What I Learned Leading Transformation for 33,000 Employees at HSBC
Created on 2026-02-06 09:20
Published on 2026-02-23 09:45
The methodology that achieved 93% awareness and 72% participation at enterprise scale
I was one of five managers selected globally to deliver a transformation program at HSBC.
The scope was staggering. 33,000 employees serving 10 million customers across Asia Pacific. Multiple countries. Multiple languages. Multiple cultures. Entrenched ways of working that had developed over decades.
The goal was behavior change at scale. Not awareness. Not training completion. Actual change in how people worked.
Most transformation programs at this scale fail. They produce impressive training statistics and negligible behavior change. People attend the sessions, tick the boxes, and return to doing exactly what they did before.
We achieved something different.
93% awareness. 72% active participation.
Not attendance. Participation. People actually engaging, changing behavior, sustaining new practices.
This article shares what I learned from that experience and how it applies to AI transformation today.
Why Most Transformation Programs Fail
Before explaining what we did, let me explain what most programs do wrong.
They push instead of pull.
Most transformation programs push content at employees. Mandatory training. All-hands announcements. Email campaigns. Intranet articles.
The organization pushes. Employees receive passively. Nothing changes.
Push creates compliance, not commitment. People do the minimum required to avoid consequences. They attend the training. They acknowledge the email. They continue as before.
Real transformation requires pull. People must want to engage. They must see value in participation. They must choose to change, not be forced to comply.
They broadcast instead of network.
Most transformation programs treat communication as broadcasting. Senior leaders record videos. Corporate communications drafts messages. Content flows from center to periphery.
This ignores how change actually spreads.
People change because of other people they know and trust. They watch what peers do. They listen to colleagues they respect. They follow examples from their own teams.
A video from a senior executive they have never met changes nothing. A colleague they trust adopting new practices changes everything.
They measure activity instead of adoption.
Most transformation programs measure what is easy to measure. Training completion rates. Communication reach. Event attendance.
These metrics are meaningless. 100% training completion with 0% behavior change is failure, not success.
Real measurement requires tracking adoption. Are people actually using new practices? Is behavior actually changing? Are results actually improving?
This is harder to measure. That is why most programs avoid it.
They launch instead of sustain.
Most transformation programs are events. Big launch. Initial energy. Gradual fade.
The organization moves on. New priorities emerge. The transformation becomes yesterday’s initiative.
Real transformation is sustained. Not a moment but a movement. Not a launch but a journey.
Sustaining requires ongoing attention, resources, and reinforcement that most organizations are unwilling to provide.
The Sparks Methodology
At HSBC, we took a different approach.
We did not push transformation at 33,000 employees. We built a network of people who pulled their colleagues toward change.
I call these people Sparks.
Finding the Sparks:
In every function, every team, every location, there are opinion leaders. People others watch. People whose adoption signals that something is worth paying attention to.
These are not necessarily senior people. They are respected people. Credibility comes from competence and character, not just position.
We identified Sparks through a systematic process:
Who do people go to for advice? Not officially, but actually.
Who tries new things first? Who experiments while others wait?
Who influences how their team works? Not by authority, but by example.
Who communicates well? Who can explain things in ways that resonate?
We asked managers to identify these people. We asked peers to identify these people. We triangulated until patterns emerged.
The result was a network of Sparks across the organization. Not a large percentage. Perhaps 5-10% of the workforce. But the 5-10% who influenced the other 90%.
Equipping the Sparks:
We invested disproportionately in the Sparks.
While typical transformation programs spread resources thinly across everyone, we concentrated resources on opinion leaders.
Sparks received deeper training. Not just what to do, but why it mattered and how to help others.
Sparks received early access. They experienced new practices before their colleagues. They could speak from experience, not just instruction.
Sparks received ongoing support. Regular check-ins. Problem-solving assistance. Recognition for their efforts.
This investment was not fair in the traditional sense. We were giving more to some than to others. But it was effective. The investment in Sparks multiplied through their influence on peers.
Activating the Sparks:
Equipped Sparks became active agents of change.
They demonstrated new practices in their daily work. Colleagues watched and learned.
They answered questions from peers. Informal peer support is more trusted than formal training.
They shared stories of success and struggle. Real stories from real colleagues are more persuasive than polished corporate messages.
They created social proof. When respected colleagues adopt something, it becomes safer for others to follow.
