Enterprise knowledge Case study Shipped

Knowledge-Centered Service (KCS) Transformation

Led global KCS and Knowledge Management strategies at Broadcom through major acquisitions, reducing support volume by 20-25% and generating $30M+ in cost savings.

Project
Broadcom
Role
KCS Program Manager
Timeline
2019–2025
Broadcom company logo
$30M+
Cost savings (2024)
20-25%
Case volume reduction
Mid-90s
CSAT score
160K+
Articles managed

Scaling enterprise KCS through acquisition

I came to Broadcom through the Symantec acquisition, and served as KCS Program Manager from November 2019 to February 2025 — scaling the KCS program across global support organizations, covering strategy, roadmap and execution for a knowledge base of more than 160,000 articles.

That knowledge base was the engine behind self-service and chatbot-assisted resolution across every support channel. When it worked, customers solved their own problems. When it didn’t, they opened a case.

Running KCS as a measured program

Knowledge-Centered Service is the methodology underneath the work. Its core idea is that knowledge gets captured while the work is happening, by the people doing it, so articles are a byproduct of solving cases rather than a separate documentation project.

Making that stick at enterprise scale takes measurement, so I helped establish the program’s operating rhythm: recurring KCS health checks, article quality reviews, and leadership dashboards built on deflection, usage, quality and search metrics. Those dashboards are what turned knowledge from a thing people believed in into a thing the business could see.

Coaching 1,800 people to write differently

KCS succeeds or fails on adoption, and adoption is a coaching problem.

I coached and enabled 1,500+ publishers and 300+ KCS coaches across the global support organization, working on article quality, consistency and the standards that make an article reusable. That work is what moved first-contact resolution — an engineer who finds a good article resolves the case on the first touch instead of escalating it.

Findability, hands on

I analyzed search and knowledge performance to find where content was failing people: gaps with no article behind them, queries returning nothing useful, articles nobody could surface.

Then I acted on it. I tuned search relevance personally, guided Knowledge Engineers and support teams on what to fix, and drove content gap closure so the missing answers got written. Better findability shows up in three places at once — self-service, chat resolution, and the cases that never get opened.

Change management through two acquisitions

I went through this from both sides. In 2019 Broadcom acquired Symantec’s Enterprise Security division for $10.7 billion and I moved across with it, as part of the organization being absorbed. In 2023 Broadcom acquired VMware for $69 billion, and that time I was on the receiving side, helping bring an incoming organization in.

Each company arrived with its own tools, its own processes and its own culture around knowledge. I played a central role in the KCS change management that brought them together — aligning the tooling, reconciling the processes, and doing the slower work of getting people from different companies to agree on what a good article looks like.

What it produced

  • 25% year-over-year reduction in global support case volume in 2024.
  • $30M+ in support cost avoidance that year, from actual deflected case counts at approximately $400 per case.
  • Mid-90s CSAT, held through both acquisition transitions.
  • 1,500+ publishers and 300+ coaches enabled in KCS practice.

One governed source, every channel

The structural result is that the knowledge base became the system of record. Articles fed search, the chatbot and the support engineers from one governed source, so improving an article improved every channel at once.

That is what makes knowledge usable by AI, and it’s the same principle behind the retrieval work I do now.

Questions about this work?

Happy to go deeper on any of it — the decisions, the trade-offs, or what I'd do differently.