calendar
Edited on October 10, 2026

70-20-10 Learning Model: Origin, Evidence and Examples

Where the 70-20-10 learning model really comes from, what the evidence shows, and a worked 24-week development plan for one role with two free templates.

Carolina Martin, Learning Experience Manager at Coursebox AI
Carolina Martin
Customer Success Manager
Key takeaways

The 70-20-10 learning model is a rule of thumb for planning staff development: about 70% from challenging experience on the job, 20% from other people, especially managers, and 10% from courses and reading, with the numbers as reminders, not measurements.

  • The 1988 book behind the model never states 70-20-10, and the numbers first appear in a 1996 planner by Michael Lombardo and Robert Eichinger.
  • Coursework made up 38 of the 616 key career events that 191 executives recalled in the original Center for Creative Leadership research.
  • In the same research, 90% of the people named in person-centred key events were organisational superiors, so a plan should give the manager a defined job.
  • The model has no controlled evidence behind its numbers, so treat it as a design checklist rather than a budget split.
  • Of 21 workplace course requests typed into Coursebox over 90 days, none gave the learner's manager any part.
Does your development plan go beyond the course?

Five questions, about a minute. Answer for one role you develop, thinking about what actually happens in a new person's first few months. This is a planning checklist, not a validated measure of learning or effectiveness.

1
2
3
4
5
Does the plan name the real tasks the person will do in the job, in order from low risk to harder?
Instructional Designers

Go from storyboard to an interactive course in minutes

Picture a new tenancy officer at a community housing provider, two weeks into the job. They passed the two-day induction and the policy quiz, and on Monday they took over a full caseload of about 120 tenancies. Several tenants are already behind on rent, and the officer has not rung any of them. They do not know what to say when a tenant asks for more time, or which repayment plans the team will accept. Under the old approach, the team leader only sees the problem in the monthly arrears figures.

I'm an instructional designer at Coursebox, where we make an AI tool for employee training that turns a brief into a course, so I'll mention our product where I tested it. One planning lens for that officer's problem is the 70-20-10 model, which says most development happens in the job and through other people rather than in a course. It is also often misquoted. Here I trace where the numbers really came from, what the evidence does and does not show, and how to use the model for one role, with a 24-week plan and two free templates.

What is the 70-20-10 learning model?

The 70-20-10 learning model is a rule of thumb for planning staff development: about 70% from challenging experience on the job, 20% from other people, especially managers, and 10% from courses and reading, with the numbers as reminders, not measurements.

The 70-20-10 learning model connects experience, other people and formal input around one capability. Work provides material for feedback, feedback identifies learning needs, and formal input prepares the next task. Equal-sized panels avoid implying measured proportions; the numbers are planning reminders rather than learning or budget shares.

View the full-size diagram

The numbers were first printed in 1996 by Michael Lombardo and Robert Eichinger in The Career Architect Development Planner, a development guide published by Lominger. Their sentence reads:

"The odds are that development will be about 70% from on the job experiences, working on tasks and problems; about 20% from feedback or working around good and bad examples of the need, and 10% from courses and reading."

Look at the hedges: "the odds are" and "about". The same pages warn that "experience by itself is a lousy teacher" and that "Neither courses nor reading will work alone." Their point was that the three sources work together: "No one type does the job alone, but in concert they are powerful."

The model came from senior executives' careers, but L&D teams now apply it to almost any role, as a reminder to plan beyond the course. The 70:20:10 Institute, co-founded by Charles Jennings, Jos Arets and Vivian Heijnen, calls it "a Reference Model" and says on its model page that the specific ratio "will vary, depending on the work environment and the organisational results required." That is how I use it here: as a design checklist, not a measure of where learning happens.

Here are the three parts at a glance.

