How an AI-powered CRM Increased User Satisfaction and Retention

In this case study, we’re going behind the scenes to take a look at how improving specific features of this CRM powered by AI increased user satisfaction and retention.

Overview

This project is about an AI-powered CRM (Customer Relationship Management) that integrates messages from Meta platforms like WhatsApp, Instagram, and Facebook into a single platform where users can manage conversations, automate tasks, get data-driven insights, and simplify their sales process.

The team

Business Analyst
UX/UI Designer
Frontend Developer
Backend Developer

My role

UX/UI Designer

My contribution

Conducted discovery and evaluative research.
Facilitated affinity and empathy mapping workshops.
Produced data-driven personas and user journey maps.
Created wireframes, UI designs, and interactive prototypes.

Problem Statement

Despite strong adoption in its first year, the AI-powered CRM platform started to experience less user activity, resulting in a 28% decline in monthly subscription renewals.

Objectives

Increase user retention by reducing churn and achieving at least a 15% lift in subscription renewals in 6 months.

Increase feature adoption by ensuring 80% of users engage with the top 3 core features in 6 months.

Improve user satisfaction by raising NPS/CSAT scores from 70 to 85 in 6 months.

Problem-solving approach

Our process combined discovery research and team alignment activities to map user pain points, prototype potential solutions, and validate them with a sample of churned users.

Discovery research

The first step was to create a survey for users who were actively using the platform, paying a month by month subscription and then cancelled.

The goal of this survey was to identify causes of disengagement and uncover primary drivers of cancellation like price, usability issues, missing features, poor support, or lack of perceived value.

This is the survey the users saw at the end of the cancellation flow, and this same survey was sent by email one week later to anyone who didn’t respond the first time.

To incentivize respondents, we offered a promo code discount after completing the survey to encourage them to return.

Survey results indicated that the primary reason users canceled their subscriptions was the absence of key features they needed in the CRM.

This survey helped us focus our efforts on addressing the missing features first.

User interviews

After reviewing the survey results, I started finding participants for the interviews by screening survey respondents and selecting users who canceled their subscriptions because the features they needed were not available, but were willing to return if those features were improved.

To encourage potential participants, we offered a gift card and promo code discount to support reactivation rates.

The objectives of these interviews were to:

Understand how and why users manage messages from Meta platforms in their business context.

Discover what are the highs and lows of their experience while managing messages from Meta platforms.

Find out what features users desire while managing messages from multiple Meta platforms.

Discover what other apps they use to manage messages from multiple Meta platforms.

This is the script I used for the interviews.

After conducting 10 interviews, I created a summary report of the findings for each user that was interviewed, and these reports were used later as a starting point for team alignment activities like affinity maps, empathy maps, data-driven personas and user journeys.

The purpose of these collaborative activities was to understand user needs, workflows, pain points, context in their business environment, and uncover opportunities for innovation or improvement.

Affinity map

We organized the notes, quotes, and observations from users by grouping similar insights and issues into clusters using sticky notes in order to spot patterns and identify themes.

This affinity map helped us review and discuss the clusters and vote on the most relevant insights and issues.

These were the issues we decided to address first:

1- Users can’t view customer info or adjust inclusion when scheduling a campaign for a funnel stage.

2- Users struggle to find the “Schedule a message” option and to manage scheduled messages.

3- The process of updating notes for each customer is tedious and time-consuming.

Data-driven personas and empathy maps

After the affinity map workshop, we continued with the creation of data-driven personas and empathy maps to represent our users by segment, and to review user attitudes and behaviors.

This activity helped us summarize qualitative data from the user interviews and articulate what we know about each type of user in a more organized and empathetic way.

These are the data-driven personas and empathy maps we created to understand our users, prioritize their needs and support our decisions about the next steps.

Click the image below to view it full screen.

FigJam link

User journey map

After updating the empathy maps, we proceeded to create the user journey map.

Here’s the user journey map that helped us clarify what’s working well in the current journey and where to focus our time, budget and energy for the best results.

Click the image below to view it full screen.

FigJam link

Wireframes

Following the completion of the empathy-mapping workshop, I created wireframes so we could compare layouts, interactions, and user flows between the current and future states of the features we decided to improve.

Schedule a message feature (current state):
Figma link

One of the pain points identified during research was that users struggle to find the “Schedule a message” option and to manage scheduled messages.

Schedule a message feature (future state):
Figma link

With this update, we intended to help users schedule messages faster by placing the “Schedule a message” option in the message input area, next to the most used actions.

Prototypes