Jade Parrish

Data & Service Design · 2024 to 2025

Getting everyone back to the same numbers.

Bringing reporting scattered across five tools into one place that brands and internal teams could both trust.

Role

Senior Product Designer

Research

8 stakeholder interviews

Outcome

Up to 8 hours, halved

Audit found

Stock overstated by 80%

Setting the Scene

In 2024, I led the redesign of SoPost’s Reporting Area. The honest version is that the Commercial team didn’t trust the numbers anymore, and I got to be the person who went and found out why.

Before this work, results were scattered across Campaign Manager, spreadsheets, standalone Looker dashboards, Google Docs, and a complex internal tool called Mission Control. Mission Control and Campaign Manager refreshed at different rates, so their totals did not match, and everything built from them became inconsistent and sometimes incorrect. Reports were time-consuming to compile and confidence in the numbers was low.

The goal was to bring all reporting into one reliable suite, so every team from Commercial to Operations, and the brands we worked with, had the same trustworthy view of performance and value.

Fig. 01Before the redesign: one report for the brand, two systems that disagreed, and the tools each market built to optimise their own workflows.
BRANDOne performance reportLINE OF VISIBILITYSYSTEMS OF RECORDMission Controlbuilt for engineersCampaign Managerreporting split across pages!the numbers did not agreeWORKAROUNDSSpreadsheetspersonal and team-sharedNORTH AMERICAAUSTRALIAUKFRANCEGERMANYGoogle Docsnumbers pasted inLooker boardsspun up ad hocASSEMBLYAssembled by hand, every time!Up to 8 hoursof Commercial’s time on a single complex brand report!Fail pointBuilt by teams, owned by no one

Scroll to see the rest

Part One

Auditing and Aligning the Foundations

What was wrong

Reporting had become fragmented and unreliable. Mission Control was built for engineers, and Commercial had ended up relying on it anyway. Campaign Manager split reporting across isolated pages. Between them the two systems refreshed at different rates, duplicated charts, and disagreed on metrics that shared a name, and nobody held a clear picture of how the datasets connected.

The effect was cumulative. Every brand report had to be compiled by hand from sources that did not agree, which took hours and left the numbers open to question once they arrived.

Process

The audit started with the people using the data. Commercial sat across North America, Australia, the UK, France and Germany, and each market had settled into its own way of reporting. They kept coming to me with gaps and differences in their numbers, which made their decks hard to understand and talk through. Building the WMS had already uncovered discrepancies in the data, including mismatched refresh rates between the two systems.

I audited every existing reporting view across Campaign Manager and Mission Control, documenting duplication and logic gaps. The picture became much clearer: the discrepancies ran deep, and there was no single, central reporting hub.

In parallel, I worked with the Commercial Team to review the reports they shared with brands, pinpointing where manual edits and visual inconsistencies created risk.

The audit didn’t say Mission Control had to go. Engineers found it useful, and it was built for them. The problem was that Commercial had been given access to a developer tool, and were pulling numbers out of it for brands. My proposal to the CTO was to keep Mission Control for engineering, build Commercial reporting of their own, and treat that as product work rather than maintenance.

Fig. 02A page from the audit: a Customer Success Director’s commentary on Mission Control, including two campaigns where the stock it reported was close to double what was actually there. The figures are changed.
A page from the Mission Control audit. A Customer Success Director's comments are pinned around a screenshot of a campaign page in Mission Control. Two of them record the stock discrepancy: on one campaign the system reported 61,400 samples with 42,800 remaining when there were really 34,000 to start and 15,600 left, and on another it reported 34,200 with 23,100 remaining against 19,500 and 8,400. Her other comments say she uses only the approved filter, that the data excludes customer service samples, that these stats are the ones referenced in campaign reports, that Campaign Manager is a better place to gather top level stats, and that she cannot see what samples remain for a particular activation without going elsewhere.