PartThe 1996 wordingThe question it asks of your planFor a new tenancy officer
70: experience"on the job experiences, working on tasks and problems"Which real tasks build the capability, and in what order?Shadowing arrears calls, then leading them with a buddy present
20: other people"feedback or working around good and bad examples of the need"Who gives feedback on that work, and when?A weekly 20-minute debrief with the team leader
10: formal learning"courses and reading"What formal input is needed, and when will it help most?A 45-minute course in two parts, timed to the first calls
note:Not the budget rule or the performance curve

Several unrelated ideas share these numbers: the 70-20-10 personal budgeting rule, the 70-20-10 split some firms use for marketing budgets and innovation portfolios, and the 20-70-10 split used to rank staff performance. None of them comes from the research behind this learning model.

A peer-reviewed study in the source search examined organisations using the model shows what happens when they adopt the label and leave the three parts to chance.

Five colleagues sit around a meeting table with a laptop, papers and notebooks
cottonbro studio / Pexels

Case study: Middle managers in the Australian public sector

This was a study of organisations already using the model, not a programme designed by the researchers. The Australian Public Service Commission began promoting 70:20:10 in 2011 and encouraged Commonwealth agencies to adopt it. The question was whether middle managers developed under the framework actually carried their learning into their work.

  • What the researchers did: They interviewed 5 senior managers and 18 middle managers in one Commonwealth organisation, then ran 13 group interviews with 122 middle managers from the Queensland, Victorian and Northern Territory public services, 145 people in all. The managers' experience came mainly through acting (higher duties) roles, alongside formal programmes and informal mentoring.
  • Evidence and result: Johnson, Blackman and Buick reported in Human Resource Development Quarterly in 2018 that the framework "is failing to deliver desired learning transfer results". The first of four misconceptions they found was "an overconfident assumption that unstructured experiential learning will automatically result in capability development". Managers described acting roles with "no one to backfill their position and no one to support them in their role", and learning from others depended on individual managers' habits rather than any design.
  • Limitation: It is a qualitative study of what managers said, in one sector, with the organisations unnamed. No performance, promotion or retention figures were measured, and the authors note their reliance on group interviews from one context.
  • Lesson: Adopting the label was not enough on its own. An acting role without backfill, feedback or a link to learning can add workload without providing the intended development, so plan the three parts together and name who will debrief the experience.

Where the 70-20-10 model comes from

The numbers come from a Center for Creative Leadership (CCL) study of how successful executives said they had developed. Lombardo, one of the study's authors, and Eichinger later regrouped and rounded them. The study itself never states the split.

WhenWhat happenedSource
Early 1980sCCL researchers ask executives to name at least three key events in their careersMcCall, Lombardo and Morrison, 1988
1987CCL publishes the event counts: 616 events from 191 executivesLindsey, Homes and McCall, Technical Report 32
1988The Lessons of Experience reports the study, with no 70-20-10 splitMcCall, Lombardo and Morrison
1996Lombardo and Eichinger print "about 70% ... about 20% ... and 10%"The Career Architect Development Planner
2001The Leadership Machine (first published 2001) explains the split as the events left after removing hardshipsLombardo and Eichinger
2007Reuters, where Charles Jennings is global head of learning, is reported to use "the 70:20:10 approach"TrainingZone profile, April 2007
2009 to 2010Morgan McCall, the study's lead author, calls the rule "folklore"McCall, Recasting leadership development

What the executives were asked

The researchers asked 191 successful executives from six major corporations one question: "Please identify at least three key events in your career, things that made a difference in the way you manage now." Interviews with 79 executives in three firms and open-ended surveys of 112 more in three other firms produced 616 events and 1,547 lessons. The companies hand-picked people "judged to have the best shot at the top jobs". Five of the firms were in the Fortune 50 industrials, and the sixth was a large Canadian subsidiary. The sample was "almost all white males, averaging in their early to mid forties" (McCall, Lombardo and Morrison 1988, p. 196).

The researchers flagged the limits themselves. Their technical report on the event data says "All of these data were self-report" and encourages "healthy skepticism of the results we report" (Lindsey, Homes and McCall 1987, p. x).