We did not ask Sparks to become trainers or change agents as a formal role. We asked them to be visible practitioners. The influence happened naturally.
Letting them pull:
Here is what most programs miss.
You cannot push people to change. You can only create conditions where they pull themselves toward change.
The Sparks created those conditions.
When a respected colleague is using new practices successfully, others want to know more. They approach the Spark. They ask questions. They try things themselves.
This is pull, not push. The energy comes from the periphery, not the center.
Our role was to support this pull. Ensure Sparks had answers. Provide resources when people sought them. Celebrate adoption when it occurred.
We stopped trying to make change happen. We started enabling change to happen.
The Results
93% awareness. 72% active participation.
Let me break down what these numbers mean.
93% awareness means that nearly everyone in the 33,000-person scope knew about the transformation, understood what it meant for them, and could articulate key elements.
This was not awareness from a single email blast. This was awareness built through personal networks. People heard about the transformation from colleagues they trusted, in terms that made sense for their role.
72% active participation means that nearly three-quarters of the workforce engaged in new practices. Not attended training about new practices. Actually adopted them.
This number mattered more than the awareness number. Participation meant behavior change. Behavior change meant results.
The 72% was not uniform. Some functions reached higher. Some reached lower. But across the full scope, nearly three-quarters of employees changed how they worked.
For enterprise transformation, this is exceptional. Most programs consider 30-40% adoption a success. We more than doubled that.
Why It Worked
The Sparks methodology worked for specific reasons that apply to AI transformation.
Trust transfer:
People trust people they know more than institutions they do not.
When transformation comes from corporate headquarters, it carries institutional credibility but lacks personal trust. People comply without committing.
When transformation comes from respected colleagues, it carries personal trust. If Sarah is doing this and Sarah is excellent at her job, maybe I should pay attention.
The Sparks transferred institutional initiative into personal recommendation. They made corporate transformation feel like peer advice.
Visible proof:
Sparks provided visible proof that new practices worked.
Not case studies from other companies. Not promises from leadership. Actual proof from colleagues in the same context, facing the same challenges, achieving real results.
This proof was more persuasive than any presentation. People could see it working. They could ask questions. They could examine closely.
Safe experimentation:
When a corporate program demands change, failure feels risky. What if I cannot do it? What if I look incompetent?
When a colleague demonstrates change, experimentation feels safer. I can try what Sarah is doing. If I struggle, I can ask her for help. It is learning, not performance.
Sparks created safe space for experimentation by being accessible sources of support, not evaluators of performance.
Sustained energy:
Corporate programs launch with energy that fades. Sparks provided sustained energy.
Every day, Sparks demonstrated new practices. Every week, they answered peer questions. Every month, their influence expanded.
The energy did not come from the center in periodic bursts. It came from the network continuously.
Applying This to AI Transformation
Everything I learned at HSBC applies directly to AI transformation.
Your Sparks exist already.
MIT’s research found that over 90% of workers are already using personal AI tools. This is your Shadow AI Economy.
Within that 90%, some are using AI exceptionally well. They have figured out effective prompts. They have integrated AI into their workflows. They are getting results that others are not.
These are your Sparks.
Find them. Who is already using AI effectively? Who are others asking for AI help? Who experiments and shares what they learn?
Your Sparks are not future trainees. They are current practitioners. Find them where they already exist.
Equip your Sparks disproportionately.
Do not spread AI resources thinly across everyone. Concentrate on Sparks.
Give Sparks access to better tools than others have. Give them advanced training. Give them support and attention.
This feels unfair. It is effective.
Sparks who are well-equipped become more influential. Their success becomes more visible. Their ability to help peers expands.
Let Sparks pull their peers.
Do not mandate AI adoption. Create conditions where people want to adopt.
When Sparks demonstrate what AI makes possible, others become curious. When Sparks share how they are saving time and improving quality, others want to learn.
Your role is not to push AI at everyone. Your role is to ensure that when people become curious, they find answers.
Support the pull. Ensure Sparks have time to help colleagues. Create easy ways for curious people to learn. Celebrate adoption when it occurs.
Measure adoption, not activity.
Do not report how many people completed AI training. Report how many people are actually using AI.
Better yet, report what using AI is producing. Time saved. Quality improved. Decisions enhanced.
Activity metrics let programs claim success while changing nothing. Adoption metrics force programs to produce actual change.
