AmpUp · Product Designer · July 2025 · B2B SaaS · Enterprise Sales Enablement Platform
AI Sales Coaching Platform
I designed an AI-powered B2B sales coaching platform that scales top-performer knowledge across enterprise teams.
Challenge
When 95% of sales coaching moments disappear and new reps take 6 months to ramp, enterprise sales organizations lose millions in revenue. I needed to design a system that captured what top performers do and made that knowledge useful without replacing sales judgment.
+34%
conversion-rate lift for pilot teams
6x
faster ramp time for new reps
100%
coaching coverage across every rep
10+ hrs
saved per manager each week
Illustrative artifacts
A product story with tangible surfaces.
These panels explain the design decisions around the supplied project surfaces.
Research synthesis
Illustrative portfolio artifact based on the case-study narrative, not a direct client screenshot.
Three coaching moments
Illustrative portfolio artifact based on the case-study narrative, not a direct client screenshot.
AI recommendation model
Before
Generic advice
After
Team-specific coaching with source transparency
Illustrative portfolio artifact based on the case-study narrative, not a direct client screenshot.
Product surfaces
Real interface artifacts from the work.
Project artifact supplied by Savita for this case-study narrative.
01 · Context & Stakes
I designed AI coaching for enterprise revenue teams.
I worked as the Product Designer in July 2025 on a B2B SaaS sales enablement platform. I partnered with a Product Manager, ML engineers, sales enablement consultants, frontend engineers, and enterprise sales leaders to shape the product from research through pilot validation.
Owned research synthesis, product flows, prototypes, usability testing, and handoff.
Designed for sales reps, managers, and leaders who needed coaching to scale without adding more manual review time.
Kept the AI experience transparent so users could understand the recommendation and still make their own call.
02 · Constraints & Risks
The hardest choices were about trust.
We deliberately cut features that looked attractive on paper but created risk in real enterprise teams. The goal was not to make AI louder. It was to make coaching easier to trust.
We did not build live call monitoring because reps described it as invasive and compliance varied by industry.
We skipped a generic sales training library because clients already had LMS tools and needed team-specific playbooks.
We avoided public leaderboards because research showed they could create toxic competition.
We deferred CRM integration to Phase 2 so the MVP could validate core coaching value first.
03 · Research & Discovery
The coaching problem was expensive, but mostly invisible.
I started by interviewing 15 stakeholders across 3 enterprise client organizations. The pattern was clear: managers wanted to coach more, top reps had valuable habits they could not easily explain, and new reps were left to learn slowly through trial and error.
I spoke with 8 sales managers, 5 top-performing reps, and 2 newly ramped reps.
Managers reviewed only 5% of calls because they simply did not have the time.
New reps took 6+ months to hit quota, while top performers reached strong performance in 3 to 4 weeks.
One manager said Sarah closed deals differently, but those strategies disappeared the moment each call ended.
04 · Key Decisions & Trade-offs
We chose three moments that fit the sales workflow.
I ran a 2-day workshop with Product, Engineering, and client stakeholders. We mapped the sales workflow and agreed that the product should support reps before calls, during practice, and after calls instead of becoming another dashboard they had to remember to open.
Pre-Call Intelligence gave reps company research, prospect insights, and personalized talking points.
Practice Partner created a safe place to rehearse objections using realistic client scenarios.
Post-Call Analysis reviewed calls with actionable coaching and top-performer comparisons.
We aligned on a 6-month delivery plan and 3 design partner clients committed to the pilot.
05 · Design Solutions
Every screen had to create value in 30 seconds.
I explored 20+ concepts across the three coaching flows. The first direction showed a lot of insight in one dashboard, but client feedback changed the direction quickly: sales reps live in calendars and CRMs, not analytics screens. Time-tracking with client users showed reps spent about 8 minutes preparing for calls. That became the design constraint: every screen needed to be useful in under 30 seconds.
The first concept showed every insight in one place, but it felt disconnected from the sales day.
Client feedback pushed the design toward a calendar-first model with one-click coaching access.
Sales trainers helped align the coaching model with MEDDIC and Challenger Sale methods.
The product framing changed from AI telling reps what to do to AI showing what top performers do.
Pre-call screens used a clear hierarchy: company context, prospect insights, then talking points.
Practice flows used real objections from recent lost deals instead of generic roleplay prompts.
Post-call analysis compared reps with their own team's top performers, not a generic benchmark.
The experience worked on mobile so reps could prep between client meetings.
06 · Testing & Iteration
Specific, transparent AI changed user trust.
In moderated testing with 6 sales reps, the first version failed in familiar AI ways. The advice felt generic, the source of recommendations was unclear, and Practice Partner felt too much like standard training.
Generic advice like ask more questions became team-specific coaching about budget timing and approval process questions.
Recommendations added context such as sample size and source calls from top performers.
Trust scores increased from 42% to 87% after the transparency layer was added.
Practice Partner weekly usage increased from 12% to 68% after scenarios used real objections from client deals.
07 · Outcomes & Impact
The platform made coaching coverage practical at scale.
After 3 months with client pilot teams, users could find key insights quickly and managers had more time for strategic coaching instead of manual call review.
Conversion rate increased by 34% across pilot teams.
New-hire ramp time became 6x faster, moving from months to weeks.
Every rep received personalized feedback instead of only the 5% of calls managers could review manually.
Sales cycles became 35% shorter and managers saved 10+ hours each week.
08 · Reflection
The breakthrough was the learning loop, not just the AI.
This project taught me that AI products earn trust through context. The platform worked because it learned each client team's playbook and explained why a recommendation mattered.
Context beats generic advice when every client team sells differently.
Transparency matters more than polish when users are deciding whether to trust AI.
Time pressure can improve clarity because every screen has to earn attention quickly.
If I restarted, I would involve client sales reps in weekly design sprints from Week 1 instead of Week 5.
Tools and learnings
Built through craft, constraints, and handoff.
Tools
Outcomes
Conversion rate increased by 34%
Ramp time became 6x faster for new reps
Coaching coverage increased to 100%
Sales cycles became 35% shorter