What the 1988 book found

The events fell into assignments, bosses and other people, hardships, and other events, including coursework. Two findings shape how I plan with the model.

The first is about people. "Almost 20 percent of the key events" featured a specific person rather than an assignment (p. 67), and of those people, "fully 90 percent" were "organizational superiors" (p. 68). The 20 was mostly managers, good and bad.

The second is about courses. Coursework was a small share of what executives remembered.

38 of 616
key career events recalled by 191 executives in the original CCL research were coursework (6.2%); counts of remembered events, not a measure of learning
Source: Lindsey, Homes and McCall, CCL Technical Report 32, 1987, p. 177

When coursework had an impact, the book says, "it dealt with a relevant issue, and it occurred at a good time for the manager" (p. 180). The authors also warned against reading their work as a formula: "Development is not magic. It is not a science. It does not lend itself to neat formulas, precise measures of this and that" (p. 183). A full-text search of the book finds no 70-20-10 split, and Jefferson and Pollock made the same point for the Association for Talent Development in 2014: "the phrase never actually appears in the book".

How the numbers were made

Lombardo and Eichinger later explained the arithmetic in The Leadership Machine: "When we take out hardships (because hardships are not something you put into a development plan), the remaining events cited by the executives were as follows: 70 percent were related to on-the-job experience, 20 percent were related to learning from others, and 10 percent were related to learning from courses" (p. 286 of the Korn Ferry printing).

The 1987 report's own counts show how much rounding that took. Our arithmetic from its per-event tables: removing the 107 hardships leaves 509 events, of which about 68% were job-based (assignments, early work and first supervision), about 22% involved other people, about 7.5% were coursework and about 3% were purely personal. So 70-20-10 is a rounded share of remembered career events, with one category left out. It was never a measured share of learning.

McCall, the lead author, said as much in a 2010 paper in Industrial and Organizational Psychology. He wrote of "the resulting folklore that there is a '70-20-10 rule' (I have not found an original published source, though the percentages clearly come from data reported in McCall et al., 1988, and Lindsey, Homes, & McCall, 1987)". He credited the "rule of thumb" with "increasing the emphasis on on-the-job experience", but said it "also misleads by suggesting that coaching, mentoring, and programs are effective when used as stand alone interventions."

Which sources to read

  • Lindsey, Homes and McCall, Key Events in Executives' Lives (CCL Technical Report 32, 1987). Read this one. It is free from CCL and gives the method, the caveats and every event count.
  • McCall, Lombardo and Morrison, The Lessons of Experience (Lexington Books, 1988). Read it for the executives' accounts, and chapter 6 for when coursework helped.
  • McCall, "Recasting leadership development" (2010). Read it for the lead researcher's own view of the rule.

The worked example: new tenancy officers

Here is the hypothetical example I'll plan with. A not-for-profit community housing provider manages about 1,800 homes in two regional towns, with a tenancy team of 9 officers and a team leader in each town. The learners are new tenancy officers in their first 24 weeks, managing rent, inspections, repairs and support referrals for a caseload of tenancies. Most come from customer service, real estate or community work. They are new to this provider's arrears policy and have not held hard conversations about money here.

The capability the plan builds is handling rent arrears: contacting the tenant early, holding the arrears conversation (listen, check what has changed, explain the position plainly, agree an affordable repayment plan, offer support), recording it and following up. The routine property inspection is a second, lower-stakes task for the early weeks. The provider's own rules, which are its policy rather than legal guidance, are:

  • contact the tenant within 5 business days of arrears reaching 2 weeks' rent;
  • record every repayment plan in the housing system and confirm it in writing;
  • never issue a breach notice or apply to the tribunal without the team leader's approval;
  • offer a support referral in every arrears conversation.

The people around the officer are the team leader, who holds the weekly debriefs and approves every notice; a senior tenancy officer as buddy for the first 3 months; the income officer, who runs the weekly arrears report; colleagues at the fortnightly case review; and the provider's tenant support worker.

