Gigi Kuo Back to work

Webex Contact Center — Unified Flow Designer

Two flow builders. One flow designer.

Webex sold two separate tools for building call and messaging flows. Customers couldn't tell which one to buy, and internally, two teams were building overlapping solutions. I led the design of the single, unified Flow Designer that resolved the conflict and replaced both legacy tools.

The unified Flow Designer in light mode: a node palette on the left and a contact center flow on the canvas, with menu, queue and error-handling nodes connected by color-coded paths

Timeline

2025

Role

Product Designer
Contact Center, Webex (Cisco)

Owned the future-state experience, core workflows and interaction patterns. Product Management and leadership owned customer migration.

Collaborators

Sales, Design System, Product Management, Engineering, Data, UX Research, UX Writing, and Leadership — 30+ people across the project

Summary

Webex Contact Center's voice-only Flow Builder and the omnichannel Webex Connect were two separate products solving the same problem. I defined the future-state experience for a single Flow Designer: the core workflows and interaction patterns that kept voice specialists' muscle memory intact while opening the door to full omnichannel workflows. In a 2-week prototype study with 20 enterprise admins, task completion time dropped 27% compared with the legacy dual-tool workflow, satisfaction scores rose 38%, and 18 of 20 admins completed setup unassisted on their first try.

What do you sell a customer who can't tell which product to buy?

A cloud contact center helps businesses manage inbound and outbound communications across voice, chat, SMS, email, and social. Inside Webex, the piece that built those workflows had a split personality.

Grown In-House, Bought In

The voice-only Flow Builder was native to Webex Contact Center. Webex Connect arrived through the acquisition of a smaller company — different codebase, different design heritage, same job to do.

No Shared Language

Different node styling, different connection logic, different panel structure. Learning one taught you nothing about the other.

An Unanswerable Sales Question

A prospect moving off an on-premise system had to pick one. Nobody — including us — had a clean answer for which.

Webex Connect's flow canvas showing a prescription-refill workflow with Epic, HTTP request, and SMS nodes connected in a straight line

Webex Connect

Webex Contact Center's Flow Designer canvas showing a main flow with queue contact, condition, and error-handling nodes in a different visual style

Webex Contact Center Flow Designer

Two products built to do the same job, with no shared visual language, node system, or interaction model.

The question we kept hearing

"Which one do I actually need?"

Every sales conversation hit the same wall — and every time, the honest answer was that it depended on a product distinction customers shouldn't have had to know existed.

Business goals

01

Unify the flow creation experience

Create a single product to eliminate duplication and streamline operations.

02

Simplify flow creation

Let both technical and non-technical users design customer journeys quickly and intuitively.

03

Boost efficiency & innovation

Let teams build smarter workflows — from voice to omnichannel, chat, email, SMS, and integrations.

The design challenge

How do we build a single, intuitive product that serves both voice builders and omnichannel builders—without compromising the experience for either?

Research & Discovery

Mapping the UX gaps

I put both existing canvases side by side to see exactly where they diverged — node styling, connection logic, panel structure, error handling — and used that gap map to scope what a unified product would actually need to solve.

Research comparison table of Voice Flow Builder and Webex Connect across flexibility, variable management, error tracing, learnability and flow readability, with strong preferences flagged

What each product got right

Research with participants and sales partners surfaced a clear, consistent pattern: each tool had real strengths worth protecting, not replacing.

Voice Flow Builder — strong preference

  • ✓More intuitive and easier to learn
  • ✓Non-intersecting connection lines made flows easier to read
  • ✓Sub-flows were more intuitive to manage

Webex Connect — strong preference

  • ✓Custom nodes and JavaScript made integrating with any tool possible
  • ✓Variable categorization made large flows easier to navigate
  • ✓Clear display of a node's connected variables

The mandate this set for the unified product: keep Voice Flow Builder's learnability and readability as the visual foundation, and build in Webex Connect's flexibility rather than sacrificing it.

The trade-off

One foundation meant one group had to adapt. Webex Connect builders moved to a canvas that wasn't theirs. To make that worth it, I kept what they valued most (custom nodes, JavaScript and variable categorization), and with engineering and product we pushed analytics and debug tooling to later releases, so the shared canvas could ship first instead of waiting for full parity. Moving existing Connect customers onto the new product was owned by Product Management and leadership; my job was to make the destination worth moving to.

Competitive analysis & task prioritization

I assessed industry-leading flow design tools to identify best practices, feature gaps, and opportunities for differentiation — focused on usability, scalability, and cross-platform integration. From there, I worked with engineering and product to identify and prioritize the build: canvas, nodes, and node connections first; variable configuration and sub-flows next; analytics and debug tooling after that.

