AI-driven demand planning that cut dead stock 62% and stockouts 71% — in the same quarter.

A major Tirupur knitwear exporter replaced gut-feel demand forecasting with TextileERP's AI-driven planning — and simultaneously crushed both dead stock and stockouts in a single quarter.

Tirupur, India 1,950 employees 13 weeks rollout Live since August 2025

Headline Results

62%

Less dead stock

₹22Cr → ₹8.4Cr

71%

Fewer stockouts

in 1 quarter

87%

Forecast accuracy

from 54%

₹14Cr

Working cap freed

first year

Knitwear production floor with circular knitting machines

The Challenge

Aged inventory of ₹22Cr sat unsold for over a year while bestsellers stocked out 38 days a year. Forecasting was a merchandiser's intuition, not a system.

The Solution

AI demand forecasting trained on 7 years of historical sales, buyer behavior, seasonality, and macro signals — integrated directly into procurement, production, and capacity planning.

The Outcome

62% less dead stock in 13 weeks.

Challenge

Two mistakes, compounding every season

Every knitwear exporter makes two mistakes: they overbuy what won't sell, and they underbuy what will. For Tirupur Knitwear Group, those two mistakes compounded into ₹22Cr of aged inventory on one side, and 38 stockout days a year on their bestsellers on the other.

Forecasting was done by merchandisers with spreadsheets and decades of gut feel. Some were great. Some were guessing. Either way, nobody could explain why the system worked when it worked — so when it broke, nobody could fix it.

Key pain points

₹22Cr in aged inventory — over 12 months unsold
38 average stockout days per bestseller SKU per year
54% forecast accuracy — below industry average
Planning horizon limited to 4 weeks

Solution

Pattern recognition at a scale humans can't do

We trained TextileERP's forecasting model on 7 years of their transaction data — SKU-level sales, returns, seasonality, buyer-specific patterns, weather, and macro trade signals. The model found patterns no merchandiser had articulated, and blended them with the ones they already knew.

The output wasn't 'the AI says buy X.' It was an interactive planning cockpit where merchandisers could see the forecast, see the reasoning, adjust assumptions, and commit. Humans still decided — but with 100x more information than before.

What we deployed

AI forecasting trained on 7 years of historical data
SKU-level weekly forecasts with confidence intervals
Interactive planning cockpit for merchandisers
Integrated with procurement, production, and capacity

AI Demand ForecastingInventory OptimizationProcurement PlanningCapacity PlanningSales & Operations Planning

Results

Both mistakes — fixed at the same time

The non-obvious result: dead stock and stockouts both dropped. In most companies, fixing one makes the other worse — you stock more to avoid stockouts, which creates more dead stock. AI forecasting let them stock smarter, not more.

₹14Cr of working capital freed up in year one. The CFO reinvested it in automation. The merchandisers finally had time to think about the future, instead of firefighting the past.

62%

Reduction in aged inventory (over 12 months old)

71%

Reduction in stockout days per SKU

87%

Demand forecast accuracy (was 54%)

₹14Cr

Working capital freed up in first year

28%

Improvement in inventory turnover

19 days

Buffer stock reduction on top 200 SKUs

12 weeks

Earlier visibility into capacity constraints

“For 30 years our head merchandiser was our crystal ball. Now the system is the crystal ball, and she's our strategist. That's a better use of her 30 years of expertise.”

Senthil Kumar

Managing Director, Tirupur Knitwear Group