Sales CRM
Building a sales platform for the human service customers want
One Workplace is a global platform designed to connect SoftwareOne sellers with customer requests from around the world. The platform tracks all interactions and requests, making them accessible to sales agents across various departments.
Enable sales agents to manage their tasks more effectively within the digital platform instead of spreadsheets, respond to requests more quickly, and have more time to pursue new customer opportunities. To accomplish that, I identified the following success metrics:
Allow sales agents to easily learn the new system without time waste or sacrificing efficiency
Help sale agents process sales faster and refocus their energy on chasing new leads
The One Workplace team struggled to identify high-impact improvements, relying on internal processes over real user needs. Without tools to track user behaviour, gathering data-driven design requirements was challenging.
One Workplace relied on Microsoft templates, which limited customisation, restricting visual simplification, text accessibility, and menu hierarchy adjustments.
A key learning was the importance of aligning early with the engineering team and existing UI libraries to streamline collaboration and support the overall workflow. Simplifying the visual design to its essentials, especially when the current UI limits creative solutions, can be challenging. This experience underscored the value of team collaboration and showed that, even in constrained environments, strong design fundamentals can lead to effective, functional outcomes.
Design Thinking is a human-centred approach to creatively solving problems and building meaningful experiences.
Without available usage data to begin the investigation and no integrated technology to support my analysis, I tested the existing platform with sales agents from different countries to draw the starting point of the experience. The content for the test included the main user journeys of the existing platform, such as: finding assigned tasks, claiming new customer requests, tracking the team's progress, find newly generated requests.
I reported test results based on users' success or failure to interact as expected, revealing an average success rate of only 57% across both expert and inexperienced users. This prompted the business to recognise the urgency of usability improvements to enhance adoption rates.
During the test sessions, I dedicated time to observing people's challenges and investigating their sentiments, so as to have both a quantitative and qualitative picture of real users' needs.
The business jargon posed a significant challenge for users, requiring familiarity with acronyms and company-specific terms for successful navigation. Only highly trained users found the experience comfortable.
The Notifications component deviated notably from industry standards, causing users to overlook it. In tests, 100% of users failed to notice the feature, despite its prominent placement, due to their preconceived expectations of where it should be found.
.png)
In addition to usability tests, I interviewed 5+ sales agent to learn about their processes and expectations. I collated that information to generate three pillars that have driven and kept the design on track moving forward.
I simplified the labels of each critical menu area and in-page navigation, aligning them with their core functionality and grouping similar features. This approach, inspired by the Gestalt design principle of proximity, enhances and facilitates navigation.
I relied on common industry-standard patterns for notifications and positioned the access point of the feature (the bell icon) in the top-right area of the main menu. This ensures users are familiar with its location, facilitating navigation.
I chose a code-based prototype to enhance the realism of navigation, a level of detail I couldn't achieve with Sketch mockups for every aspect of the app. I aimed to give users the freedom to explore every corner of the application and immerse themselves in a genuine experience while running tests. Moreover, due to the limitations of the template-based technology used to build One Workplace, I couldn't utilise the company's design system, leading me to opt for a code-based prototype to minimise the risk of extensive styling and streamline my creative process.
.png)
I validated the design concept by conducting a short round of usability tests with sales agents to capture major inconsistencies. Feedback was generally positive, and I was able to gather additional requirements by observing their interactions.
Despite receiving positive feedback from users and Stakeholders, concerns arose within the engineering team regarding the previous design's liberal approach to decluttering visuals, which technology limitations made unfeasible.
To address this, I adjusted my strategy by incorporating alternative existing components that better aligned with design requirements, ensuring usability within technological constraints. With the engineering team's support, we established a dedicated testing environment using the existing technology to create a Proof of Concept.
I overall conducted 15+ individual testing sessions lasting 15 to 30 minutes each, and involving participants of varying expertise levels. 11 out of 12 journeys have been fully optimised, resulting in user success rates ranging from 65% to 100%. This marks a significant improvement from the initial performance, where the execution of 5 journeys failed critically and 3 were performed with errors.
Using the coding language R and a customised database, I meticulously tracked and analysed results, monitoring the evolution of design experiences. By recording every participant's journey test results from each session and feeding them into the script, I could identify trends and fine-tune journeys that were underperforming to optimise adoption upon release.
.png)
The business complained that incoming application requests could not scale. The CEO's brief was one line: work on the customer journey. Everything else had to be discovered, so I turned an ambiguous request into clear problems, explored several approaches, and identified who needed to be involved.
Help sales agents handle applications more easily, while keeping the high-quality, human service that customers asked for and the market complained about when it was missing.
The work ran in seven steps: research, define, align, design, validate, adjust and release. Each step produced evidence that shaped the next.
The brief from the CEO was one line, so I planned and ran the discovery on my own. I asked two questions: what do we know about our customers, and what major pain points do they deal with?
With direct prospective customers.
From GA and Heap, including conversion rates of digital funding requests.
On Reddit, Facebook and similar channels, for generic market complaints.
Interviews were blocked, because there was no budget for direct end-user interviews. I interviewed the people closest to customers instead: the sales agents.
The website targeted the wrong segment, with many personal loan seekers, sole traders and start-ups.
The website asked for registration too early, causing a high drop-off at the last step.
High costs, opaque declines and poor customer service were the most-complained topics.
Agents were taking the path of least resistance. High service was a sales strategy, because when people talk about money they want to talk to humans.
The next question was how well the current technology supported agents in serving customers. The answer was poor, for four reasons:
These became the proposed improvement areas for the project.
I started from a blank canvas, with first intuitions about responsibilities and tasks, and used the project to assess AI tools for design for the first time.
I ran multiple rounds of usability tests with sales agents. The tests exposed new requirements, so I fixed the usability issues and added the requirements into the design.
Just before release, the stakeholder changed and brought unexpected requirements. Among them was a request to add back an area known to be unused.
The issue: I needed to understand why. What added value did that area bring, in the eyes of the new stakeholder?
What I found: the sales team had changed its way of working halfway through the project. Tasks were now distributed to each individual, not to specific roles within the team. The request had become scope creep, driven by a real shift in how the team worked.
What I did: I went back to the canvas.
I used AI to generate all official, user-focused system documentation, working with Claude through its Confluence connection and Claude Code.