The problem is the old approach: a two-day classroom induction on tenancy law and policy, then a full caseload of about 120 tenancies from week 2. New officers passed the policy quiz, then avoided early arrears calls, agreed plans tenants could not keep, and asked for notices too early or too late. So that decisions can be judged, officers work with the policy, the weekly arrears report, the repayment plan form and three short cases:

  • Case A: a tenant 3 weeks behind after their work hours were cut offers a repayment plan that would leave nothing for food.
  • Case B: a tenant 2 weeks behind has not answered two calls.
  • Case C: a tenant 1 week behind after a one-off car repair is already back at work. This is below the contact trigger, so a friendly reminder is enough.

The intended outcome by week 24 is an officer who contacts tenants on time, agrees plans that are affordable and recorded, offers support every time and brings notice decisions to the team leader with the facts. It is not certified competence or authority to issue notices. The example suits 70-20-10 because most of this capability can only be built in real arrears conversations, the people who can give feedback are clear, and the course has a narrow job.

The three parts, worked through one role

70: Challenging experience on the job

The 70 is real work that stretches the person: in the 1996 wording, "working on tasks and problems". Stretch alone has limits. In a 2009 study of 60 leaders and 225 work experiences, DeRue and Wellman found that skill development showed "a pattern of diminishing returns" as challenge rose, but that "access to feedback can offset" them. That is why the 70 cannot be planned without the 20.

  • In the plan: In weeks 1 and 2 the officer shadows the buddy on 4 routine inspections and 3 arrears calls and reads 5 closed arrears files. In weeks 3 to 8 they take a caseload of 40 and lead 3 arrears conversations with the buddy present. Independent calls follow only when the buddy and team leader have checked timely contact, affordability, support, records and authority limits in observed work. In weeks 9 to 24 the caseload grows to about 120, with one stretch assignment: running the arrears contact list for a week with the income officer.
  • How to check it: The team leader checks every repayment plan before it is confirmed in weeks 3 to 8, looking for contact within 5 business days, an affordable plan, a support offer and a complete record.
  • In the course: A plan at a glance and a log where the officer records each real task and one thing they learned from it.

20: Other people, starting with the manager

The 20 is feedback and examples from other people: "feedback or working around good and bad examples of the need", in the 1996 wording. In the original data those people were mostly bosses. McCall wrote that they "almost always were either excellent or terrible bosses and senior executives who, more often than not, were neither good coaches nor mentors." So the 20 starts with the learner's manager, not a social platform. Bad examples teach too. Peers and mentors are sensible additions, but the 1988 data say little about them.

Mentoring also costs senior staff time. CVS Group, a UK veterinary practice group, built its new-graduate programme on the 70:20:10 model, and a 2025 study co-authored by its staff found that graduates reported positive experiences of mentoring, while mentors said they needed more time to support graduates. It found "no significant differences in overall attrition rates" between cohorts before and after the redesign. Only the abstract was available to us, and it does not give the number of graduates analysed.

  • In the plan: The team leader holds a 20-minute debrief every week, moving to fortnightly after week 12, and approves every notice. The buddy supports the officer for the first 3 months, the officer brings one case to the fortnightly case review, and the income officer and tenant support worker are named contacts.
  • How to check it: Debriefs happen and are recorded, and each one ends with the next real task agreed.
  • In the course: A one-page debrief checklist for the team leader, with questions for each stage. It is a tool for the manager, not a quiz for the learner.

10: Courses and reading

The 10 is formal input: courses, reading and structured explanation. The 1988 book's verdict on coursework was that courses "do not substitute for on-the-job learning and should not be asked to serve that purpose" (p. 182), and that the coursework executives remembered was relevant and well timed. Small does not mean optional. It means the course should prepare for specific real tasks.

