Case Study · Heavy Plant & Equipment

AI Analytics That Bent a 3-Year ROI Decline

How a £250M UK heavy plant distributor used Microsoft Copilot, structured Excel, Python pipelines and Power BI to lift ROAS from 2.4× to 2.75× — the first uplift in three years.

+15%
ROAS uplift
2.4× → 2.75×
~£1M
Incremental margin
Year 1
10–12×
Engagement ROI
Blended Media ROAS
5-year trajectory
First uplift in 3 years
01 · The Client

A UK heavy plant & equipment leader

Revenue
~£250M peak
Employees
~325
Network
15 UK depots
Portfolio
4 trading brands
Model
Sales · Rental · Service · Parts
02 · The Problem

Pressure on two fronts

Market pressure

Revenue down ~15% YoY

High interest rates and a slower housebuilding market squeezed demand. Every pound of marketing spend was under scrutiny.

Data pressure

A 15+ month ERP transformation

NetSuite migration was underway, but commercial leadership needed answers immediately — not after the warehouse was ready.

Disconnected sources
Google Ads
Meta Ads
GA4
OEM Portals
Legacy ERP
NetSuite
Messy middle
No cross-cutting ROI visibility
03 · The Brief
“Break the three-year ROI decline now. Don't wait for the ERP.”
£8M annual media spend£20M marketing envelope4 brands15 depotsNo in-house data teamNo unified ROI visibility
04 · Solution Architecture

Microsoft Copilot for Commercial Analytics

A semantic foundation under the AI — not the other way around.

Step 1
Data Sources
Google Ads, Meta Ads, GA4, OEM portals, legacy ERP, NetSuite.
Step 2
Python Pipelines
Scheduled API pulls, transformations and refresh logic.
Step 3
Structured Excel Layer
AI-readable source of truth with strict conventions and clean structure.
Step 4
Context Engineering
Business definitions, brand taxonomy, depot structure and product hierarchy.
Step 5
Copilot + Power BI
Plain-English Q&A plus selective dashboards where visuals beat conversation.
AI isn't the hard part. The foundation under it is.
05 · Copilot Interface

Ask commercial questions in plain English

Commercial Copilot
Connected · 6 sources · Last refresh 04:00 GMT
Live
What's our margin-weighted ROAS across the four brands this quarter?
Insight

Brand-protect campaigns are consuming budget across all four brands, but margin-weighted return is stronger in selected non-brand Shopping categories.

Recommended action

Reallocate 12–18% of spend toward high-margin categories with available stock.

Confidence 92%Source · margin_weighted_roas_v3.xlsx
Suggested next:
06 · Timeline

Value from week one

Week 1
First pilot insight from a single static Excel export
Insight before infrastructure
Week 2
Google Ads, Meta and GA4 snapshots connected
Week 4
Dynamic Python pipeline live with daily refreshes
Live pipeline
Week 8
Depot-level sales and lead reporting layer
Week 12
NetSuite stock, margin & depot inventory integrated
07 · Findings

What the AI analyst found

Channel fatigue isolated

40% of the ROI decline traced to specific campaign types and audience segments.

Impact · Focused remediation, not blanket cuts.

Brand-protect overspend

Budget reallocated to higher-margin non-brand Shopping campaigns.

Impact · Spend follows margin, not vanity.

Out-of-stock ad waste

Live NetSuite sync blocked ads for unavailable SKUs.

Impact · Wasted impressions eliminated.

Margin-weighted ROAS

Revenue-heavy brands were not always margin-heavy brands.

Impact · Re-prioritised portfolio investment.

Depot-region skew

National bidding was misaligned with concentrated regional demand.

Impact · Geo-tuned bid strategy.

Faster optimisation cycles

Underperformers were killed in days, not quarters.

Impact · Decision velocity 10× faster.
08 · Impact

The ROI curve finally bent the right way

2.4× → 2.75×
Blended media ROAS
+£4.2M
Incremental attributable revenue
~£1.05M
Incremental gross margin
10–12×
Return on engagement
Scenario comparison
Decline scenario Actual
A +0.52× ROAS gap vs. the do-nothing scenario.

On £8M annual media spend, the uplift translated into approximately £4.2M incremental attributable revenue and ~£1.05M incremental gross margin in Year 1.

09 · Bigger Picture

Faster decisions today. De-risked foundation for tomorrow.

While the NetSuite programme continued its 15+ month delivery cycle, the commercial team had a working AI analyst from week 1 — designed to plug into NetSuite when ready.

Immediate value

First insight delivered in week 1.

Operational adoption

19 authenticated users across marketing, sales and operations.

Future-ready foundation

Semantic layer ready for NetSuite integration when the ERP stabilises.