Case Study 03 · AI in Action

Custom
Risk Matrix
Power BI Visual

Using AI to build reusable reporting components — without paying for marketplace visuals.

Project teams needed a clear, configurable way to present risk by Likelihood and Impact in Power BI. Standard visuals didn't fit; marketplace visuals introduced cost, dependency and governance issues. AI was used as a development partner to scaffold and refine a custom Power BI visual — built once, configured per project, owned by Arup.

Power BI GitHub Copilot M365 Copilot Claude

£0

Marketplace subscription dependency

~5h

AI-assisted development time

5×5

Likelihood × Impact matrix cells

∞

Reusable across Arup

01 · The 30-Second Story

Why This Exists

PMO teams need to show risk exposure clearly in Power BI, but standard visuals couldn't render a true Likelihood × Impact matrix. Suitable marketplace visuals were paid, limited or not customisable. AI supported the development of a bespoke custom visual that plots risks dynamically, supports tooltips and click-to-filter — built once, reusable across every Arup Power BI dashboard.

Team

Tapan Parmar

Project context

PMO, project controls and risk management reporting in Power BI — applicable to every Arup project that uses Power BI.

Problem

Power BI defaults are limited; marketplace visuals create dependency, cost and governance risk; risk reporting is inconsistent across projects.

Users

Risk managers · project managers · PMO analysts · leadership teams.

AI role

Visual scaffolding · risk-positioning logic · layout & formatting refinement · interactivity wiring (tooltips, click-to-filter).

Non-AI role

Power BI custom-visual integration · data binding · code review · packaging · ongoing maintenance.

Main value

Reusable Arup IP · consistent risk presentation · no marketplace subscription dependency · fully configurable per project.

Evidence

~5h AI-assisted build · £0 subscription · reusable across future dashboards. Proven · build Production reuse to track

Status

Custom visual built and available for reuse across Arup Power BI dashboards. Live

Reusable pattern

AI-assisted custom-visual / component development for bounded, testable reporting needs.

02 · Project Context

Why This Mattered for Arup PMO Reporting

Risk matrices are a staple of project controls reporting — they show what's high-likelihood, high-impact at a glance. But Power BI's defaults can't produce one, and marketplace alternatives carry cost, vendor risk, and customisation limits.

What the visual needed to do

Why it mattered

Plot risks by Likelihood × Impact

A proper matrix view, scaled 1–5 on each axis — not a table or scatter chart.

Show Risk IDs in-cell

Each risk identifiable at a glance. No clicking required for the headline view.

Surface tooltip detail

Title, description, dominant impact, owner, mitigation, status — on hover, without crowding.

Cross-filter the report

Clicking a risk highlights related data across the entire Power BI report.

Match project branding

Configurable colours per risk classification (Low / Medium / High / Critical).

Why this is a delivery story, not a tools story

Standard Power BI charts could show risk data in tables but couldn't lay it out as a proper matrix. Marketplace visuals added cost, dependency and limited customisability. The right answer wasn't "find a visual" — it was "build the right visual once, and re-use it everywhere."

03 · The Problem

Stuck Between Paid Visuals and Limited Defaults

Risk reporting was caught between two unattractive options. Three problems compounded.

Problem 1 — Cost

Marketplace risk-matrix visuals were mostly paid. Per-seat costs scaled with project headcount and recurred per year.

Problem 2 — Customisation

Marketplace visuals were limited in formatting and not customisable to project requirements (colour schemes, classification thresholds, tooltip fields).

Problem 3 — Reusability

Without an internal asset, every project rebuilt the matrix manually. Inconsistent presentation made portfolio-level views harder to compose.

The real question

How do we build a clear, configurable, reusable risk-matrix visual — without paying for a marketplace product or rebuilding it on every project?

04 · Why AI Was Appropriate

Custom-Visual Development Has a Cost Floor — AI Lowers It

Building a custom Power BI visual from scratch normally takes specialist developer time and a multi-day effort. The build needs scaffolding, positioning logic, styling, packaging — none individually hard, but collectively expensive. AI compresses each step.

Standard Power BI visuals can

Show risk data in tables and basic charts
Support category and value display
Provide basic filtering across the report

Standard Power BI visuals cannot

Render a proper risk-matrix layout with cell zones
Plot risks dynamically by Likelihood × Impact pairs
Allow full design customisation per project
Combine in-cell IDs with rich hover tooltips and click-filter interactivity

The gap AI closes

Translating a business concept into custom-visual logic. AI converted the risk-matrix idea into structured visual code, accelerated layout and formatting work, and surfaced edge-case design decisions that would otherwise have eaten the build window.

05 · What Was Built

A Custom Risk Matrix — Built in Hours, Reusable Forever

A bespoke Power BI custom visual that plots risks by Likelihood and Impact, with configurable styling, rich tooltips and report-wide cross-filtering. Configurable per project; no marketplace dependency.