  • In the plan: A 45-minute course in two parts. Part 1 in week 1 covers the arrears policy in plain words, what an affordable plan looks like, the conversation steps, the support referral and what to record. Part 2 in week 3, before the first led conversation, has two example conversations to critique and a repayment plan to check.
  • How to check it: Open-text answers on Cases A, B and C, compared with the policy: support and a smaller amount for A, repeated contact attempts within 5 business days for B, and a reminder only for C.
  • In the course: An arrears conversation card and a repayment plan checklist to keep open during calls.
tip:Plan the course last

Write down the real tasks and the debriefs first, then ask what the course must do to prepare for them. If the course could be taken in any week without changing anything, it is not timed to a need.

How the three parts fit together

The parts are not a sequence or a target mix. They are three sources the plan needs, linked so each supports the others. Johnson, Blackman and Buick put it plainly: "the three elements need to work together in a coherent, designed way" (2018, p. 396).

In the tenancy plan, the links are specific. Part 2 of the course lands in week 3, just before the officer leads a conversation. The buddy sits in on the first 3 led conversations. The team leader debriefs each of the first 5 solo ones and checks the repayment plan before it is confirmed. Remove any link and the plan slips back to a course followed by a caseload.

warning:The numbers are not a budget or a timesheet

Read as time, 10% of an illustrative 1,800-hour working year is 180 hours of formal learning, far more than most plans intend. Read as money, it says nothing about what a course or a manager's time costs. The 70:20:10 Institute says the ratio varies with the work, and organisations interviewed for a 2012 DeakinPrime white paper reported using 40:30:30 and 50:30:20 in practice. To be fair, Lombardo and Eichinger did use the split for time themselves, suggesting that the time in a review be spent the same way.

How to use the 70-20-10 model to plan development

Lombardo and Eichinger's planner offers development tips, not a design process, so this workflow is my own, carried through the tenancy officers' first 24 weeks.

  1. Pick one role and one capability. Not "new starters" but new tenancy officers handling rent arrears.
  2. List the real tasks, from low risk to stretch. Shadowing calls, then leading them with the buddy present, then alone, then the stretch week on the contact list.
  3. Name the people and what each one does. Start with the manager: what they check, how often and what they approve. Then the buddy, peers and specialists.
  4. Decide the formal input needed and when it lands. Time each part of the course to the first task that needs it, as with part 2 in week 3.
  5. Add job aids. Give the learner something to use during the work, such as the conversation card and the repayment plan checklist.
  6. Decide how progress is checked in the work. The team leader's check of every repayment plan does more than a quiz score.
  7. Review after the first group. Ask the first officers and team leaders what helped and what got in the way, then change the plan.

The 24-week plan for new tenancy officers

Here is the plan in one table. Each stage ends with a check before the officer takes on more.

StageJob experience (70)People and what they do (20)Course and job aids (10)Check before moving on
Weeks 1 to 2: watch and prepareShadow the buddy on 4 routine inspections and 3 arrears calls; read 5 closed arrears filesBuddy explains each call; team leader debriefs at the end of each weekCourse part 1; conversation cardOfficer can explain the four policy rules and the conversation steps using a real call they shadowed
Weeks 3 to 8: lead with supportCaseload of 40; 3 inspections alone with reports reviewed; lead 3 arrears conversations with the buddy present, then the rest aloneBuddy reviews; team leader debriefs after each of the first 5 solo calls and checks every plan; one case at the case reviewCourse part 2 before the first led call; repayment plan checklistPlans are affordable, recorded and confirmed on time, with support offered every time
Weeks 9 to 24: fuller caseload and stretchCaseload grows to about 120; arrears contacts led independently; one week running the arrears contact list with the income officerWeekly debriefs, fortnightly after week 12; present the stretch week at the case review; notices still need the team leaderJob aids kept open on callsTeam leader agrees the officer handles arrears contacts on their own, with notices still referred
70-20-10 development plan for one role

A planning sheet that uses 70-20-10 as a reminder, not a ratio: for each stage, the job experience, the people and what they do, the course and job aids, what the learner records and the check before moving on. Filled in for new tenancy officers handling rent arrears, with blank pages for your own role. Editable Word file, no email required.

