Reflections on a Bank of America Action Learning Initiative — and What the Field Has Learned Since
Reflections on a Bank of America Action Learning Initiative — and What the Field Has Learned Since
Building on original research published in The Change Champion's Field Guide, Jossey-Bass/Pfeiffer, 2013.
Keywords
Action learning; leadership development; reflection; organizational learning; learning transfer
Abstract
In 2003, Bank of America launched Crack the Code — a large-scale action learning initiative in which 27 high-potential learning leaders worked in two teams over three months to address two critical enterprise learning challenges while simultaneously developing as leaders. The program met its business objectives. This account revisits it through the lens of what the author — an external consultant and program designer at the time — knows now. It examines what the program did well, what it missed, and what a redesign would include today. Four structural gaps are identified: the absence of individual leadership competency baselines and development goal-setting; the lack of real-time capture and structured between-session reflection infrastructure during the program; the absence of a prepared receiving organization; and the absence of a post-program continuity infrastructure — leaving insights and behavioral commitments to fade once the teams disbanded. The account introduces the Reflective Practice Framework — a four-mode framework for designing reflective practice across individual and team dimensions — and proposes Augmented Action Learning as an enhanced design approach incorporating structured pre-program, during-program, and post-program elements. Implications for action learning practitioners and program sponsors are offered.
Case Study Synopsis
Background
By 2003, Bank of America had grown through acquisition from $147 billion in assets operating in 10 states to $736 billion in 21 states and 30 countries. Each acquired organization brought its own learning function, its own processes, and its own priorities — none of them integrated, none of them aligned. The result was a learning infrastructure that reflected how the company had been assembled, not what it had become.
Kenneth D. Lewis, who had succeeded Hugh McColl as CEO, sponsored an initiative to address this directly and put the bank's next generation of learning leaders at the center of solving it. He called it Crack the Code.
Perceived Current State at the Time
Bank L&D processes
The learning organization was under significant pressure. More than 70% of training was classroom instructor-led. Content deadlines were met only half the time. Average program shelf life was three to six months, against an industry norm of six to twelve. Multiple non-standardized processes across lines of business created variation in user experience, unnecessary cost, and confusion for employees seeking learning. There was no central learning function and no single head of learning.
Participant leadership capabilities
The 27 participants were high-potential learning leaders selected for their performance history and potential for greater organizational responsibility. Most had deep expertise within their own lines of business but limited experience operating across organizational boundaries — leading without authority, navigating senior stakeholder relationships, and driving change in a large, complex organization. For most, the Six Sigma DMAIC methodology they would be asked to use was new territory.
Program Goals and Process
Business goals and leadership development goals
The program was designed to achieve two goals simultaneously. The original case study stated the intent directly:
"The project was designed to be both an important business initiative whose outcomes would have a significant impact on the capabilities of the organization, and a learning initiative that would accelerate and drive the leadership capabilities of the participants."
The business goals were specific: recommend approaches to optimize the investment in and management of key learning processes, tools, programs, and technology across the enterprise. Two business challenges were defined — Learning Administration and Learning Design, Development, and Delivery — and one team was assigned to each. Documented leadership development goals asked participants to deepen their understanding of the business challenges, enhance their ability to apply Six Sigma methods, and build knowledge of best practices in learning.
This dual-objective framing — explicit, documented, and owned by senior leadership — was one of the program's genuine design strengths. It is also the lens through which its design gaps are most clearly seen.
CTC process description
Crack the Code ran over approximately three months through three face-to-face sessions with intensive individual and team work between them. Session 1 defined the business challenges and launched each team's Six Sigma work. A Benchmark phase of data gathering, stakeholder interviews, and best practice research followed. Session 2 was a check-in at which teams reviewed findings, sharpened their recommendations, and assessed team effectiveness — a designed moment for developmental reflection that did not always receive the attention it warranted. Final recommendations were developed in the weeks before Session 3, at which teams presented to senior Learning Leaders and Line of Business partners. Recommendations were accepted and implementation plans were put in place.
Program Structure
The overall program architecture is illustrated below.
Figure 1. Crack the Code Program Architecture. Source: O’Brien & Terrell (2013).
Each team was supported by a cross-functional structure including a Program Sponsor, steering committee, Team Sponsors, an external Process/Technical Coach, Six Sigma resources, and a Program Office. External learning expertise was provided by three consulting firms, including the author's.
Reported Program Outcomes
Both teams produced data-grounded, stakeholder-validated recommendations that senior leadership accepted and moved into implementation: a single Learning Management System, a Learning Content Management System, a Learning Standards Committee, and a cross-functional Process and Standards Working Group.