Visual scaffolding

AI generated the structure for a custom Power BI visual project — capabilities file, render logic, properties pane.

Risk-positioning logic

AI worked out the cell-mapping logic that places each risk by Likelihood (1–5) and Impact (1–5).

Layout & formatting

Matrix layout, cell colour rules, axis labels, marker style and font choices — refined iteratively with AI.

Interactivity wiring

Click-to-filter behaviour and hover tooltips configured so the visual participates in cross-report filtering.

Workflow

How AI plugged into the build

Input

Risk register data — Risk ID · Likelihood (1–5) · Impact (1–5) · tooltip fields (Title, Description, Dominant Impact, Owner, Mitigation, Status)

Define the visual requirement User briefs AI on the matrix concept — axes, cell zones, plotting logic, interactivity expectations.

AI scaffolds the custom visual Project structure, capabilities file, render template, and properties-pane definitions generated as a working starting point.

AI works out positioning logic Cell-mapping for each (Likelihood, Impact) pair, marker placement, classification colour rules.

Wire tooltips & cross-filter Tooltip fields, hover popup styling, click-to-filter integration with the rest of the Power BI report.

Output

A working custom Risk Matrix Visual — matrix display · configurable colours · rich tooltips · click-to-filter · reusable across future Power BI dashboards

06 · How the Process Changes

From Searching the Marketplace to Owning the Asset

The headline is cost — no subscription dependency. The deeper story is reuse, customisation, and report-wide interactivity.

Before · Marketplace

After · AI-assisted custom visual

Paid marketplace visual or limited default chart.

Custom AI-assisted visual, owned by Arup.

Risk display: tables or basic charts.

Matrix-based plotting by Likelihood × Impact.

Customisation: limited.

Fully configurable — colours, labels, markers, classification thresholds.

Cost: dependent on paid visuals.

No paid-visual dependency; reusable for free across projects.

Interactivity: basic filtering only.

Click-to-filter across the report · rich hover tooltips.

Reporting feel: less visual, less project-specific.

Clear, interactive, configurable per project.

Who feels the change

Risk manager

Clear matrix view at every risk review meeting — no more re-creating the same visual from scratch on each project.

Project manager

Ready-made risk visual for project reports — drop in three fields, the matrix populates.

PMO analyst

Configurable component, no marketplace dependency, no licensing chase. Owns the asset, not licence renewals.

Leadership

Consistent risk presentation across the portfolio — one visual language, easier portfolio-level synthesis.

Developer

A reusable custom-visual pattern they can adapt for the next reporting need — not just a one-off matrix.

Bid / account lead

A PMO reporting differentiator — Arup ships a clean, governed risk visual without the marketplace carrying cost.

07 · Value Delivered

Each item is labelled by evidence quality so the case study doesn't overclaim — what's proven on the build, what's still being tracked across reuse.

Time Proven · build

~5 hours of AI-assisted development time for a custom Power BI visual that would normally take days of specialist developer time. Concept-to-working-asset compressed by an order of magnitude.

Quality Proven

Proper matrix layout (not a chart approximation). Configurable per project. Standardises how risk exposure is presented across Arup PMO dashboards.

Cost Proven · cost avoidance

£0 marketplace subscription dependency. Removes a per-seat / per-project recurring cost and avoids vendor lock-in for risk-matrix visualisation.

Risk Proven · governance

AI-suggested code reviewed before publishing. Built using illustrative (not client) data. The asset ships clean — projects bind their own data when they install it.

Governance Proven

Standard code review pattern applied to AI output. Visual versioned and ownership defined. No client information embedded in the published asset.

Commercial Targeted · pattern

Foundation for an internal PMO visual library — a reusable asset library that strengthens Arup's project-controls reporting offer. Adoption to be tracked across projects.

08 · Governance & Responsible Use

Why It's Safe to Reuse on Live Projects

A custom visual is code. The audit story has to cover what was built, with what AI, on what data, with what review.

No client data in development

The visual was built using illustrative risk-register fields, not live project data. The artefact ships clean — projects bind their own data when they install it.

Human review of every plot

AI generated the layout and positioning logic — the risk owner still validates that each risk lands in the right cell before any review meeting.

Code reviewed before publishing

AI-suggested code went through standard review before the visual was packaged for re-use — no AI output went straight into a client report.

Reusable, traceable, defensible

One published visual with clear field mappings — every project that adopts it inherits the same logic, the same audit trail, and the same review pattern.

For client conversations

The same shape that lets this visual be defended on a live engagement is the shape that scales: AI accelerates the build; humans validate the data; the artefact carries no client information until each project binds its own.

09 · The Technical Bit

For Specialists — Components, Validation, and What to Confirm

A deeper-dive section for technical readers. The case study is readable without it; this section adds depth on the build components, validation activities, and what we still need to confirm with the developer.