Step 7 is easy to skip. A Nigerian health agency got its clearest result from it.

A man in a blazer speaks at the front of a bright training room while seated professionals listen, some taking notes
Pavel Danilyuk / Pexels

Case study: A leadership academy at Nigeria's primary health care agency

This programme was designed with 70-20-10 by name. In 2018, Nigeria's National Primary Health Care Development Agency (NPHCDA) recognised that gaps in its staff's skills limited the technical help it could give state primary health care boards.

  • What they did: The agency set up a Leadership Development Academy built on what it called the 70-20-10 learning philosophy: on-the-job practice and a capstone project (the 70); coaching, mentoring, job shadowing and external rotations (the 20); and classroom and e-learning courses (the 10). After the first cohort it changed the design, adding three months, letting mentors choose mentees, dropping rotation hosts that did not fit and paying data and power allowances for virtual sessions.
  • Evidence and result: In Aminu and colleagues' 2023 paper in Global Health: Science and Practice, two cohorts of 29 and 27 fellows graduated between 2019 and 2020. The authors report that the proportion acquiring intermediate to advanced competencies rose "from 41% in the first cohort to 57% in the second cohort", which the authors credit to "Incorporation of feedback from the first cohort".
  • Limitation: The authors work for the agency and its implementing partner, the programme team rated competency on a 5-point scale, there was no control group, and the second cohort started from a lower baseline. The gain is credited to the redesign, not to 70-20-10 itself.
  • Lesson: The model worked as a checklist for what the academy had to contain. The authors attribute the improvement to feedback-driven changes in time, mentor matching and rotation hosts. The uncontrolled cohort comparison cannot isolate their effect. Treat your first plan as a draft and revise it from what learners and mentors report.

What workplace course requests leave out

To see how people brief workplace courses, I looked at course requests typed into Coursebox in the 90 days to 7 October 2026: one per person, from English-language browsers, with no uploaded file. Of 266 prompts, 104 were course requests. Two independent AI readers each classified all 104 against written definitions, and every positive call was re-read by hand. They agreed that 21 were for developing people in a work role, such as onboarding, safety inductions, supervisor courses and professional development.

19 of 21
workplace course requests asked for nothing beyond on-screen content: no task in the learner's job, no job aid, no follow-up and no other people; none of the 21 gave the learner's manager a part
Source: Coursebox course requests, PostHog, 90 days to 7 October 2026

Only two asked for anything more. One asked leaders to work on their own organisation through the programme, and one asked for a checklist to use in a real audit. Three requests were for courses aimed at supervisors or managers, and even those gave the learners' own managers no role. None of the 104 mentioned 70-20-10 or informal learning. In 70-20-10 terms, the briefs described the 10 and left the 70 and the 20 out.

The sample is small and covers requests, not finished courses, and organisations may run manager check-ins and stretch work that never appear in a prompt. With borderline requests counted, the workplace total rises to 27, of which 25 asked for nothing beyond the screen. A course builder only sees what the brief describes, so the job tasks, the manager's part and the job aids have to be written into it.

Is there evidence for the 70-20-10 model?

The source search found no defensible measurement establishing that 70% of all learning happens on the job, and I found no controlled study showing that a 70-20-10 design beats another approach. The evidence supports something narrower: experience is a large source of development at work, and it works better with feedback and structure.

Clardy reviewed five research traditions behind the wider claim that most work-based learning is informal. His 2018 review in Human Resource Development Review concluded, in its abstract, that "the evidential basis for the 70% rule is weak" and that "the 70% rule of informal learning needs to be set aside." Jefferson and Pollock reached a similar verdict on 70-20-10 itself: it is "neither a scientific fact nor a recipe for how best to develop people."

The best-known survey figure near 70% measures something else.

