Scale is a navigation problem.
Group, filter, collapse, and aggregate lines. Reveal detail when the seller needs it.
ZUORA CPQENTERPRISE SAAS · B2B · AI PRODUCT DESIGN
Making complex quoting feel clear.
Leading design for an AI-powered quoting experience that helps sales teams build, review and refine complex deals with confidence.
↓
30–50%
faster first-pass quote creation
25–40%
faster large-quote review
40%+
of advanced sessions used AI workflows
01 / THE CONTEXT
The quote shouldn’t
be harder than
the deal.
Enterprise quoting is a decision-making problem as much as a data-entry problem.
Sellers work across hundreds or thousands of quote lines, subscription changes, multiple agreements, locations, and billing accounts. When all that complexity is exposed at once, deals become difficult to scan, review, and maintain.
I led design direction and execution across the core quoting flows, helping sellers see the business impact of their actions and use AI without losing context.
MY ROLE
Lead Product Designer
Sole product designer, from direction to execution
THE TEAM
PM, engineering lead, 3 engineers
Cross-functional product and engineering team
TIME & STATUS
14 months
Shipped as Phase 1
THE CHALLENGEMake complex deals understandable,
Make complex deals understandable,
actionable, and accurate.
01 / SCALE
Too much to scan
Large line sets and dense configuration fields made important changes difficult to isolate.
02 / TIME
Context was hard to follow
Agreement terms and future-dated changes added another dimension to every quoting decision.
03 / CONFIDENCE
Errors surfaced too late
Fragile save behaviour and incomplete paths reduced trust. Spreadsheet workarounds drifted from catalog data and business rules.
02 / DISCOVERY
Find the friction.
Give it structure.
The strongest patterns came from real deal work.
I reviewed recorded customer feedback, support themes, and internal sales pain points using Glean, then interviewed Zuora sales reps and customer sales reps to validate assumptions and prioritise the first release.
Not every field deserves equal weight.
Keep primary information on the main row, with expandable secondary detail and flexible visible fields.
Time needs to be visible.
Use period selection and agreement filters to focus review on the relevant changes.
AI belongs beside the work.
Help sellers generate and shape quotes in context, with changes they can see and review.
03 / DESIGN DECISIONS
Test the structure.
Adapt to real deal work.
I began with a single-product quote, then introduced bundles, pricing adjustments, approvals, multiple contracts, and high-attribute configuration to understand where each interaction model held up.
Moved forward: clearer stages supported complex products, validation, approvals, and contextual help.
Better suited to simpler catalogs: immediate feedback was useful, but dense attributes and policy requirements were harder to scale.
WHY THE STEPPER MOVED FORWARD
Give complex configuration a clear sequence.
Product selection, configuration, pricing, and review became distinct stages. A persistent sidebar kept the quote total in view. This offered stronger guidance for complex products and clearer places for validation, approvals, and contextual assistance. The split-view concept stayed relevant for simpler catalogs, where immediate editing mattered more than configuration depth.
Fast selection met dense configuration.
The first prototype tested adding products and assigning their attributes. Starting with the simplest quote helped establish the core flow before introducing the demands of more complex products.
Concept A: adding products and moving through the guided quote flow.
WHAT TESTING REVEALED
Selection was intuitive; attributes were harder.
Users found product selection fast and intuitive. Products with 50 or more configuration fields made manual attribute assignment difficult.
DESIGN RESPONSE
Support density without overwhelming sellers.
This exposed a need for clearer configuration hierarchy and led to the later exploration of AI-assisted configuration, with sellers able to review and adjust the setup.
A useful target did not make the flow universal.
Goal seek let sellers work backward from a target quote value. Testing also examined pricing insights and approval guidance within that workflow.
Early goal-seek prototype: target-based pricing, insights, and approval guidance.
WHAT TESTING REVEALED
Value varied between sellers.
Some users valued working backward from a target; others saw complexity they were unlikely to use. Pricing insights and approval guidance were useful, but users wanted them throughout the workflow.
DESIGN RESPONSE
Make guidance persistent and navigation flexible.
Signals needed to remain available beyond goal seek. Testing also revealed that adding products after reaching pricing was too difficult, pushing the design toward more flexible navigation.
Confidence mattered more than extra steps.
Pricing guidance helped reps estimate a quote around a prospect’s needs, including during early discovery conversations.
