Havish M Consulting
Customer Case Study · Supplier Evaluation Matrix
Consumer Durables Manufacturer × Havish M Consulting

Choose suppliers on evidence, not on memory

The AC manufacturer builds room air conditioners and their components at scale, with hundreds of suppliers behind every unit. HMC built a supplier evaluation workspace for the procurement team: a weighted scoring matrix the category heads agree on, a 3D view of component risk, supplier segmentation, a volume split optimiser, a disruption simulator, and GenAI that reads and compares quotes.

ClientConsumer Durables Manufacturer FunctionProcurement · Strategic Sourcing ThemeSupplier Evaluation Matrix
Supplier evaluation cycle
0days
▼ from about 3 weeks
Incoming defect rate, key suppliers
0% lower
▼ scorecards drive improvement plans
Material cost saving, evaluated categories
0%
landed-cost comparison and volume splits
Single-source critical parts
0left, from 3
second sources developed

Figures indicative; validated metrics available on request.

The Challenge

Sourcing decisions on scattered data

Quality lived in the QMS, delivery in the ERP, prices in emails and spreadsheets. Supplier reviews were long, subjective and hard to compare across categories.

Data in silos

Defect rates, on-time delivery, prices and certificates sat in different systems and files.

Subjective selection

Rankings depended on who was in the room; criteria and weights changed from review to review.

Concentration risk

Critical parts depended on one supplier or one country, with little view of the exposure.

Quotes hard to compare

Different currencies, Incoterms and payment terms made true landed cost slow to work out.

HMC Approach

Agree the rules, then let the data rank

A transparent method the category heads sign off, applied the same way to every supplier.

Select a step to see what happened.

Discover · How suppliers were judged

Interviews with category managers, quality and stores.

  • Data sources for quality, delivery, price, capacity and certificates mapped.
  • Past sourcing decisions and their outcomes reviewed.
  • Critical components and single-source exposure listed.

Deliverable: Criteria catalogue and data map.

Design · A matrix everyone signs

Eight criteria, clear formulas.

  • Scoring rules from raw metrics (defects per million, on-time-in-full, price index).
  • Weights agreed with pairwise comparison (AHP) and a consistency check.
  • Segmentation rules: partner, develop, maintain, exit.

Deliverable: Signed-off evaluation framework.

Build · The sourcing workspace

From scorecard to sourcing decision.

  • Automated monthly scorecards from ERP and QMS extracts.
  • Volume split optimiser and disruption scenarios.
  • GenAI quote reader and supplier review letters.

Deliverable: Working tool and Power BI views.

Adopt · Quarterly reviews on the tool

Same data in every meeting.

  • Supplier reviews run from the scorecard.
  • Improvement plans with targets for the "develop" quadrant.
  • New RFQs compared on landed cost.

Deliverable: Review cadence and playbook.

Scale · Every category

From pilot to standard practice.

  • Remaining categories onboarded.
  • Second-source development tracked to closure.
  • Scorecards shared with suppliers each quarter.

Deliverable: Roll-out plan and governance.

The Solution

Six tools, one supplier base

A synthetic supplier base for a 1.5-ton inverter split AC: 8 component families, 12 anonymised suppliers across India, China, Thailand and Vietnam. Everything below is live: change the weights and every module updates.

✦ Live GenAI

Templates

Optional. Every ✦ button already works on built-in templates. Add an OpenAI key to read any quote you paste, ask the supplier base anything and have AI write review letters.

The key goes only from this browser to OpenAI and stays in memory unless you tick "remember". Only the synthetic data on this page is sent.

01 · 3D component explorer

Where is the supply risk inside the product?

An exploded split AC, every component family coloured by supply risk. Drag to rotate, scroll to zoom, click a part (or pick it from the list) to see who supplies it.

Low riskMediumHigh

Component families

Simplified 3D model for illustration. Risk = single-sourcing, import share, supplier scores, lead time and criticality.

02 · Evaluation matrix

Agree the weights, then let the scores rank

Twelve suppliers on eight criteria, each scored 0–100 from raw metrics. Set weights with sliders or presets, or switch to pairwise comparison (AHP) the way a sourcing committee would. Rows re-rank live.

Weights

0–4040–6060–8080–100Click a supplier for its 360 view
03 · Supplier 360 & segmentation

Who to grow, who to develop, who to replace

Each supplier placed by weighted performance and business impact. Play the quarters to watch suppliers move as improvement plans land. Click a supplier for its 360 card.

04 · Volume split optimiser

How much should each supplier get?

Pick a component. The optimiser splits annual volume across qualified suppliers by landed cost, cost of poor quality and supply risk, within capacity, and keeps a minimum share with a second source.

Component

Minimum second-source share
Cost ⟷ resilience

Effective cost per unit = landed price + cost of poor quality (defects per million × 40 × price) + a supply-risk premium scaled by the resilience setting.

05 · Disruption simulator

What happens when supply breaks?

Trigger an event and see which suppliers and parts are hit, whether inventory covers the gap, the cost impact and the recommended switch.

06 · Procurement copilot

Read quotes, write reviews, ask questions

Compare quotes in any format on true landed cost, draft a quarterly review letter from the scorecard, or ask the supplier base a question in plain English.

Templates

Quotes for a 1.5 TR inverter rotary compressor

Landed cost assumes USD ₹84, EUR ₹91, CNY ₹11.6, THB ₹2.45, sea freight ₹120 per unit from FOB ports, insurance 0.5%, import duty 10% on CIF, inland freight ₹40, and 10% a year cost of money for payment terms. Illustrative only.

Normalised comparison

No quotes yet. Load the samples or paste a quote.

Synthetic quotes and suppliers. Template mode reads the sample formats; with a key, the AI reads any quote text. Buyers verify every extracted value.

Impact Dashboard

Better suppliers, fewer single points of failure

Scorecards and improvement plans narrowed the spread of defect rates, and second sources closed most of the single-source gaps.

Incoming defect rate by component family

Violin plot of monthly defects per million across suppliers

Who supplies what

Arc diagram linking component families to suppliers (arc width = share)

Single-source partMulti-source partNew second source

Figures indicative; validated metrics available on request.

Before vs After

How suppliers are evaluated now

DimensionBeforeAfter
Supplier dataQMS, ERP, emails and spreadsheetsOne monthly scorecard per supplier
Criteria and weightsChanged from meeting to meetingEight criteria, weights agreed by AHP
SegmentationInformalPartner, develop, maintain, exit, reviewed quarterly
Volume allocationHistorical sharesOptimised for cost, quality and risk within capacity
Disruption responseFirefightingPre-planned switches with cover and cost known
Quote comparisonManual, days per RFQLanded cost in minutes
Tools & Tech Used

The sourcing analytics stack

ERP and QMS extracts monthly scorecards AHP agreed, consistent weights Python scoring and optimisation Three.js 3D component risk view Large language model quote reading and letters Power BI category dashboards
Key Takeaways

What any procurement team can borrow

01

Agree the weights before the scores

When the committee signs the weights first, the ranking is accepted, not argued.

02

Compare on landed cost

The cheapest quote on paper is often not the cheapest at the plant gate.

03

Design for a second source

Know the cover days and the switch before the disruption, not after.