Helping Support Reps. Get to Answers Faster

Helping Support Reps. Get to Answers Faster


Customer Support Agents had to navigate separate systems—Salesforce for caller data, an internal tool for account info and McAfee websites for product queries.
Disconnected tools, too much content, and internal documentation that didn’t reflect how agents actually worked.


Customer Support Agents had to navigate separate systems—Salesforce for caller data, an internal tool for account info and McAfee websites for product queries.
Disconnected tools, too much content, and internal documentation that didn’t reflect how agents actually worked.

Ethnographic Research

Prototyping

Context

Observations in the field. An eye-opening trip to LATAM

Observations in the field. An eye-opening trip to LATAM

In February 2025, I visited our call center in Guatemala to shadow support agents over 3 days.

Initially, I was paired with experienced agents—resourceful, quick, and relying heavily on personal workarounds (Word docs, sticky notes, pricing cheat sheets).

It looked smooth… until I asked to observe newly onboarded agents.

That’s when the cracks surfaced.

In February 2025, I visited our call center in Guatemala to shadow support agents over 3 days.

Initially, I was paired with experienced agents—resourceful, quick, and relying heavily on personal workarounds (Word docs, sticky notes, pricing cheat sheets).

It looked smooth… until I asked to observe newly onboarded agents.

That’s when the cracks surfaced.

Each agent cohort organised information differently

Each agent cohort organised information differently

Experienced agents each had unique methods of organising information

Insight - Agents were finding workarounds to poor processes

A Word doc with every relevant link organised by subject

Sticky notes with price points and shortcuts to commonly asked questions

Taping a pricing chart to the wall


Real Insight Came From Watching New Agents Struggle

New agents set a timer for 1 minute and placed the caller on hold while they searched for answers.

Agent looks up SKU inside of tool

Agent places the customer on hold

A given topic could have 6-9 articles

1 min. Eternity for the customer. Not enough time for an agent.

01
min
00
sec

Freshly trained agents couldn’t find the right KB article, interpret feature names, or explain how a product actually helped a customer without switching multiple tabs and going over the marketing page.

Freshly trained agents couldn’t find the right KB article, interpret feature names, or explain how a product actually helped a customer without switching multiple tabs and going over the marketing page.

Insights

Content needed consolidation

I observed agents jumping between multiple open tabs because articles were written by multiple authors over months/years.


Quicker access to answers in the tool

The tool only showed them what SKU the customer was on.

  • No brief on features

  • No link to relevant article

Marketing jargon

Customers didn't care that the software would handle 1 million threats. They wanted to know how it would impact them.

Missing visuals and no simulators

With frequently changing UI and no simulation tool, agents had to guess what the customers were seeing.

Senior Agents Had Hacks. Juniors Had Gaps.

This contrast revealed a deeper issue: Knowledge wasn’t organised.

Agents were building their own systems because ours weren’t working.
If we wanted consistency and faster support, the fix couldn’t be more tabs or tools. It had to be built into the flow.

Every feature/topic had mutliple articles associated with it

What they currently trained with:

Handwritten posters

A cheatsheet - to print + digital

FAQ style prep sheet

Final Solution

Meet the users where they are

Meet the users where they are

Instead of creating yet another resource (like a flashcard) for the agent to handle, I saw an opportunity within the tool itself.

What could the pop-up modal look like?

Ultimately the easiest means for Engg to implement the change was a stripped down version

I proposed embedding feature summaries directly within the internal tool’s "View Assets" section.
Each product feature now:

  • Explained what it does, in human terms

  • Linked to the most relevant (and updated) KB article

  • Reduced the need to switch tabs or guess

A small change, but a significant impact.

Quick wins vs stretch goals

Quick wins vs stretch goals

Understanding that several improvements could be made but not all of it could be implemented

Understanding that several improvements could be made but not all of it could be implemented

While the proposed changes could be a quick win, the internal tool required Dev resources

Prototyped, living in JIRA

Prototyped, living in JIRA

If we’d used Lovable onsite, this idea could’ve been tested the same-day with agents.
Instead, I documented everything and filed it into JIRA the following week—alongside a request to consolidate our KBs and enable agent comments directly under articles.

If we’d used Lovable onsite, this idea could’ve been tested the same-day with agents.
Instead, I documented everything and filed it into JIRA the following week—alongside a request to consolidate our KBs and enable agent comments directly under articles.

Challenges

Challenges

Temporary solution for a tool that will be sunsetted

Temporary solution for a tool that will be sunsetted

Built to Bridge a Gap

Built to Bridge a Gap

Salesforce AI is being integrated, and soon agents will be able to just ask for answers.
But this interim solution gave them clarity, confidence, and faster workflows—when they needed it most.

Salesforce AI is being integrated, and soon agents will be able to just ask for answers.
But this interim solution gave them clarity, confidence, and faster workflows—when they needed it most.

Outcomes

Outcomes

Almost Shipped

Almost Shipped

The AI integration work took priority

The AI integration work took priority

On an impact effort matrix, the CST improvements would qualify as a quick win unfortunately the Dev team didn't have the bandwidth.

On an impact effort matrix, the CST improvements would qualify as a quick win unfortunately the Dev team didn't have the bandwidth.

The project currently lives in JIRA

The project currently lives in JIRA

The Engg. team clarified that the lift was fairly straightforward but due to the entire teams focus being redirected to AI.

The Engg. team clarified that the lift was fairly straightforward but due to the entire teams focus being redirected to AI.

What I'd do differently if I were to repeat the project

What I'd do differently if I were to repeat the project

Block a day to test solutions on site

Block a day to test solutions on site

With how easy Lovable + ChatGPT made it to mock up a solution, it should have been validated on site

With how easy Lovable + ChatGPT made it to mock up a solution, it should have been validated on site

Work with KB to immediately address gaps in knowledge

Work with KB to immediately address gaps in knowledge

The KB team wasn't able to come with us on site and it would have helped them to understand how our agents were perusing content compared to our users.

The KB team wasn't able to come with us on site and it would have helped them to understand how our agents were perusing content compared to our users.

nashok.design@gmail.com

Open to opportunities!

Reach out for kuchen mit conversation

nashok.design@gmail.com

Open to opportunities!

Reach out for kuchen mit conversation