Work

Case study deck · 9 slides

Agri & dairy

Procurement across 8 collection centers. One 30-minute refresh.

A D2C dairy brand scaling across 3 Indian cities

Process Redesign & BuildAdoption & Enablement
01 / 09

The challenge

What was broken

Procurement ran on eight spreadsheets, one per collection center, stitched together by hand every month-end, two to three days of rework and still not clean. A single mistyped BMC code could route 400 litres to the wrong vendor. With 20,000+ farmers across three cities, the variance that fell through the cracks showed up as missing farmer payments, not a line in a report.

Hours spent reconciling procurement data across 8 BMCs per month-end cycle, before the rebuild
02 / 09

The approach

Scored on the Durable AI Index

We mapped how milk actually moved before touching a dashboard, and found three centers used a different litre-rounding convention than the rest. We built a connector pulling BMC and vendor data into one table, added variance flags that fire the moment a mismatch appears, then built the view procurement checks daily instead of the one they dread monthly.

DURABLE AI INDEXImpact90Feasibility84Stickiness8642at startafter redesign
High impact, low stickiness. The gap was the workflow, not the model.
03 / 09

What we built

The system, not just the model

We redesigned the workflow around the model, then built the pieces that make it run every day.

  1. 1

    ERP/IoT connector unifying BMC intake logs and vendor invoices

  2. 2

    Variance flags on any center-vendor mismatch, in near real time

  3. 3

    Farmer-level tracking across all 20,000+ farmers, every VLCC

  4. 4

    Power BI dashboard replacing the eight-file manual reconciliation

04 / 09

Inside the build

The eight-center reconciliation view

One live view of milk moving from center to vendor to farmer payment, with the points where litres used to go missing flagged instead of buried.

Refresh time: 30 minRework cycles: ZeroFarmers tracked: 20,000+

Tools & systems

ERP/IoT connector layerPower BI dashboardBMC-level data capturevariance-flagging logicvendor invoice matching

Collection centers

  • BMC intake logging
  • Litre-rounding normalization
  • Center-vendor variance check
  • Daily volume roll-up

Vendors

  • Invoice ingestion
  • Invoice-lag adjustment
  • Vendor-wise variance flagging
  • Payment release trigger

Farmers

  • VLCC-level litre attribution
  • Farmer payment reconciliation
  • Churn-risk flagging
  • Network-wide farmer view
Highlighted nodes are where the new visibility or incentive was added.
05 / 09

Results

What good looks like

30 mins

End-to-end data refresh, down from 2–3 days, with zero rework cycles

Near-zero

Reconciliation variance, with faster and cleaner farmer payments

20,000+

Farmers tracked on a single dashboard across all collection centers

0 crashes

Replaced the cross-center Excel that failed every month-end

END-TO-END DATA REFRESH60Before (2–3 days)0.5After (30 mins)
06 / 09

How it stuck

Adoption is the deliverable

The two-day, phone-off month-end ritual disappeared after one cycle, not by mandate but because the dashboard was simply faster. What kept people in it was the flagging: a mismatch now surfaces the same week instead of the same quarter, so procurement stopped reconciling and started investigating.

050100month 6This engagementTypical pilot
Weekly active use of the dashboard by procurement and finance, months 1-12 post go-live
07 / 09
We used to close the books not knowing if the number was right. Now if a center's off by two hundred litres, I know which one, and I know it Tuesday.

Procurement operations lead, dairy engagement

08 / 09

Want this for your business?

Book a free 30-minute AI opportunity assessment. You will leave with at least one concrete idea.

09 / 09