Plutofy

Redesigning an AI sales training tool to create more business and user value.

Web App

Web App

Conversational AI

Generative AI

Generative AI

Startup

Education

B2B

B2B

Product Design

TEAM

2 Engineers, Marketing

ROLE

Product Designer

DURATION

July - October 2024

(3.5 months)

Context

Expanding the scope from UI to UX

The founder initially hired me to refresh the UI of the AI sales training tool that simulated real life scenarios and provided feedback. However:

  • Current features brought little user value because it did not target problems unique to our demographic.

I built trust with the founder and expanded the scope to encompass both UX design and product strategy.

Old UI

Problem Identification

Finding better product opportunities to target our audience

While Plutofy was built for individual freelancers, small businesses, and casual learners; The current product did not cater to this user group. I was able to shift the founder's mindset to better align with our users’ needs.

Key Findings:

  1. Currently, the product merely copies the competitor’s demos.

  2. Our audience is inherently different and requires different approaches.

  3. Competitors assume sales experience and familiarity with industry techniques, all luxuries our users do not have. 

  4. Through interviewing, I discovered that general 'sales training’ is unhelpful for most users.

Ideation

Adding in setup for custom simulations, instead of theoretical ones.

Through interviews, I discovered that each niche in which people sell services or goods has unique challenges, making general "sales training" ineffective.

To maximize product effectiveness, users need custom simulations to practice upcoming important sales calls before they happen.

Allow users to upload relevant documents to auto-create scenarios.

Customizable personality, objection level, and type.

Ideation

Pitching a new value proposition of real-time coaching

To mimic the feedback and support you would get from an experienced sales mentor, I pitched 3 different ways to simulate real-time coaching:

Concept #1

An AI mentor character in the call to guide direction and correct mistakes real time

Concept #2

Real-time tips that pop up when users make mistakes during calls.

Concept #3

Give users suggested approaches when they fail to complete certain objectives.

Challenges

Reducing API Costs

With limited data, the tips required near-constant API calls, making it prohibitively expensive. To reduce costs we came up with a 2-pronged approach:

Preset Tips

Gave general guidance depending on the phase of the call. (Easy to train and deploy)

Real-time Tips

More specific but costly tips (Given only when users press the tip button to limit cost)

Product Strategy

Training the Model

I identified the need for better data to train our model after discussing the product’s technical limitations with the founder. To address this, I proposed two data collection strategies within Plutofy:

Call Collection Feature:

We offered to score users' real sales calls for free, allowing us to gather valuable training data.

Tip Rating System

Users rated real-time and sales report tips, helping us refine suggestions based on usefulness.

Data Collection Feedback Loop Flow Chart

Prototype & Testing

The concept resonated but most users felt overwhelmed by live tips

The concept tested well, however all participants had one complaint…

Key Findings:

  1. 5/5 users felt overwhelmed with information when both tips and the call happened simultaneously

  2. Users DON'T want to be interrupted during the simulated calls.

  3. Generic preset call scenarios are not helpful for experienced users since they are only interested in improving in their selling niche.

5/5 users felt overwhelmed with information when both tips and the call happened simultaneously

Insights from Testing with Users

After synthesizing the key findings from testing. I had 3 realizations.

01

Novices and experienced users had completely different needs.

Novices and experienced users had completely different needs.

Novices and experienced users had completely different needs.

02

Live Tips overwhelmed users.

Live Tips overwhelmed users.

Live Tips overwhelmed users.

03

Generic preset call scenarios offered no real value to users.

Generic preset call scenarios offered no real value to users.

Generic preset call scenarios offered no real value to users.

Product Strategy

Solving Insight #1: Doubling Down on Novices

Since novices and experienced users had such different needs, we had to prioritize our product towards one group.

Given that the majority on our waitlist was novices and no other beginner trainer tools existed in the market, I convinced the founder to focus even more on novice salespeople.

Solution

Solving Insight #2: How to Train Novice Users without Overwhelming them

After our users firmly rejected the idea of ‘real-time tips’, I pitched the new 'learning flow'. A seamless experience that guides novice users step by step through the sales process to get them ready for a full simulated call.

Solution

Solving Insight #3: Customizing Product Experience Based on User's Level and Niche

The founder agreed to prioritize novice users, but he also wanted to find a way to repurpose the existing custom simulations to suit their needs.

As I explored this challenge, I came up with a simple yet effective solution: tailor the product experience to the user's skill level and sales niche.

Challenges

Connecting discrete parts of the product.

After designing the lesson flow and custom simulations, the product felt a bit separated. To bridge the gap between both parts of the product and introduce everything the product has to offer to both user groups. I added interactions for experienced users to benefit from the lessons when needed.

  1. Prompt users to revisit specific lessons when associated mistakes are made

  1. Introduce users to create custom simulations as part of the lesson curriculum

Conclusion

5/5

Final solutions tested well with users.

100%

Founder loved the shifted direction and proposed changes.

Next Steps

Product under development, planned launch during Q1 2025.

© Bryan Huang 2024

© Bryan Huang 2024