosophy behind their system,
Development generally begins with a realization of current or future need and the motivation to do something about it. This might come from feedback, a mistake, watching other people's reactions, failing or not being up to a task β in other words, from experience. The odds are that development will be about 70% from on-the-job experiences β working on tasks and problems; about 20% from feedback and working around good and bad examples of the need; and 10% from courses and reading.
You might be wondering, how did they get to this magic ratio? You won't find the methodology in the Planner.Β
Decades later, Eichinger provided his own retrospective account, explaining that theyΒ took the executive interview data from Lessons of Experience, recoding the 616Β experiences into five categories:
- Learning from challenging assignments
- Learning from other people
- Learning from coursework
- Learning from adverse situations
- Learning from personal experiences outside of work
They realized that learning from adversity and personal experiences couldnβt be designed into a development program, so they threw out those two categories, which accounted for 25% of the overall experiences. The remaining three categories: challenging assignments, other people, and coursework, accounted for 69%, 22%, and 9%, respectively.Β
The authors rounded those numbers to 70%, 20%, and 10%, βfor memorability and adoption.βΒ
Youβve got to give it to Lombardo and Eichinger, that magic ratio certainly is memorable. Thereβs a reason it is also used for everything from personal finance to Googleβs approach to innovation.Β
What you should do with the 70-20-10 model
That memorability is what has made the 70-20-10 model stick. And whatβs made it so hard to push back against. (I mean how many people are still convinced theyβre βvisual learnersβ?) So instead of a re-education campaign, I recommend treating 70-20-10 as a helpful reminder, not a formula. Here's how:
1. Provide structured stretch assignments
Individuals who are thrown into opportunities without preparation or support often struggle. In some cases, they even pick up bad behaviors. If you're designing a stretch assignment, articulate what success looks like, and then make sure the individual is getting regular feedback against that target. A stretch assignment shouldn't be one opportunity to try (and likely fail), but rather a time-bound development opportunity to build capability.
2. Offer coaching and mentoring
In Lessons, the interviewees described learning from other people, including bad bosses and leaders. They learned from the people in their proximity. If you want to reinforce desired behaviors, consider designing structured coaching and mentoring programs. While each serves a different purpose, you can incorporate them into a broader development program.
3. Reinforce formal training
For a training program to stick, it needs to be reinforced in the flow of work. (I know, you know.) While formal training in the workplace has evolved over the past four decades, it is worth repeating that coursework alone won't change behavior. Give employees opportunities to practice their new skills, get feedback, and continue developing what they've learned.
4. Connect experiential, social, and formal learning
You can do each of these things independently and still fail to drive learning outcomes. Use each to complement the others. If you design an experiential learning opportunity, pair it with a mentor for guidance and formal training to build the underlying skill. Think of social learning as the connective tissue between experiential and formal learning β reinforcing what is taught, modeling desired behavior, and providing the actionable feedback necessary to improve.
How technology is changing the 70-20-10 modelΒ
The interviews behind Lessons were conducted over forty years ago, well before eLearning replaced much classroom training and executive coaches became ubiquitous.
Today, formal training and support are available on demand. Leaders can watch a quick training video on how to command a board room, and then Zoom their coach to practice what they've just seen modeled.
And then there's AI, which is making personalized learning feasible at scale, transforming not only how leaders experience eLearning, but also what knowledge they have accessible during stretch assignments. AI can also create the practice itself. A roleplay session simulating a critical conversation isn't on-the-job experience, and it isn't a course. It's practice designed to challenge a leader, like a stretch assignment, but with the feedback loop built in.
If the boundaries between these categories are already dissolving, the ratio was never really the point. What matters is whether people get the opportunity to learn and practice, with the support and feedback they need to build capabilities.

Amy Vidor, PhD, is the Learning and Development Evangelist at Synthesia, where she researches learning trends and helps organizations apply AI at scale. With 15 years of experience, she has advised companies, governments, and universities on skills.
Frequently asked questions
Is the 70-20-10 learning model backed by research?
No, it is not backed by peer-reviewed research. The 70-20-10 learning model comes from a 1996 corporate development guide, the Career Architect Development Planner. The model is grounded in interview data with nearly 200 executives from six Fortune 50 companies.
What are the alternatives to the 70-20-10 model?
Other frameworks that recommend blending formal, informal, and social learning are the Pervasive Learning model and the OSF ratio. Dan Pontefract argues for an even 33% split between the three, while the OSF ratio, based on more recent research, recommends closer to a 55-25-20 split. Both highlight that these ratios aren't universal.
How do you measure whether the 70-20-10 model is working?
Because the 70-20-10 model is more of a general framework than a prescriptive design approach, you can't really measure whether it's "working." What you can measure is whether a development program β one that combines social, formal, and experiential learning β is producing real behavioral change, and whether that change is being reinforced over time.