For technical readers

"AI scaffolds and refines. Humans review the code, version the artefact, and own the maintenance."

Components

Component

Description

Risk register / dataset

Source data containing required fields — Risk ID, Likelihood, Impact.

Optional fields

Risk title, description, dominant impact, owner, mitigation, status — surfaced in tooltips.

Custom-visual logic

Plots risks on a Likelihood × Impact matrix and supports interaction with report filters.

Dashboard layer

Embeds the visual in Power BI reports for leadership or project teams.

Validation activities

Validation area

What to test

Data mapping

Risks appear in the correct Likelihood × Impact cells.

Scoring logic

Numeric or categorical scores are interpreted correctly.

Tooltip accuracy

Tooltip fields show the correct risk information.

Filtering behaviour

Visual responds correctly to report filters and slicers.

Performance

Visual works with realistic risk-register sizes.

Security

No unintended data exposure through tooltips or export.

Version control

Reusable component has clear ownership and release history.

To confirm with developer — before publication

Whether the Power BI Custom Visual SDK was used (or another framework).
Language / framework used (TypeScript, D3 etc.).
Whether AI generated TypeScript, D3, formatting logic or packaging steps — to describe the AI contribution accurately.
Whether the visual is certified or internal-only.
Where source code is stored and version-controlled.
Who owns ongoing maintenance.
Constraints on use with client data.

10 · What Colleagues Can Reuse

Five Patterns to Lift

The reusable lesson is not just "use this risk visual". It's the underlying patterns — applicable to any bounded, testable reporting component.

Risk visual component

Drop into PMO and project-controls dashboards. Map three fields, populate matrix.

Risk dashboard pattern

Standardise how risk exposure is shown across projects — one visual language across the portfolio.

AI-assisted code development

Use the same approach for bounded, testable reporting components — anything where the requirement is specific and the test plan is achievable.

Test plan for custom visuals

Reuse the validation areas above as a baseline test plan for any new custom visual. Improves reliability before reuse.

PMO visual library

Build a library of governed reporting assets — a Risk Matrix today, a CPI/SPI radar tomorrow, a stakeholder heatmap after that.

Reusable assets — now available across Arup

Risk-matrix visual packageDrop into any Power BI report. Map three fields, populate matrix.

Likelihood × Impact logicCell-mapping pattern reusable for other dashboards.

Tooltip configuration templateStandard fields ready to wire up.

Colour-classification setupLow / Medium / High / Critical defaults — overridable per project.

Interactivity patternClick-to-filter wiring re-applicable to other custom visuals.

AI prompt approachThe brief format that produced a working custom visual in hours.

11 · Client-Facing Proposition

How This Becomes an Advisory Offer

The Risk Matrix visual is the first asset in a wider PMO reporting offer — reusable components and AI-assisted dashboard development that replace marketplace dependencies with bespoke, governed Arup IP.

Proposition statement

Arup helps clients build reusable PMO reporting components and AI-assisted dashboard development — replacing marketplace dependencies with bespoke, governed visuals that match each project's data and standards.

Offer components

Project controls dashboard review
Risk visualisation improvement
Custom Power BI components
Reusable reporting templates
Governance and data-quality checks
AI-assisted custom-visual development
Test plans for reusable visuals
PMO visual library curation

Where this lands in the playbook

Maps to Service Line 03 — Programmes & Project Management (P&PM) as programme & project risk reporting, with adoption across MC and BIA dashboards too.

12 · What's Next

From a Risk Matrix to a Risk Intelligence Layer

Each future capability inherits the same approval pattern — AI accelerates, humans approve, and the visual gets richer per release.

Dynamic risk scoring

Compute composite risk scores from multiple factors instead of relying on raw Likelihood × Impact alone.

Automated colour classification

Risk-zone colours derived from project thresholds — no manual setup per dashboard.

Inherent vs residual

Side-by-side comparison of risks before and after mitigation — does the action actually move the risk?

Risk movement over time

Snapshot-on-snapshot comparison — track which risks are escalating and which are de-risking.

Drill-through & export

Click a risk → land on its detailed page. One-click summary export for board packs.

AI-assisted risk insights

Natural-language commentary on the matrix — pattern detection, top-of-mind risks, suggested mitigations.

Built by Tapan Parmar Use cases Project risk dashboards · Design risk reviews · Construction risk monitoring · Organisational risk registers

Final Takeaway

AI transformed the process from searching for a paid visual to creating a custom, interactive, project-specific Power BI risk matrix.

Custom visual + AI-assisted build + Reusable across projects

The visual didn't just replace a marketplace dependency — it introduced customisation, interactivity, and reusability into Arup risk reporting in Power BI.

Continue

CustomRisk MatrixPower BI Visual