Elevating Support Through Knowledge Engineering

Support that scales without more agents

Symantec’s enterprise support organization was growing faster than it could hire. The usual answer is more agents. The better answer is customers resolving issues themselves, which only works if the knowledge base actually answers the question they came with.

I co-led the knowledge work that made that shift, and this is where the KCS practice I still run today was built.

Knowledge written by the people solving cases

Knowledge-Centered Service starts from a simple observation: the person who just solved the problem is the person who should write it down, while they’re solving it. Not a technical writer weeks later. Articles become a byproduct of doing the work rather than a separate documentation effort.

That only holds if support engineers are trained and supported to work that way. So the program was mostly about people. I helped establish and mentor a network of 1,500+ KCS publishers and 150+ coaches across global support teams, with standardized training on content refinement, adoption practices and quality assurance. That training became the foundation the program ran on — and the model that later scaled through the Broadcom acquisitions.

Fixing the content customers actually hit

Working with publishers, I refined articles for accuracy, usability and findability, so global enterprises could resolve issues through support.symantec.com without opening a case.

Deciding what to fix first came from the data. I analyzed support trends to find where knowledge was missing and where customers got stuck, then used that to prioritize what got written and what got rewritten. That turns content work from guesswork into a queue ordered by what’s costing the most.

A chatbot on top of governed knowledge

An Azure-powered chatbot extended self-service further, and I managed the knowledge side of it: the intents, entities and answers that decided what it could actually resolve. It drew on the same knowledge bases the support site used, so improvements to article quality raised the bot’s answers at the same time.

What it produced

• 20–35% case deflection globally through optimized self-service.
• 25% improvement in first-contact resolution, from standardized training and quality processes.
• 1,500+ publishers and 150+ coaches trained in KCS methodology.
• Data-informed practices that tied content decisions to measurable support outcomes.

The lasting result was structural. Symantec’s knowledge program became the base that survived acquisition and scaled across Broadcom, VMware and legacy Broadcom teams — because the practices were built to be taught, not to depend on any one person.