On the development side, the original case study documented that action learning worked as a live testing ground — certain individuals stepped up to lead and influence, while others remained in the background. Participants built cross-organizational relationships that extended beyond the project and reported increased confidence in stakeholder management and executive-level influence. These were meaningful outcomes, honestly reported through participant and stakeholder feedback. What the program did not produce was a systematic, pre-established measure of behavioral change against which post-program growth could be demonstrated. The development that occurred was real; the evidence of it was anecdotal.
Critical Issues Observed
AL team members returned to an unchanged environment
When the program ended and the teams were disbanded, participants returned to organizational roles and environments that were largely as they had left them three months earlier. The culture had not changed. The systems, norms, and expectations shaping daily leadership behavior remained unchanged. Every action learning program faces this tension: the program creates a bounded space of intensity, accountability, and genuine development — but the organization outside that boundary continues on its own logic. When participants cross back, that logic reasserts itself quickly.
The receiving organization was not prepared
No formal preparation was made for the managers, supervisors, or immediate teams that participants would return to. Sponsors had not been asked to create conditions that would support the application of new learning. Without that preparation, the organization that had invested in developing 27 leaders was structurally unable to benefit from what those leaders brought back. What happened beyond the program boundary was left to chance.
The organization re-absorbed team members without infrastructure for integration
There was no formal mechanism to help participants integrate what they had experienced, learned, and committed to during the program into their returning work lives. The original case study's final statement was simple: "The action learning teams were disbanded after the recommendations were presented and accepted."
Insights and behavioral commitments were not captured in any structured way, not consolidated across teams, and not shared with organizational leadership as a source of institutional learning. They faded at the rate that memories and notebooks do. The organization received the project deliverables. The developmental investment beyond that produced no lasting artifact and no ongoing infrastructure.
Challenges and Successes
Challenges
The Six Sigma/DMAIC methodology was unfamiliar to most participants and the learning curve was steep. The more consequential challenge was the pull toward task performance at the expense of developmental attention. The original case study captured it directly:
"Talented, high potential people are known to seek new, difficult challenges and to learn rapidly from experience. We found — as expected — that team members became very focused on the issue very quickly, as they wanted to solve the real, meaty, important business problem for the company. However, it was less natural for them to reflect on their individual or team learning objectives, or to explicitly focus on deriving 'lessons learned' from their experience."
This dynamic was observed and documented at the time. What was not recognized then is how structurally inevitable it was — and how much more the design needed to counteract it.
Successes
The program met its business objectives and demonstrated that action learning could operate effectively at this scale in a large, complex organization. The stakeholder engagement strategy was particularly strong — stakeholders were briefed throughout, consulted on preliminary recommendations, and met with teams the evening before final presentations. No senior leader was surprised on presentation day. The recommendations moved because the relationships had been built.
Reflections and Perspective
Knowing What We Knew at the Time
What we missed or didn't design for
We neglected to establish leadership competency baselines. No pre-program 360° assessment was conducted. Participants entered without a documented picture of their current capabilities and exited without a systematic measure of what had changed. Nor did the program translate its developmental objectives into specific, documented individual goals — goals tied to a baseline and carried into Session 1 as personal commitments, not only cohort-level intentions. Development was treated as something that would emerge from the experience of doing the work, rather than something designed for with the same rigor applied to the business challenge.
What we could have done better
Built a focus on learning mindset and development goals into the program's structure from the outset — not as an orientation add-on, but as a commitment participants encountered throughout, with the same accountability as the task deliverables. Built a more rigorous and systematic focus on reflection into the between-session periods, not only during formal sessions. Provided more active coaching between sessions, focused on developmental patterns and team dynamics. Captured individual and team insights at key points during the program, and synthesized them at close across both teams — a final combined synthesis presented to leadership would have transformed private experience into shared organizational learning.
Knowing What We Know Now
Two areas — one available but underused in 2003, one developed since — would have changed the program's design significantly.
Learning Mindset. A learning mindset — the disposition to approach experience as a source of learning rather than primarily as a performance challenge — was available as a framework but was not built into the program's design. The pull toward performance in high-stakes environments is powerful and predictable. Building that orientation into the program's front end, prior to Session 1, would have created a shared vocabulary for developmental attention that coaches could reinforce throughout.
Reflection as a key tool. The original case study's observation — that reflection was present when coaches created it and seemingly absent when they did not — pointed not toward a need for more coach-led reflection sessions, but toward a need to design for the full range of reflective practice deliberately, across both individual and team dimensions, and throughout the program rather than only in formal sessions.
