Maximizing Sales Productivity with AI/ML Prioritization
As Lead Product Designer, I envisioned transforming Google's fragmented ads seller experience into a unified "single pane of glass". Taking over a stalled P0 project with no PM, I partnered with Engineering to build the CRM from scratch, delivering an AI/ML-powered alpha that made the Portfolio dashboard the central hub for managing customer relationships, campaigns, and revenue targets each quarter.
Goals
Make Portfolio dashboard the single starting point where all seller workflows and team insights converge
Replace fragmented tool-hopping with seamless workflow management
Automate routine processes so sellers can focus on building customer relationships
Role
Lead Product Designer
Scope
End-to-End Product Design, Product Strategy, Implementation Support, Cross-functional Leadership
Collaborators
Product, Business, Research, Content Design, Accessibility, and Engineering
Timeline
Q1 2023 - Q3 2023
BACKGROUND
What Drove the Redesign
Connect Sales is Google's internal CRM for Ads sellers. The platform provided one-size-fits-all account views for sellers managing large books of business (500-600 customers per quarter) across distinct workflow phases. This mismatch forced manual prioritization across fragmented tools, hurting both seller performance and business revenue.
Business Need
Prevent lost ad revenue from sellers missing high-value opportunities due to fragmented prioritization.
User Need
Enable sellers to focus on building customer relationships rather than manually sorting through hundreds of accounts.
PROBLEM
The challenge
Fragmented account prioritization across multiple tools makes it hard for sellers to focus on high-impact customers and move quickly, directly impacting their ability to achieve target's and Google's revenue outcomes.
RESEARCH
Understanding Seller Workflows
I approached this as a workflow design challenge, not just a prioritization problem. My research with 15 sellers revealed they weren't just managing accounts – they were navigating three distinct phases (outreach, prep, and performance optimization) that each demanded different information. This insight fundamentally shifted how I designed the information architecture, moving from static lists to dynamic, workflow-aware interfaces.
Mental model mismatch
Sellers think in phases, system thinks in lists.
Context matters
Same account needs different info per phase
Scattered signals
Critical data spread across 5+ tools
DESIGN CHALLENGE
How might we
Design a smart prioritization experience that adapts to sellers' dynamic workflows so they can focus more time on building customer relationships?
STRATEGY
Strategic Approach
My research findings converged on a focused direction: smart prioritization that adapts to seller workflows for sellers involving managing large books of business across dynamic phases. Three core principles guided my approach:
Cut to the chase
Surface high-impact accounts with relevant workflow context
See the finish line
Make quarterly targets and progress visible to sellers
Skip the grunt work
Automate prioritization so sellers focus on relationships
IDEATION
Mapping Seller Workflows
To design an effective prioritization experience, I mapped how sellers navigate their daily work when managing large account portfolios. This workflow showed how sellers move between performance assessment, account prioritization, and taking action throughout their day.
Design Explorations
Based on the workflow mapping, I explored different interface approaches for key workflow stages. I took a progressive approach to smart prioritization, starting with research-based heuristics that would evolve into ML-driven intelligence as we collected user data.
Smart Priorities
Transforms static account lists into intelligent, ranked prioritization by using heuristic-based signals that dynamically rank accounts with transparent rationale, enabling user feedback collection to inform future ML development while adapting to workflow needs.
Contextual Flyout
Provides detailed account context and prioritization rationale without disrupting workflow, enabling sellers to understand why accounts are prioritized while maintaining their position in the main list.
Time-sensitive Insights
Alerts sellers to critical, time-bound opportunities and risks that demand immediate action using threshold-based heuristics to surface only the most urgent signals that could significantly impact performance if missed.
Performance Summary
Provides an at-a-glance progress dashboard of seller goals across key metrics: points, UAAs, and revenue, enabling sellers to quickly identify performance gaps, maintain momentum, and adjust their strategy accordingly.
USER TESTING
Alpha Launch Validation
The alpha with 50 GCS sellers across EMEA and APAC validated that intelligent prioritization improved workflow efficiency but revealed control and guidance gaps. Feedback led to enhanced alert management and planning for integrated action flows to build user confidence and workflow continuity.
Alert Management Controls
INSIGHT: Sellers wanted better threshold control to avoid alert fatigue. ACTION: Implemented snooze/dismiss controls for better alert management.
Actionable Guidance Integration
INSIGHT: Users expected recommended next steps with insights. ACTION: Planned guidance integration for future releases due to engineering constraints.
FINAL DESIGN
The MVP
The alpha established intelligent prioritization as a viable solution for high-volume seller workflows while revealing enhancement opportunities. User feedback drove immediate alert management improvements and identified requirements for integrated action flows, creating a clear beta roadmap and capturing the behavioral data necessary to evolve toward ML-driven recommendations.
IMPACT
The Results
The alpha validated the workflow-aware approach through strong engagement and seller adoption, establishing Connect Sales as the reliable planning tool for high-volume sellers while laying the foundation for future AI/ML development. The designs for the product vision were also presented at International Ads summits in Dublin, Tokyo, and Singapore to align global sellers and leadership.
24%
Increase in WAU vs. pre-alpha baseline
28
Avg. monthly interactions per seller with prioritized accounts
80%
Adoption replacing spreadsheet workflows
71%
CSAT indicating early product-market fit and seller confidence
LEARNINGS
What I learned
ML maturity impacts user trust
The underlying models were still maturing, meaning some signals lacked depth. This affected seller confidence and highlighted the importance of aligning design ambition with technical capability.
Explainability builds trust in AI
Some of our most intelligent prioritization signals were difficult to unpack for users. Designing for transparency in how suggestions are generated is as critical as the intelligence itself.
Power users need customization flexibility
We defaulted to a one-size-fits-most model, but power users wanted more control over how they filtered or weighed signals. Future iterations should explore more flexible personalization.
Cross-functional success requires early alignment
PM churn taught me that UX must sometimes fill leadership gaps, and earlier collaboration with enablement teams would have improved adoption despite organizational instability.