Redesigning How Tax Teams Track Work
Tax teams managing thousands of deliverables were spending the beginning of every session configuring the table before they could do any actual work. The platform had the data they needed. The Deliverables app wasn't built around how tax work was done.
The Deliverables app wasn't built around how tax work was actually done. Teams had developed workarounds in Excel to compensate, using Sightline mainly to log status before client check-ins rather than to manage their actual work.
Background
Sightline is PwC's tax compliance platform, designed to help tax teams collaborate in real time while giving clients visibility into filing status and engagement progress. At the center of that experience is the Deliverables app, the primary hub for tracking e-file returns, monitoring deadlines, and managing status across the full tax lifecycle.
A single engagement could contain thousands of deliverables spanning income tax filings, extensions, estimates, and other specialized tax deliverables. Each had its own timeline, status, and filing requirements. The Deliverables app was meant to be the command center for all of it. In practice, it wasn't.
The problem
Teams were working around the platform Teams were regularly turning to Excel to handle work the platform couldn't support well, updating status in Sightline before client check-ins and then returning to spreadsheets to do their actual day-to-day tracking. The pattern was consistent enough to show up in interviews and in the habits teams had built around the platform.
Previous fixes had been blocked by technical cost I had tried to address this previously, but improvements were deprioritized when the technical cost of building on top of the existing Bootstrap-based table was too high to justify.
A technical shift created the opening Engineering identified a need to migrate the Deliverables table to AG Grid for performance reasons. When evaluating its capabilities, I saw that the features previously too expensive to build were now within reach: inline filtering, column customization, and grouping. The performance problem and the UX problem had the same solution.
Understanding users
Before the AG Grid migration became a defined project, I was already working to understand where the platform was failing our users. I ran research sessions with preparers and reviewers across staff and manager levels, observing them complete real tasks and documenting where the workflow broke down.
How might we help tax teams get to the right work without having to configure the platform before every session?
From that question, I set out to observe how users verify engagement readiness ahead of e-file deadlines. I chose this scenario because time pressure made every friction point more costly. If the table slowed you down in the week before a filing deadline, there was no slack to absorb it.
"I spend more time setting filters than actually reviewing my tasks."
— Associate, Preparer
What the analysis revealed was that most of the friction was happening before users reached any real verification work. The heaviest cost was in the steps just to get the table into a useful state.
What we found
Mental model mismatch
All deliverables appeared in a single undifferentiated list, organized around data architecture rather than how tax teams move through their work. Users thought in terms of deliverable type (extensions, estimates, returns) not just by workflow status. The flat structure meant every session started with manual filtering just to see relevant work.
Every session started with setup, requiring users to filter out irrelevant items before they could focus on the work that actually applied to them.
Filtering friction
The filter experience worked against users. Results didn't update until an apply step was triggered, so any unexpected output meant starting over with no way to preview what a selection would return. The quick filters designed to help were routinely ignored because their criteria didn't match how teams naturally segmented their work.
Filters that required submitting before showing results put the cost on the user. One unexpected return meant retracing every step from the beginning.
"We categorize deliverables using the primary group field, which is the most common filter I'm using. I never use these filters."
— Senior Associate, Preparer
Users wanted faster ways into their work, but quick filters provided a range that was too broad to be useful.
Information buried behind overlays
Finding key tax information like e-file system or tax due amounts required opening a detail panel for each record. For multiple records, that meant opening the panel, noting the relevant details, closing it, and repeating for each deliverable. Comparing records meant cycling through that sequence over and over.
"When we are setting up our e-file systems, it's important to have more details than just the status available at a quick glance."
— Manager, Reviewer
The slideout made sense for editing a single record, but slowed down anyone who needed to scan and compare data across multiple records at once.
Design directions and validation
With the problems understood, I explored three directions and brought them to users for feedback. The directions covered how deliverables were organized, whether the quick filter tiles could be more useful, and giving users more direct control over the table.
Narrowing deliverable focus
Breaking down the full list into meaningful segments based on deliverable type (extensions, estimates, returns) gave users a faster path to the work most relevant to them. The phase-based framing matched their mental model in a way the existing structure never had.
Reducing visual noise helped users orient faster and focus on immediate priorities.
Grouping by workflow status (preparing, reviewing, filing) didn’t align with how teams segmented day-to-day work.
Rethinking quick filter tiles
The quick filter tiles offered shortcuts users rarely used; the criteria weren’t specific enough to match how teams actually segmented their work. I explored whether surfacing more relevant data could turn them into a more useful starting point.
Users appreciated seeing their own assignment data as a top-level filter, finding it more relevant than the existing category options.
Adding data visualization did not improve user utility. The tile criteria weren't mapped to how users thought about their work.
Putting users in control
Column-level filtering, customizable layouts, and row groupings teams could reuse across sessions. This generated the most enthusiasm. Users immediately saw how it would reduce the overhead of setting up their workspace each time and let teams establish consistent starting points across an engagement.