Sustain, do not launch.
AI transformation is not an event. It is a journey.
The Sparks methodology provides sustained energy. Sparks continue demonstrating, helping, and influencing long after launch events fade.
Design for sustainability from the beginning. How will energy be maintained at month six? Month twelve? Year two?
The Prosumer Path
One specific pattern deserves attention: the Prosumer Path.
Prosumers are people who use AI for personal productivity before organizations provide official tools. They are consumers of AI who have become sophisticated users.
In most organizations, Prosumers are your best Sparks.
They have already overcome the learning curve. They have experimented with prompts. They have integrated AI into how they work. They have developed the Auditor Mindset through practice.
Prosumers represent capability that already exists. You do not need to develop it. You need to recognize and leverage it.
Finding Prosumers:
Ask directly. “Who is using AI tools like ChatGPT or Claude for work, even if informally?”
You may be surprised by how many people raise their hands. And you may be surprised by who they are.
Prosumers are not necessarily young or tech-savvy by stereotype. They are curious, willing to experiment, and looking for better ways to work.
Learning from Prosumers:
Prosumers know things your official AI program does not.
They know what AI tools are actually useful for. They have discovered use cases through experimentation.
They know what does not work. They have tried things that failed.
They know what barriers exist. What limitations are frustrating. What support would help.
Learn from Prosumers before designing your program. Their experience is invaluable.
Empowering Prosumers:
Prosumers often operate in shadow because official channels do not support them.
Bring them out of shadow. Legitimize what they are doing. Provide better tools than they can access personally.
When Prosumers feel supported rather than hidden, their influence expands. They become visible examples rather than quiet practitioners.
What Mid-Market Can Learn
The HSBC experience was enterprise scale. But the principles apply to mid-market organizations with important differences.
You have fewer Sparks, but they are more accessible.
In a 200-person organization, you might have 10-20 natural Sparks. You can know them personally. You can support them directly.
This is easier than managing a Spark network across 33,000 people. Use this advantage.
The CEO can be a Spark.
In mid-market organizations, the CEO is visible to everyone. Their behavior sets the tone.
If the CEO visibly uses AI, it signals importance. If the CEO asks “what did AI say about this?” in meetings, it normalizes AI use.
In enterprises, CEO behavior is distant. In mid-market, it is immediate. Use this leverage.
Pull happens faster.
With fewer people and closer relationships, pull dynamics accelerate.
When respected colleagues adopt AI, word spreads quickly. Curiosity emerges fast. Adoption can cascade rapidly.
This is the 90-day advantage. What takes enterprises quarters can happen in weeks for mid-market.
But resistance is also more visible.
In mid-market, skeptics and resisters are also more visible. Their influence can be significant.
Address resistance directly. Understand concerns. Either convert skeptics or ensure their resistance does not block willing adopters.
You cannot hide from resistance in mid-market. Face it.
The Numbers That Matter
Let me revisit the numbers from HSBC.
93% awareness. 72% participation.
These numbers mattered because they measured what actually changed.
For AI transformation, comparable numbers might be:
What percentage of employees can accurately describe what AI tools are available and what they can do? (Awareness)
What percentage of employees are actively using AI tools for meaningful work? (Adoption)
What percentage of AI usage is producing measurable value? (Impact)
Most organizations can answer none of these questions. They know how many licenses they purchased. They know how many people completed training. They do not know whether anything actually changed.
Measure what matters. Awareness. Adoption. Impact.
The technology is the same for everyone. The adoption approach is what differs.
At HSBC, we did not have better technology than other banks. We had a better approach to getting people to use technology.
The Sparks methodology created pull instead of push. It leveraged networks instead of broadcasting. It measured adoption instead of activity. It sustained energy instead of launching events.
The same principles apply to AI transformation today.
Find your Sparks. Equip them disproportionately. Let them pull their peers. Measure what matters. Sustain the energy.
This is how 93% awareness and 72% participation happens.
This is how transformation actually transforms.
Who are the Sparks in your organization? Who is already using AI effectively and could influence peers?
The AI Readiness Scorecard helps identify capability across your organization, including where natural Sparks might exist. It takes ten minutes and shows where your Human Layer is strong and where it needs work.
Comment “SCORECARD” below and I will send you access.
Transformation is not pushed. It is pulled. The question is whether you are creating conditions for pull.