The 70-20-10 learning model

The ratio comes out of research at the Center for Creative Leadership in the 1980s, where Morgan McCall, Michael Lombardo, Robert Eichinger and colleagues asked successful executives where their most important lessons had come from. Most named a job: a turnaround, a first command, a visible failure, a difficult boss. Very few named a course. Lombardo and Eichinger later attached rough numbers to that pattern, the shorthand 70-20-10 stuck, and corporate learning has been arguing about it ever since.

What each number covers

Read as a description of where learning has historically come from, the model is hard to argue with. Read as a spending rule, it does real damage. Both readings are common, which is why the same three numbers get used to justify opposite decisions.

The 70 is not "doing your job"

This is the misreading that wastes the model. Ordinary work repeated competently for three years is not 70% of anything. It is practice at what someone can already do. A developmental experience is one where the outcome is genuinely uncertain and the person is visibly accountable for it.

Which means the 70 has to be designed. Left alone, most jobs settle into repetition within a year, and the person who has had ten years of it has had one year ten times.

What makes an experience developmental

• The person owns a real outcome, in front of people whose opinion of them matters.
• The task needs a capability they do not have yet, not more of one they already have.
• Failure is genuinely possible, and there is a net underneath that stops failure being career-ending.
• Somebody helps them make sense of it afterwards. Experience without reflection produces confidence, not competence.
• It ends. Open-ended stretch becomes ordinary work in about a quarter, and stops teaching anything.

The 70 is not free. It is paid for in senior people's attention and in the willingness to let someone do a job worse than you would for a while.

The 20 is the part organisations leave to chance

Most organisations mandate the 10, hope for the 70, and do nothing at all about the 20. Yet the 20 is what converts experience into learning. Two people can run the same failed project and come out with completely different lessons, and the difference is almost always whether anyone sat down with them afterwards and asked good questions.

Mentoring schemes fail here because they are built as relationships rather than around work. A mentor with no visibility of what their mentee actually did can only offer general advice, the least useful kind.

• Attach a named person to every stretch assignment whose job is to debrief it, not to rescue it.
• Make feedback specific enough to act on. "Be more strategic" is unusable; "you gave three options and no recommendation" is a behaviour someone can change. There is more on this in how to give constructive feedback at work.
• Put peer learning in the calendar: one hour a month where two people each bring a live problem, not a polished presentation.
• Let people watch experienced colleagues work rather than hear them describe it later. Most expertise is invisible in the retelling.

The 10 does more work than its number suggests

The smallest share is the one people cut first, usually with the line that 90% of learning happens on the job anyway. That gets the causality backwards. Formal learning is what gives people the language and the models with which to interpret the 70. Someone who has never been given a frame for negotiation notices nothing useful in a negotiation, however many they sit through.

Formal learning also does something experience cannot: it lets people rehearse without consequences. A manager's first redundancy conversation should not be their first attempt at one. Neither should a team leader's first serious conflict be the moment they discover they have no method for it.

• Use sessions to hand people a portable frame: a structure for a difficult conversation, a model for feedback, a checklist for a decision they will make under pressure.
• Time training to sit just before the experience that needs it, rather than annually because the calendar says so.
• Judge a course by what changed in the following month, not by the scores collected in the room.
• Spend the 10 on rehearsal rather than transmission. Content can be read; practice cannot.

How to build a development plan with it

1. Start from a capability, not a calendar. Name one thing this person should be able to do in six months that they cannot do today. Be concrete enough that you would recognise it if you saw it.
2. Find the 70 first. Which real piece of work would force that capability? If no such work exists, the plan is decoration and everyone will know it by week three.
3. Attach the 20. Who observes, who debriefs, how often, and what specifically will they be watching for? Agree it with both people, out loud.
4. Add the 10 where it earns its place. Where the person lacks vocabulary, or where live practice would be too costly to get wrong.
5. Set a review date and a visible marker. Something that either happened or did not: they ran the client review, they chaired the meeting, they made the call.

Note what the sequence does. It puts the assignment before the course, which is the whole point of the model, and it makes somebody accountable for the 20, which is the part that otherwise evaporates.

Where the model breaks down

Five honest caveats, because most pages on this model have none.

1. The numbers were never measured. They come from successful executives recalling their own careers. Recall favours the vivid, and nobody remembers the workshop that quietly changed how they run meetings. Treat 70-20-10 as a rough description of a pattern, not as data.
2. The sample was senior and specific. It describes people who had already succeeded, in one kind of professional life. An apprentice, a newly qualified clinician, or anyone working in a safety-critical setting needs a different balance, with formal instruction firmly in front.
3. Experience teaches whatever happened to work. Unfacilitated experience embeds bad habits as efficiently as good ones. A manager who bullied his way to a result has learned, accurately, that bullying works. That is a learning outcome nobody wanted.
4. It gets used as a budget argument. "Ninety per cent of learning is free" is the most expensive sentence in learning and development. The 70 and the 20 consume senior diaries, tolerance for error and deliberate design. They are cheaper in cash and dearer in attention.
5. The categories do not separate cleanly. A good workshop is 10 by label and mostly 20 in the room. A well-run project review is 70 and 20 at once. Counting hours into three buckets produces reporting, not learning.

A quick test of whether you are actually using it

Three questions, answered honestly, will tell you more than any dashboard.

• Can you name the specific piece of work each of your people is currently learning from? If not, you do not have 70-20-10, you have 0-0-10.
• When did you last debrief one of those pieces of work out loud, for more than ten minutes, without turning it into a status update?
• What did anybody do differently in the month after your last training day? If nothing, the problem was rarely the training. It was the absence of anything around it.

This is also the argument for experiential learning, and its limit. Put a group into a real task with real pressure and you compress the 70 into an afternoon: the stretch, the visible accountability, the chance to fail without a customer paying for it. What makes the afternoon count is exactly what makes a stretch assignment count, the debrief afterwards, where a facilitator points at what actually happened and asks why it happened. Skip that and you have run an activity. Keep it and you have run the 70, the 20 and the 10 in the same room, in a single day, with everyone watching.

Tour De Force runs this as live, experiential training for teams worldwide — online and in person. Talk to us, or play Gamified learning appsThe Weekly Challenge to see the method in ten minutes.