Research and Testing

To validate direction, I interviewed eight people across Product, Commercial, Engineering and Operations. Using a Figma-based testing matrix, I mapped their workflows and captured pain points. The findings were consistent:

  • People were piecing together insights across multiple tools
  • Each team defined “reporting” differently (fulfilment vs engagement vs exports)
  • No one had a full, end-to-end picture of campaign performance

That was enough to get the rebuild properly backed.

Part Two

Redesigning the Reporting Experience

Problem

Campaign Manager’s reporting pages (Overview, Orders, Demographics, Feedback and Reviews) existed in isolation. There was no hierarchy, shared data model or visual consistency.

Design Approach

I defined a single Reporting Hub that could:

  • Unify data sources under one structure
  • Standardise metrics, terminology and visualisation styles
  • Simplify navigation, so people could find things without being told where they were
  • Show where a number came from, so people could check it rather than take it on faith

Exploration

I created a series of prototypes exploring:

  • Mapped how data moved between systems as it was, and as it needed to be
  • Card-based layouts grouping metrics by performance type
  • Standardised filters, comparison modes and accessibility patterns
  • Error and low-data states, so a gap looked like a gap rather than a zero
  • Early versions of the brand-level summaries, and the structures the ROI work would later need

Each iteration was reviewed with Product, Commercial and Engineering to ensure the hierarchy reflected how insights were actually used and shared.

Fig. 03Orders as shipped: one page per reporting area, card-based, with filters, a time series and a rejection breakdown in a shared structure. Figures are illustrative.
The Orders page as shipped in Campaign Manager. A left navigation lists Campaign Overview, Orders, Activities, Email, Feedback, Reviews and Demographics. The page carries filters for date range, activity and product, with the product filter open as a dialog. Order overview shows submitted, approved against a cap, and dispatched. Below it a chart plots approved and dispatched orders over time with a hover readout, and under that a row breaking order rejections into duplicate, intelligent filter, invalid address and other.

Impact

The new Reporting Area established a single source of truth for campaign performance.

A complex brand report took up to eight hours to assemble. The redesign halved it, and introduced consistent visual language and data standards across every team.

Both figures came from the Commercial team themselves, in face to face sessions. Nobody was tracking report build time, so these are their estimates, consistently given.

The work carried on well past 2024. That foundation became SoPost’s ROI and Impact Dashboards, Comparison Tables and the Reporting Hub, which I picked up with our Product Manager the following year.

By 2025 it had become a new hub layout rather than a reworking of the old pages, prototyped in shadcn so the components were real rather than drawn. This is as far as it got.

Fig. 04The hub as proposed: summary metrics, channel performance and campaign stock in one structure. Designed in 2025 and not built, and the figures in it are illustrative.
The proposed hub, its brand overview page. A left navigation lists Overview, Orders, Stock, Distribution, Emails, Clicks, Consumers, Feedback and Ad performance. An insights strip carries a low stock alert and a top campaign. Below it, four summary cards give total orders, reach, engagement rate and cost per consumer, beside a chart comparing reach against engagement across six social channels. A campaign performance table then lists five products with their region, channel, stock, usage and stock status. Several section headings are still placeholders, because the page was a proposal rather than a built screen.

Reflection

Staff trusted the numbers again. The tools were joined up, Engineering and the Commercial Team were working from the same definitions, and reporting stopped being something you rebuilt by hand every time.

It became the foundation for everything that followed. The stock work and the ROI dashboards both trace back to that audit. What started as an argument about whose numbers were right ended with one set everybody used.

“I had the pleasure of working with Jade for nearly three years at SoPost, where she was a driving force behind the evolution of our design and reporting capabilities.”

Isabel Llamas (opens in a new tab) · Senior Product Manager, SoPost
ContactAvailable part-time

Let’s improve something together.

Open to part-time roles and fractional contracts, two to three days a week.

hello@jadeparrish.me

Previous

Bringing stock and fulfilment data into one clear view

Next

Designing notifications that scale beyond a single alert

Back to selected work