70%
of employer-provided training hours were informal, May to October 1995, in US private establishments with 50 or more employees; self-learning was not counted as training
Source: Frazis, Gittleman, Horrigan and Joyce, Monthly Labor Review, 1998

The Bureau of Labor Statistics survey behind it drew on questionnaires from 1,074 employees and a 10-day training log. It measures hours of training, not where people learned, so it cannot confirm 70-20-10.

Other studies point the same way without confirming the numbers. CCL asked a similar question of 393 senior executives in China, India, Singapore and the United States, and its Grooming Top Leaders report found that coursework and training were mentioned by just 9 to 15 percent of interviewees. These are still remembered events. A 2018 meta-analysis by Cerasoli and colleagues of 49 samples and 55,514 people linked informal learning behaviours to knowledge and skill (rho = .41) and performance (rho = .42), which supports taking informal learning seriously but says nothing about the ratio.

Limitations of the 70-20-10 model

  • It counts memories, not learning. The data are self-reported key events, rounded and regrouped with hardships removed.
  • The sample was narrow. 191 hand-picked executives, almost all white men, at six very large North American firms in the early 1980s. Applying the numbers to tenancy officers, or anyone else, is a leap the data cannot make.
  • Unstructured experience can fail. In the Australian public sector study, managers described acting roles without support, feedback or links to training.
  • It does not fit every role or stage. New starters in safety-critical or licensed work often need more formal input before they touch live work, and some decisions stay with a senior person, as notices do in the tenancy example.
  • It is easy to misuse. Read as a budget rule, it can justify cutting training without funding the managers' time the 20 depends on.

70-20-10 vs other models: which to use

70-20-10 is a planning mix, not a theory of how people learn. The table is my synthesis, not a ranking.

ModelBest forWeak spotUse it when
70-20-10Planning the mix of job tasks, people and formal input for a roleNo evidence for the numbers; says nothing about how to design each partYou are planning development for a role over weeks or months
Experience, exposure, education (3E)The same mix under plain labels, usable without the percentagesSame limits, and the labels hide that the original 20 was mostly managersPeople in your organisation keep reading the numbers as a budget
Kolb's experiential learning cycleTurning one experience into learning through reflection and trying againSilent on who supports the learner and whenYou are designing the debrief after a real task
Situated learningSeeing what newcomers can reach of a team's real workA lens on learning, not a design methodThe problem is onboarding and access to the work
Kirkpatrick or Brinkerhoff's success case methodChecking whether development changed the workEvaluation models, not design modelsYou need to show whether the plan worked

Some organisations, including Cisco according to CCL's leadership development handbook, relabel the parts as experience, exposure and education. For the debrief itself, Kolb's experiential learning theory and Schön's reflective model help the team leader turn a call into a lesson. Situated learning theory explains why the shadowing weeks need real tasks rather than paperwork. To find out whether the plan changed how officers handle arrears, use the Kirkpatrick model or Brinkerhoff's success case method. For the tenancy officers, I would use 70-20-10 to plan the 24 weeks and Kolb's cycle to shape each debrief.

Using the 70-20-10 model with an AI course builder

On 9 October 2026 I briefed Coursebox to build the tenancy officers' course. Coursebox is our product, so weigh my account with that in mind. The brief kept the course as the small part. It described the officers, the four policy rules, Cases A, B and C and the people by role, and asked for seven sections in order: a short "Before you start", the 24-week on-the-job plan with a log, the arrears conversation, practice on the three cases in open text, a checklist for the team leader, job aids, and a wrap-up last. I asked for about 45 minutes of screen time.

Much of it came back as briefed. The first outline had all seven sections in my order, including "The 24-Week On-the-Job Plan" with a practice log of its own, a section for the team leader's checklist and a job aids section.

The issue: it turned tools meant for the job into material to test. The team leader's debrief checklist, which I had said was for the team leader and not a quiz, ended in a 3-question quiz for the learner. The job aids ended in a 4-question quiz, the open-text case practice became a 3-question quiz, and the conversation plan had a quiz too: four quizzes in all.