The Reflective Practice Framework
Reflection is not a single act. It is a family of practices that differ along two dimensions: how structured they are, and whether they happen individually or collectively. The Reflective Practice Framework maps these dimensions onto a simple grid.
<table> <colgroup> <col style="width: 20%" /> <col style="width: 39%" /> <col style="width: 39%" /> </colgroup> <thead> <tr> <th><strong>REFLECTION MODES</strong></th> <th style="text-align: center;"><strong>Individual</strong></th> <th style="text-align: center;"><strong>Team</strong></th> </tr> </thead> <tbody> <tr> <td><strong>Ad hoc, opportunistic</strong></td> <td><p><strong>Q1 · Reframe</strong></p> <p>In-the-moment noticing; Schön's reflection-in-action. A mental note after a hard conversation, adjusting course mid-meeting, naming a reaction as it rises.</p></td> <td><p><strong>Q3 · Huddle</strong></p> <p>Spontaneous shared sense-making in the flow of work. A five-minute regroup after a meeting, a corridor debrief, a team catching itself and course-correcting.</p></td> </tr> <tr> <td><strong>Planned, structured</strong></td> <td><p><strong>Q2 · Examine</strong></p> <p>Scheduled, framework-guided self-examination. A weekly reflective journal, a coaching conversation, a 360° debrief tied to a development plan.</p></td> <td><p><strong>Q4 · Review</strong></p> <p>Facilitated, documented team reflection producing commitments. An After-Action Review, an action learning set, a project post-mortem. Research links this mode directly to team learning and performance.</p></td> </tr> </tbody> </table>Source: Terrell, Steve (2026). The Reflective Practice Framework.
Crack the Code operated almost entirely in Q4 · Review — facilitated collective reflection during formal sessions. The design built no infrastructure for Q1 · Reframe (individual noticing in real time), Q2 · Examine (individual structured reflection between sessions), or Q3 · Huddle (spontaneous team sense-making between formal meetings). Reflection happened when coaches created it. When coaches were not present — during the intensive between-session project work that occupied most of the program's calendar — the task objective ran uncontested. A program that designs for all four modes changes this: Q1 · Reframe and Q3 · Huddle through explicit norms set at the outset; Q2 · Examine through a structured individual practice such as a journal or coaching touchpoint — each mode operating through the same three-step process: Notice what is happening, Name it specifically, identify the Next action or insight to carry forward; Q4 · Review through more deliberate facilitation and documentation of outcomes.
AL Team Coaching. Coaching was active during sessions and more limited between them. A more complete model would have included structured between-session touchpoints — brief, focused conversations aimed not at advising on the task but at surfacing what was happening within the team and what individuals were observing about their own leadership patterns.
If We Could Do It Again: Augmented Action Learning
Augmented Action Learning retains everything that made Crack the Code effective and adds a structured support layer — before, during, and after the formal program. The core design remains unchanged: real problems, dual objectives, Six Sigma rigor, skilled coaches, and the stakeholder architecture that makes accountability real. The infrastructure wrapped around it is what changes. The full architecture is shown below.
Figure 2. Augmented Action Learning: The Crack the Code Program Architecture.
Pre / Before
Four to six weeks before Session 1, each participant completes a behavioral 360° assessment, receives feedback, and sets specific individual development goals — recorded and revisited at each session. A sponsor briefing establishes both objectives explicitly. Line managers are briefed on the program's developmental aims and asked to prepare to receive returning participants constructively — creating the transfer climate: the organizational conditions that either support or inhibit the application of new learning, which the original program left to chance.
During
A mobile-enabled, offline-capable journal gives participants a place to capture observations and insights in real time — in a branch visit, an executive's office, a stakeholder meeting — before the specificity of the insight is lost to memory. This supports Q1 · Reframe: the individual, in-the-moment noticing that the original program had no infrastructure to capture. Structured individual reflection prompts between sessions support Q2 · Examine: the planned individual practice that translates raw observation into insight. Between-session team touchpoints — brief, informal — support Q3 · Huddle: the spontaneous collective sense-making that builds shared understanding without waiting for a formal session. Session closes, with documented insights and behavioral commitments, complete the arc at Q4 · Review. The lifecycle runs: Reframe → Examine → Huddle → Review — and each completed cycle changes what gets noticed in the next. The platform also tracks each participant's progress against the individual development goals set at program outset, surfacing them at each session close as a prompt for reflection on growth alongside task performance. Structured reflection prompts and AI coaching between sessions sustain developmental attention when it is most easily crowded out by task pressure. At each session close, participants document the insights they have generated and specific behavioral commitments for the period ahead — named, observable, and tied to the experiences that produced them.
