Enterprise Software
CRM Transformation
Architecture for High-Stakes Relationships
Partners Were Choosing Excel Over Our CRM
That was the signal I needed. During Partner shadowing sessions, one Partner told me directly that he would rather work in Excel than rely on the existing CRM. That wasn't a feature request, it was a fundamental indictment of the tool's navigation model.
When I traced why, the pattern was clear: a simple meeting review required 7+ clicks and 4 full page reloads. Partners assessing a list of 50 pitches had to click into a record, wait for the page to load, and hit back, losing their scroll position and filters every single time. The natural workaround was keeping 8–10 browser tabs open simultaneously, which created its own cognitive overhead.
The CRM wasn't broken because it lacked features. It was broken because its host environment (the company intranet) imposed structural constraints that made expert, high-frequency workflows painful. The tool was built for occasional use; Partners needed it for daily, deep-triage work managing multi-million euro client relationships.
User quote
"I'd rather use Excel." — Partner, ATOZ.
This wasn't a feature complaint. It was evidence that the navigation model had fundamentally failed expert users.
my role
Senior Product Designer
Timeline
April 2024 – August 2025
WHAT I DID
End-to-end UX strategy:
IA
interaction design
stakeholder alignment
engineering collaboration
Measured Outcomes
Results were gathered over three release cycles through Partner feedback sessions, structured interviews, and observed usage during shadowing. These are directional metrics based on qualitative evidence, not instrumented telemetry, and are presented as such.
~25
%
Faster Workflow Velocity
~60
%
Fewer Reloads
~40
%
Increase In Tool Adoption
Reframing the Roadmap
From Features to Workflow Throughput
When I joined the project, the roadmap was feature-led, a backlog of additions to an already-struggling system. My first move was to shift the conversation upstream.
The question wasn't 'what do we add?' It was 'why is the system structurally incapable of supporting expert workflows?'
I ran heuristic workflow audits, mapping the click-depth of core tasks. I shadowed Partners in their actual working environment. The evidence pointed to one root cause: the intranet's page-based navigation was destructive by design. Every click broke the user's context.
The strategic recommendation I brought to leadership was a decoupling strategy, pulling the CRM out of the intranet entirely and building it as a standalone vertical tool.
this was an architectural pivot not a visual refresh, it would give us the freedom to design for expert users without being constrained by generic intranet patterns.
The business case
Partners manage high-touch, multi-million euro relationships. Every second spent on navigation friction is a second taken from advisory.
Faster workflows = more time on client work = direct revenue impact.
The Anchor Model
Non-Destructive Navigation
The core architectural decision was replacing page-based navigation with a Non-Destructive Side Panel (Drawer) system.
The principle was simple: the list view is the Anchor. When a Partner selects a Contact, Company, or Pitch, the record slides out as a panel, the list stays untouched, with scroll position and filters fully preserved.
I evaluated alternatives deliberately:
Modals were too restrictive for the depth of data Partners need to review in a single record.
Browser tabs created tab-overload and broke the single-source-of-truth mental model Partners needed for triage.
Full page navigation was the existing pattern, and the root of the problem.
The Anchor Model meant Partners could triage an entire list of 50 pitches, open records, take notes, and close them, all without ever losing their place. The workflow that previously required 7+ clicks and 4 reloads now happened within a single session state.
Designing Around Legacy Infrastructure
The side panel architecture introduced a real engineering challenge: rendering detailed record content within a panel, on top of a live list, inside a legacy infrastructure not originally designed for this interaction model. Performance was the risk, a sluggish panel would have been worse than a page reload.
I worked closely with the engineering team to understand the constraints before finalizing the interaction model.
The key negotiation was around lazy loading: panel content would load progressively, with the most critical data fields (name, status, key financials) prioritized in the first render pass. Secondary data loaded asynchronously. This kept the perceived performance fast even when underlying data fetches were slower.
The result was a panel that felt immediate. Partners perceived it as instant because the data they needed first appeared first, a design decision that required close collaboration with engineering to sequence correctly.
The Excel Mapping Engine
Designing for Zero-Error Imports
A significant part of the CRM transformation involved migrating existing contact and relationship data into the new system. The original approach was a blind upload, users would import a spreadsheet and discover errors after the data had already entered the database.
The problems this caused were systematic:
Wrong columns being mapped to wrong CRM fields, phone numbers landing in email fields, dates in name fields.
Duplicate contacts being imported when source spreadsheets had inconsistent naming conventions.
Formatting errors, date formats, phone number structures, currency formatting, that corrupted records silently.
I designed a Pre-Import Mapping UI, a sandbox environment where users manually align spreadsheet columns to CRM fields before any data touches the database. The system supported three import pathways: direct import from Lotus (the firm's existing email solution), a structured mapping flow for Excel uploads, and a manual hand-entry path for smaller datasets.
The key design principle was: make errors visible in the sandbox, not in production. Partners could see a preview of how their data would map, catch mismatches, and correct them before committing. This transformed data migration from a stressful, error-prone process into a controlled, auditable one.
Winning Over Skeptical Partners
The most important audience wasn't leadership, it was the Partners themselves. They were the skeptics. A new UI, a new navigation model, a new data entry flow: these represented disruption to workflows they'd adapted (however imperfectly) to the old system.
My approach was to make the case through hands-on experience rather than presentations. I ran working demonstrations where Partners used the prototype on real assessment tasks, their actual pitch lists, their actual contacts. The goal was to let the speed of the Anchor Model speak for itself.
The moment that consistently shifted skepticism was when Partners realized they hadn't lost their scroll position after reviewing a record. That single interaction, returning to exactly where they were in a list, was the proof point that converted users. It was a small thing that represented an enormous structural change.
The design principle that won the room:
Partners didn't need to be told the tool was better. They needed to feel it. One interaction, returning to their place in a list, did more than any presentation slide.
On the feature complexity front, I also pushed back on stakeholder requests to add aggressive filtering controls and persistent widgets to the Pitch Management module. The proposed additions would have consumed 30% of the viewport, crushing the core data table. Using a prototype, I demonstrated that when everything competes for attention, nothing wins.
We adopted Progressive Disclosure, advanced filters behind a Smart Search bar, secondary widgets in the side panel, trading immediate visibility of every filter for maximum visibility of core data.
What This Project Taught Me
Designing for high-skilled professionals is about time economics.
Every wasted click is a hidden tax. Every second of friction is money lost.
Information Architecture IS Business Strategy
The single most impactful decision on this project wasn't a visual one, it was the Anchor Model. Choosing to treat the list view as a persistent, protected workspace rather than a transient navigation state changed how Partners worked. That's an IA decision with direct business consequences.
Experts Want Efficiency, Not Simplicity
The Excel Mapping Engine is a more complex interaction than a blind upload. But it makes the user's life simpler by preventing downstream errors that were far more costly to fix. Designing for expert users means sometimes adding a layer of controlled complexity to eliminate unpredictable downstream chaos.
Prototypes Win Arguments
Every major decision in this project, the decoupling strategy, the Anchor Model, the Progressive Disclosure trade-off, was defended with a working prototype, not a slide deck. Letting Partners experience the difference themselves was the most effective stakeholder alignment tool available.
