Analytic Partners

$13M

in additional revenue after rebuilding the flagship product from Angular to React.

25%

year-over-year revenue growth after modernizing legacy enterprise software.

85%

less time spent collecting data, once collection and review were consolidated.

01 — Overview

Three tools, one intelligence layer, one route from data to decision.

GPS Enterprise is built on a foundation of technologies that streamline data management, modelling, forecasting and scenario planning, with security at the core. It brings together three fully integrated tools — Adapta for data, AMP for analytics and Prophet for decisioning — powered by the proprietary ROI Genome intelligence layer, so the journey from data to decision-making happens in one place rather than across four.

My work ran across the platform: the dashboards and reporting surfaces, the scenario-planning flows inside Prophet, and the component patterns that hold them together. Most of it came down to one problem — taking dense modelling output like model fit, incremental revenue, diminishing returns and driver contribution, and putting it on screen so a marketer can act on it without a data-science background.

What I did

Dashboards, reporting and data-visualization surfaces
User-flow mapping for scenario planning, simulation and optimisation
Reusable UI components and patterns across the three tools
Prototyping and iterative review with analysts, marketers and engineering
Accessibility and inclusive design through the lifecycle

02 — Opportunities

The questions the platform had to answer.

01How might we take a marketer from raw data to a decision without a detour through the analytics team?
02How might we make a scenario — its inputs, its simulation, its optimisation — readable at a glance and still editable in detail?
03How might we show model quality (fit, confidence, benchmark) so someone trusts the number they are about to act on?
04How might we keep data analysts, marketers and business leaders looking at the same picture of the same quarter?
05How might we standardise documentation and specification so design intent survives the hand-off to engineering?

03 — Approach

Six areas of design work.

01

Adapta — data

Dataset creation, editing and batch review
Due dates and pending-review states, so data owners can see what is outstanding
Diminishing-returns values viewable and editable in place

02

AMP — analytics

Insight views for spend and incremental revenue, with period-over-period movement
Model-fit charts plotting actual against predicted net sales
Data score and ROI Genome benchmark score surfaced together

03

Prophet — decisioning

Scenario management as the landing surface, with datasets one step away
Input tabs for simulation, optimisation and non-marketing drivers
Output tabs for summary, driver summary, waterfall and detail

04

From whiteboard to flow

Scoping and design sessions worked through test-versus-control structure, cell counts, date ranges and confidence levels before anything reached a screen
Flows mapped end to end and colour-coded by surface type — landing page, full page, modal, tab — so engineering could read the build from the map
Work-in-progress branches marked as such rather than quietly omitted

05

A shared vocabulary

Tables, cards, modals, tabs and empty states built once and reused
Illustrated empty states that name the next action
Patterns documented so a new surface starts from the existing language

06

Setup and administration

Sub-accounts and periods, milestone configuration and executive dashboard review
Kept off the daily path but never more than a tab away
Formal review steps and documentation standards agreed with the team

04 — Outcomes

Less time collecting, more time deciding.

Faster to the number. Consolidating collection and review cut data collection time by 85% — the figure the engagement was measured on.

One picture for three audiences. Analysts, marketers and business leaders worked from the same dashboards and the same definitions, which took most of the translation work out of the middle of the process.

A vocabulary that scaled. Three tools shipped under one interface language, so a new surface inherited its patterns instead of inventing them.

“I’m happy to recommend Todd Crawford for his contributions as a UI designer. During our time working together at Analytic Partners, a data analytics company, I’ve observed his consistent delivery of high-quality designs that meet client and company expectations. Todd has a track record of translating complex data management and visualization projects into visually compelling, responsive, and user-friendly interfaces. His unique blend of creativity and technical proficiency is evident in his work, and his collaborative nature makes him a standout team member.”

Misha Holzman

Manager, UI Design

05 — Walkthrough

The Adapta scheduler landing page, laying out what the tool does as five cards - data ingestion, review, data scorecard, upload and scheduler - with the last two shown as not yet available. — 01 Adapta — what the data side does, stated up front

A client data collection view for the Nespresso team: sub-account and period selectors above grouped sections for sales data, media data and operational or non-marketing data, each row carrying contributors, a due date and a review status, with one row flagged seventy days overdue in red. — 02 Whose data is late, and by how long

The review configurator showing a sales dataset: tabs for fact data, mapping tables, questions and comments, and process data, above a preview of the fact table with a breakout column and ten dimension columns, processed and staged badges, and pagination reading one to ten of one hundred. — 03 Fact data, previewed before it is trusted

The same configurator with nothing loaded: an empty state reading no fact data has been uploaded yet, in the space the table would occupy. — 04 The empty state, in the table’s own space

The mapping tables tab: a dimensions table beside a calendar panel carrying its own empty state for unmapped calendars, with a download and upload menu open. — 05 Mapping tables, and a second empty state

A CPM trend index chart opened over the dashboard, running from 2019 to 2023 with the forecast period shaded and distinguished from actuals, the source named beneath it, and a single close action. — 06 A trend, with the forecast visibly separated

The insights dashboard configurator for a client dashboard: tabs for project parameters, two sets of data selections and dashboard formatting, previews of model fit, source of change and contribution, then general settings for metric, hierarchy, dimension members, period type and periods, with a notes field. — 07 Configuring the dashboard someone else will read

01 / 07

06 — Prototypes

01 Redesign proposal Design option one, with login screens for external users, a landing page that launches each custom dashboard and a new scenario builder; a chat exploration; and design option two’s custom dashboard.

02 Insights presentation A client insights presentation running inside the platform: the marketing mix deck opening from the dashboard, then a data restatement read out as contribution changes across paid search, digital tactics and display.

03 AMP — Agile Learning The experiment workspace: create a new experiment, then track every one of them by status, activity, primary KPI, dataset and owner in a single sortable table.

04 Adapta — checklist builder Scoping a summary across brands: filtering the checklist by industry, category, dataset and owner, with the comment thread and the results log alongside it.

01 / 04

Three tools, one intelligence layer, one route from data to decision.

The questions the platform had to answer.