Diagnose
Find the behavioural, structural and talent-system constraint behind the visible problem.
TURNOVER REDUCED
NPS IN GREATER CHINA
NPS IMPROVEMENT
Sixteen years in luxury retail building talent pipelines, leadership capability and service cultures — most of it spent on the harder half of the job: making the change survive after the training room empties.
The real work of L&D
01 / Point of viewIf learning disappears when people return to work, the programme has delivered activity — not value.
My work starts beyond the classroom. I connect learning with role expectations, manager reinforcement, talent decisions and operating data, so new behaviour survives contact with the real organisation. When that system holds, capability compounds into culture — and culture appears in retention, succession, service and growth.
In my current Hong Kong and Macau remit, I support a frontline population of approximately 120 people. Managers sat at the top, talent at entry level, and almost no one was ready in between. More training would not have fixed it. The structure had no rung to promote anyone onto, so the training had nowhere to land.
How the work moves
The sequence is deliberate: diagnose the system before designing the intervention; deliver learning, then develop the surrounding capability until the behaviour becomes the way work is done.
Find the behavioural, structural and talent-system constraint behind the visible problem.
Build the right combination of learning, assessment, roles, reinforcement and data.
Create an experience that makes the required behaviour clear, practicable and relevant.
Equip managers and talent pipelines to reinforce, coach and sustain the new behaviour.
Connect expectations to routines, talent decisions and shared standards.
Track what changes in retention, succession, service, performance and growth.
Selected transformations
Redesigned a stalled three-tier hierarchy into a five-tier progression system, then connected assessment, development and promotion decisions to rebuild internal mobility.
Translated brand heritage and personalised client connection into observable service habits, manager coaching and performance routines.
Expanded a curriculum from 3 to 19 modules and supported talent acquisition and onboarding for a retail network growing from 5 to 20 stores.
AI-enabled implementation
I build these myself. Not as a technical credential, but as a working constraint: designing a process and then having to implement it forces a precision a document never demands. Ambiguity survives in a framework. It does not survive in a database schema.
Experience trajectory
Talent cycle, structure, succession, performance, service culture and a three-person team across a Hong Kong and Macau luxury retail business.
Coaching-led behavioural change for B2B solution positioning and cross-functional performance.
Capability engine, retail expansion, leadership pipeline and service standards for Longines.
Management-trainee development and award-winning service capability at DFS.
Frontline quality assurance, facilitation and performance development across PCCW and Broadway.
Credentials & education
A deliberately connected portfolio: professional membership, assessment practice, leadership development, AI-enabled HR and formal study.
Research translated into practice
My research is not a separate academic lane. It sharpens how I diagnose behaviour, design talent systems and embed change into the operating culture.
Transfer failure is one of the most reliable findings in the training literature and one of the most consistently ignored. The gap between knowing and doing is a management problem, not a training one — which means it cannot be fixed inside the training room, and most attempts to fix it there are spending money on the wrong side of the wall.
Read the full perspectiveVITALIS™ structures observation across three distinct axes: Capability — what a person can demonstrate now; Growth Momentum — the direction and pace of development; and Core Vitality — the energy and sustainability available to keep progressing. The measures remain separate so managers can see where development action is actually needed.
The useful question is not which tasks AI can absorb, but which judgements must stay attributable to a named human. In talent decisions that boundary is not a preference — it is the thing that makes the decision defensible to the person it affects.
On evidence
02 / StandardThe figures on this page are organisational outcomes from one role over three years, across a combined intervention in structure, learning, assessment, manager reinforcement and performance design.
They should be read as the result of an operating system working together. No single workshop, instrument or dashboard produced them in isolation.
Measurement can sharpen judgement; it should not be used to manufacture certainty.
I'm always glad to talk with people working on the same problems — talent systems, behaviour change, organisational performance. A shared question, a piece of research, or a role.
joeytsaiwork@gmail.com ↗ LinkedIn / Joey Tsai ↗