A competing flow builder's canvas with a trigger node, message nodes and a widget library panel
Another competing tool's disconnect flow sample, with entry point, prompt, wait, queue and transfer nodes

Competitor flow tools reviewed for usability, scalability, and cross-platform integration.

Prioritized build backlog listing canvas, nodes, node connections and activity palette as first priority, with variables, sub-flows, analytics and debug view in later phases

The prioritized build order, agreed with engineering and product.

Node & Activity Design

Nodes are the building blocks of a flow designer — every menu, condition, and integration is one. Getting their design right meant fixing four specific, recurring pain points in the legacy tools.

Users had to click into nested panels or tiny icons to see what a node did — breaking flow comprehension.

Fonts and spacing made labels and inline logic hard to read and scan.

Error paths weren't visually distinct, and feedback was hidden and ambiguous.

Node styles and interactions were inconsistent across legacy and internal tools.

What changed

Before: a bare icon-only node labeled Rx Triggering System, with a separate small panel showing a condition's True/False branches and hidden error handling

BeforeConfiguration hidden, error paths buried

After: a redesigned SMS Consent condition-block node showing its title, its Yes and Undefined-errors outcomes, and error handling inline on the node face

AfterOutcomes and errors visible without a click

  • ✓Improved scannability
  • ✓Inline configuration clarity
  • ✓Error visibility
  • ✓One consistent visual framework

Built with the design system team, not just for it

Every node type, connector, and port state was built in collaboration with the Design System team as reusable, scalable components — start and event nodes, page connectors, and a shared success/error/timeout port language — so the pattern would hold up as the flow designer grew.

Design system reference sheet showing standardized start and event node types, page connectors, and success, error, and timeout port states, next to the preliminary canvas concept

Design process

A research-through-launch loop, with a tight validate/design/prototype cycle repeated as fast as feedback came in.

Research Define Design ⇄ Validate Prototype Build QA test Launch

UX for Configuring Logic

The real power of a flow designer comes from what happens inside each node — so how a user configures that logic mattered as much as how the node looked on the canvas. I prototyped three approaches and tested them head to head.

The shipped Flow Designer with the node palette on the left and the MainMenu configuration panel open on the right, canvas still visible

The chosen layout: node palette, canvas and configuration panel visible at once.

Pop-up modal

A centered overlay opens when a node is selected, focusing the user on editing its configuration.

Chosen

Side configuring panel

A toggleable panel on the right lets users configure the selected node while keeping the canvas and tool panel visible.

Replaces tool panel

The side panel swaps from the node library to that node's configuration options in the same space.

Why the side panel won

Users valued having simultaneous access to both the tool panel and the configuration panel more than any single-focus layout. Two reasons kept coming up in testing:

  • Contextual awareness — they could see the flow and the selected node's configuration in one view.
  • Familiarity — the side-panel pattern felt intuitive and consistent with other design tools they already used.

90%

Unassisted first-time setup success with the side configuring panel (18 of 20 admins) — a significant efficiency advantage over the other two options across every interaction metric tested.

Impact

One product, built on the strengths of both — instead of a compromise between them.

For Voice Builders

Kept the readability and non-crossing connections they already knew — nothing to relearn.

For Omnichannel Builders

Gained the same custom nodes, JavaScript, and variable categorization Webex Connect always had.

For the Business

One product to sell and support instead of two, and 27% less time per task than the legacy dual-tool workflow in testing.

For What's Next

A shared node system built with the design system team — new channels can plug straight in.

Before: Two products

  • ✗Customers couldn't tell Voice Flow Builder and Webex Connect apart
  • ✗No shared node system, visual language, or interaction model
  • ✗Configuration and errors hidden behind nested panels
  • ✗Duplicated engineering effort across two flow-building products

After: One flow designer

  • ✓One product for voice and omnichannel flows alike
  • ✓A shared, scalable node and connector system with the design system team
  • ✓Configuration and error states visible inline, without a click
  • ✓Light and dark themes, defaulting to the user's browser preference

Validated in a 2-week prototype study with 20 enterprise admins

27%

Less time to complete tasks, vs. the legacy dual-tool workflow

38%

Higher user satisfaction score (post-task)

90%

Unassisted first-time setup success (18 of 20)

All three figures come from the same 2-week prototype study with 20 enterprise admins.

Reflections

Merging isn't averaging

A blend of two tools would have been a third thing for everyone to learn. Picking one foundation and layering the other's strengths on top meant voice builders had nothing to relearn.

Test layouts, not opinions

All three configuration layouts looked reasonable in review. Putting them head to head settled it faster than another meeting could have.

Build the system with the system team

Co-designing nodes and ports with the Design System team made them shared components from day one, so the next channel plugs in without inventing new patterns.

You may also like