Post / After
The organizational reentry period — from program close through 90 to 120 days afterward — is when insight either consolidates into behavioral change or fades. Structured check-ins at 30, 60, 90, and 120 days ask participants to account for their commitments: What have you done differently? What made that harder than you expected? Between 90 and 120 days, a second 360° assessment administered to the same rater groups as the pre-program baseline produces the behavioral delta — not testimony, but data. A consolidated sponsor report gives the program sponsor a picture of behavioral change across the cohort, commitment follow-through rates, and implications for the organization going forward.
The post-program period should also address the organizational changes the teams' recommendations require. Implementation plans accepted at program close need an accompanying change enablement and sustainability framework — identifying who owns each change, what conditions must be in place for it to take root, and how progress will be tracked. Without that framework, accepted recommendations can quietly stall once the teams have disbanded and organizational attention has moved on.
Human coaches remain central throughout — to sense group dynamics, apply intuition, and read the room. The digital infrastructure does what platforms do well: consistency, documentation, and continuity across time.
Overall Lessons Learned
The more compelling the challenge, the more powerfully it pulls participants toward performance and away from reflection. Fostering a learning mindset from the outset is not a philosophical preference; it is a structural counterweight to a structural pull. Disciplined, scheduled reflection is a powerful antidote to the seduction of the problem.
Insights and commitments generated during a program are evidence of organizational learning — but only if they are captured, consolidated, synthesized, and purposely shared. Programs that end without recovering that knowledge leave the organization's investment only half-spent.
Technologies available today make the Augmented Action Learning model feasible at scale — but only with skilled human oversight. Coaches remain essential for what platforms cannot do: sensing what is actually happening in a room.
Implications for Action Learning Practitioners
Engage with program sponsors to gain clarity and commitment about the program having a balanced focus on business goals and learning and development goals, what that means in practice, and the sponsor's specific role in ensuring that balance throughout. A sponsor who has not had that conversation has implicitly authorised the business objective to take precedence.
Communicate the balanced focus to all parties — participants, AL team coaches, and specialists — so that everyone understands their role in pursuing both objectives. The developmental objective should be visible in the program's architecture at every stage, not only in its stated goals.
Engage, orient, and onboard participants in the use of a digital support platform before the program begins, so it functions as a development tool from the outset — not an administrative system they encounter for the first time mid-program.
Familiarize participants with the competency model used for their 360° assessment and give them clear guidelines on selecting raters. The assessment is most valuable when participants understand what is being measured and have selected raters with genuine knowledge of their leadership.
The original case study concluded: "This exercise proved that action learning programs can be an effective tool to accomplish both the solving of real business problems as well as developing key talent for the future." That conclusion stands. These reflections are an acknowledgment that the field has learned things since 2003 that would have made the program better — and that those learnings are available to every practitioner who designs a serious action learning initiative today.
The disbanding moment — when the teams conclude their work and participants return to ordinary organizational life — is not the end of the program. It is the beginning of the period when everything the program created will either take root or fade. Designing for that period is not optional for programs that intend to produce lasting change.
Acknowledgment
This article builds on original research published in The Change Champion’s Field Guide (Jossey-Bass/Pfeiffer, 2013). The original “Crack the Code” case study is the intellectual property of Best Practice Institute and is reproduced here with the permission of Louis Carter, Founder and CEO of Best Practice Institute® and Most Loved Workplace®. For more information, visit bestpracticeinstitute.org, mostlovedworkplace.com, and louiscarter.com.
References
O'Brien, Eryn, and Terrell, Steve. 2013. "Crack the Code: A Case Study on How Bank of America Optimised Its Investment in Learning Through Action Learning." In The Change Champion's Field Guide, edited by Louis Carter, Roland L. Sullivan, Marshall Goldsmith, Dave Ulrich, and Norm Smallwood. San Francisco: Jossey-Bass/Pfeiffer.
Schön, Donald A. 1983. The Reflective Practitioner: How Professionals Think in Action. New York: Basic Books.
Terrell, Steve. 2026. "The Reflective Practice Framework." Unpublished framework.
About the Author
Steve Terrell, Ed.D. holds a doctorate in Human Resource Development from The George Washington University and serves as Practitioner Editor of The Learning Organization: An International Journal. He is the author of Learning Mindset for Leaders (2020) and has published in People + Strategy, the Journal of Management Development, Training Magazine, and the Jossey-Bass Change Champion’s Fieldguide. He was an external consultant and program co-designer on the original Crack the Code initiative.