Users were enthusiastic about saving and reusing personalized views tailored to their tasks.
Validating the directions
After bringing concepts to users, I mapped what resonated, what fell flat, and what constraints surfaced to determine what was worth building and what needed to change before we could.
The constraint that changed the scope
Validation pointed toward clear improvements, but whether we could build them was a separate question.
UX improvements were scoped and designed, then deprioritized when the build cost was too high.
A required infrastructure migration created the conditions to revisit what had been blocked.
When the team needed to migrate the Deliverables table to AG Grid to address performance issues, I evaluated what the new foundation could support. The interaction patterns we had tried to build before were now within reach. Getting UX improvements into scope required making that case early.
The solution
The first phase delivered the core improvements: reorganizing deliverables by type and giving teams direct control over how they filtered, sorted, and grouped their data. A second phase built on that foundation with workspaces, preset and customizable views teams could return to without reconfiguring from scratch.
-
01
Laying the foundationPhase 1
Restructuring deliverables around how teams think.
-
02
Flexibility and controlPhase 1
Empowering users with tools to organize and segment their work.
-
03
Curated workspaces Phase 2
Replacing quick filters with tax-specific workspaces.
-
04
Customizable workspaces Phase 2
Extending workspaces into user-controlled views.
01 — Laying the foundation
The flat table left users with no clear starting point for the work that mattered most. The first phase addressed that directly, reorganizing how deliverables were presented and expanding what users could see without leaving the table.
Deliverables organized by type
Tabs separated deliverables into Income Tax Filings, Extensions, and Estimates, reflecting how tax teams moved through the yearly tax cycle. Instead of one large undifferentiated list, users opened the app to a focused view of the work most relevant to them.
More data, visible at a glance
Beyond reorganizing deliverables, we expanded the amount of data available directly in the table. Key filing details that previously required opening a detail panel were now visible at a glance, reducing how often users needed to navigate away just to scan basic information.
02 — Flexibility and control
Reorganizing the table solved the structure problem. The next layer gave users control over how they worked within it, replacing the rigid filter experience with tools that responded to how teams actually segmented their work.
- Contextual filtering: Filters built into column headers gave users immediate, inline results without leaving the table or applying a separate step.
- Column customization: Giving users control over column visibility and order let teams tailor the table to their workflow without affecting other users' views.
- Data grouping: Grouping deliverables by shared attributes like entity, reviewer, or status helped users break large lists into manageable, structured views.
03 — Curated workspaces
Even with improved filtering and data availability, starting a session still required manual setup. We replaced the underused quick filters with purposeful preset views built around how tax teams actually work, giving users a consistent starting point without requiring them to configure anything themselves. Teams spent less time on setup and approached reviews more consistently across the engagement.
04 — Customizable workspaces
Once teams were working with curated workspaces, we evolved the concept further, giving users the ability to build and save their own views. After configuring filters, column layouts, or row groupings, users could save those settings as a named workspace and share it across their engagement team. This strengthened consistency across reviews, reduced per-session setup, and gave new staff a faster path to getting oriented on an engagement.
Outcomes
The work landed well with both internal tax teams and the business-side partners who used Sightline to monitor their engagements. The clearest early signal was around phase separation. Teams that had been managing everything in a single flat list suddenly had a focused starting point, and finding relevant work created less friction from day one. Inline filtering addressed requests that had been sitting on the business side for some time, and the project was highlighted at PwC Tax's quarterly working session as a notable improvement to the platform experience.
The introduction of curated workspaces and the ability for teams to personalize them further turned the table from something users configured into something they could rely on from the moment they opened it.
Table load time for datasets with 8,000+ deliverables
Products across Sightline that adopted the AG Grid standard
Beyond Deliverables
The broader impact extended beyond the Deliverables app. Other teams across the Sightline ecosystem migrated their own tables to AG Grid after seeing what the migration produced. To support those teams, I built a pattern library: interaction patterns, component documentation, and implementation guidance for adapting existing table designs to the standardized table format. Two other designers and a design manager reviewed and refined it before distribution to make sure the patterns would scale across different product contexts.
A year after my layoff, a designer still on the team told me the documentation I'd created was still being used to onboard vendors onto their own AG Grid transitions. Work that outlasts the person who did it is the clearest signal that it solved the right problem.
Reflections
What I'd do differently
I'd push for a defined measurement plan before the work began. We knew what the problems were, but we didn't establish clear signals of success upfront, which made it harder to track behavioral impact over time. I'd also expand the research scope. Our sessions focused on internal PwC users, which gave us a clear picture of preparer and reviewer workflows but created blind spots around how client-side teams experienced the platform.
What this reinforced
Knowing when to move matters as much as knowing what to build. The problems in Deliverables were identifiable long before this project started, but what changed was the conditions that made them solvable. Recognizing that the opening existed and making the case early was what turned a performance migration into a product redesign.
Thanks to Joyce Lee, Bryce Newberry, Bernie Romero, and Jennifer Jennings for their collaboration on this project.
Back to work