Pricing-guidance prototype used to explore early quote estimation.
WHAT TESTING REVEALED
Experience changed the need for support.
Junior and newer sales reps found guidance especially valuable during early calls. Experienced reps relied more on their own pricing intuition.
DESIGN RESPONSE
Help newer reps without slowing experts.
The design direction kept guidance available in context, supporting confidence while avoiding a mandatory detour for sellers who already knew what to do.
WHAT THESE ROUNDS CHANGED
Keep the structure. Make support contextual.
Testing refined the initial stepper into a more flexible quote-building direction: clearer information hierarchy, pricing and approval signals available throughout, and assistance for the configuration work that stayed difficult.
04 / THE EXPERIENCE
Bring the quote together.
Keep the seller in control.
Phase 1 prioritised guided quote creation, clearer large-quote review, pricing and approval visibility, AI-assisted generation and configuration, and guardrails for complex or incomplete paths.
A walkthrough of the CPQ experience, showing how the quoting workflow comes together.
Start with a quote. Refine it with judgment.
AI generation gave sellers a first draft they could inspect and continue shaping manually. The design treated AI as an entry point into the workflow.
AI-generated quote: create an initial quote, review it, and continue refining.
WHAT TESTING REVEALED
A starting point was valuable.
Users responded positively to generating an initial quote with AI and then continuing to refine it manually.
DESIGN RESPONSE
Keep the output editable.
Preserve seller judgment by making the generated quote reviewable and open to manual changes.
Translate needs into a reviewable setup.
High-attribute products remained difficult to configure by hand. AI assistance addressed the effort of turning customer needs into product settings.
AI-assisted product configuration, following the high-attribute problems found in testing.
WHAT TESTING REVEALED
The dense-field problem needed more assistance.
After manual testing exposed difficulty with high-attribute products, users found AI-assisted configuration valuable for reducing setup effort.
DESIGN RESPONSE
Assist the translation; preserve review.
Let sellers inspect the resulting configuration and adjust the final settings before proceeding.
BEYOND THE FIRST RELEASE
Explore assistance beside the quote.
These agent prototypes explored contextual recommendations and conversational goal seek. They are distinct from the stated Phase 1 release scope.
Suggest products with a reason.
The agent used quote context, customer details, existing products, and pricing rules to suggest add-ons, upsell opportunities, and bundle-completion products within the workflow.
AI agent exploration: contextual product recommendations.
WHAT TESTING REVEALED
Context was more useful than generic search.
Users valued relevant suggestions, but expected clear rationale and compliance with pricing rules.
DESIGN RESPONSE
Make recommendations explainable.
Show why a product fits the deal so sellers can evaluate suggestions and decide what to add.
Make the target easier to express.
The earlier goal-seek flow was valuable to some sellers and too complex for others. This exploration moved the interaction into the Zuora AI agent, where a seller could request a target total contract value using pre-approved discounts.
AI agent exploration: target-value pricing through conversational goal seek.
WHAT TESTING REVEALED
The conversational approach was easier to follow.
Users found it easier to understand and faster to use when testing pricing scenarios.
DESIGN RESPONSE
Make every adjustment transparent.
Keep proposed changes reviewable and within approved discounting rules, with the seller deciding how to proceed.
05 / IMPACT & REFLECTION
Less effort.
More confidence.
Early results since launch show improvements across quote construction, review, and readiness.
Faster quote creation
First-pass quote creation speed.
Faster quote review
Review efficiency for large quotes.
Better quote quality
Improvement in accuracy and policy compliance.
Faster approval readiness
Less time getting quotes ready for approval.
Higher rep confidence
Self-reported confidence on complex deals.
AI workflow adoption
Share of advanced quote sessions using AI assistance.
WHAT I TOOK FORWARD
The goal was more than reducing clicks.
It was making dense, high-stakes commercial decisions feel understandable, controllable, and supported.
Research earlier
Partner with dedicated UX research earlier to explore the edge cases that shape enterprise quoting.
Make tradeoffs durable
Document decision rationale more consistently as stakeholder priorities change.
Extend contextual AI
Explore quote review, pricing recommendations, and error prevention as future opportunities.
LET’S WORK TOGETHER
Good conversations.
New possibilities.
I’m always open to interesting conversations—about design or what comes next.