One follow-up message fixed it. I asked it to keep the sections and their order, remove all four quizzes, treat the checklist and job aids as tools for the job rather than content to test, and make the case practice and the first conversation plan open text. After reloading the outline I checked it: seven sections, no quizzes and six written tasks, including "Three case responses" in open text. I then read the generated pages. "The Weekly Debrief Checklist" is addressed to the team leader, supports "a 20-minute weekly debrief" and says "It is not a learner quiz and does not replace approval decisions". "The 24-Week On-the-Job Plan" tells the officer to "Pause for approval when a decision needs it, especially for any breach notice or tribunal application."

I checked the outline and those pages only, not a run with tenancy officers or team leaders. The estimate rose from about 49 minutes to about 1 hour 34 minutes against the 45 I asked for, so it needs trimming, and the job aids now carry three short tasks of their own. The generated pages do not assign a real caseload or ensure a team leader holds the debrief. That part belongs to the people in the plan. With the quizzes gone, the outline keeps the course to the job of the 10: practice for real conversations, and a checklist the team leader runs in the weekly debrief.

Manager's weekly debrief checklist

A checklist for the manager's weekly 20-minute debrief in a 70-20-10 plan, with questions for each stage, what to look for in the work, when to hand over more, and a debrief record. Filled in for a team leader debriefing new tenancy officers on rent arrears, with a blank version. No scores. Editable Word file, no email required.

Putting the 70-20-10 model to work

The 70-20-10 model will not tell you where people learn or how to split a budget. Its numbers are rounded memories of 191 executives from the early 1980s, and its lead author calls the rule folklore. It is still a useful planning prompt, because it asks three questions that a course brief usually skips: which real tasks will build this capability, who will give feedback on them, and what formal input prepares them sufficiently for the first one. The evidence that does exist points one way: experience without feedback and design disappoints, so this plan gives the manager and buddy explicit feedback responsibilities.

Your next step: pick one role you develop and book the manager's first three debriefs before you build any course. Then write the course to prepare for the tasks those debriefs will cover.

Frequently Asked Questions

The 70-20-10 model is still relevant as a planning reminder, not as a research finding. The numbers come from rounded counts of career events that 191 executives recalled in the 1980s, and a 2018 review describes the evidence for a fixed share of informal learning as weak. Used to plan job tasks, feedback and short timed courses together, it still helps.

Pick one role and one capability. List the real tasks from low risk to stretch, name the people who give feedback and what each does (starting with the manager), and time the required course or reading to the first task that needs it. Check progress in the work, and revise the plan after the first group.

There is no research finding called the 70:20:10 rule for adult learning. A figure sometimes linked to it comes from Allen Tough's 1971 study of 66 adults interviewed in 1970, which found about two thirds (68%) of their deliberate learning projects were planned by the learners themselves. That measures who planned the learning, not experience, people and courses.

Michael Lombardo and Robert Eichinger first printed the 70, 20 and 10 figures in their 1996 Career Architect Development Planner. They drew on Center for Creative Leadership research reported by Morgan McCall, Michael Lombardo and Ann Morrison in The Lessons of Experience (1988). Charles Jennings later popularised the model and co-founded the 70:20:10 Institute.

No. The 70-20-10 budget rule splits personal income, and other 70-20-10 splits are used for marketing budgets and innovation portfolios. The 20-70-10 split is a way of ranking staff performance. The learning model is about where development at work comes from, and its numbers are not meant as a budget.

Carolina Martin, Learning Experience Manager at Coursebox AI

Carolina Martin

Customer Success Manager

Carolina is a highly experienced learning designer with tech start-up experience. She is based in Sydney and previously designed courses alongside Coursebox founder Travis in a previous job. She is passionate about delivering engaging learning experiences and brings the same perspective to customer experience too. She manages our customer support documentation and product testing processes.