The situation
Rangmanch had a distribution problem nobody could see on a spreadsheet.
On paper, their allocations looked balanced. Each store received inventory roughly proportional to its size and historical sales. The buying team had done a reasonable job building the range. The numbers, in aggregate, were not embarrassing.
But on the floor, something was consistently wrong.
Store managers in Tier 1 cities were reporting overcrowded fixtures — too many styles competing for the same rail space, some never making it to the floor at all. Stores in smaller markets were receiving styles that simply did not match their customer or their physical setup. Products were arriving and sitting in stockrooms rather than reaching shoppers.
The allocation system was sending inventory based on what the data said stores should receive. It had no way of knowing whether stores could actually display what they were being sent.
The core problem
Rangmanch's allocation logic at the time was sales-history-first. High-performing stores got more. That logic is not wrong — but it is incomplete.
A store's sales history tells you how much it has sold. It does not tell you how much it can show. A 1,200 sq ft store in Jaipur with 40 dedicated style slots for kurtas cannot meaningfully display 65 styles. When 65 styles arrive anyway, the store manager makes the decision about what goes on the floor. The rest sits in back. The brand has effectively made an inventory investment that is generating zero floor return.
Multiply this across 194 stores, across a season with hundreds of styles, and the problem is not a minor inefficiency. It is a structural mismatch between what the planning system thinks is happening and what is actually happening on the ground.
Two specific consequences showed up clearly in the end-of-season review:
Phantom sell-through. Styles that were allocated to stores but never made it to the floor showed up in the sell-through data as available inventory that did not sell — pulling down the ROS figure for those styles and making them look like underperformers. In the next season's planning cycle, those "underperformers" received reduced allocation. The system was penalising good products for a logistics failure it had itself created. Wasted markdown spend. End-of-season clearance was being triggered by high stock levels in stores that had never displayed the inventory in question. Rangmanch was marking down products to clear stock that customers had never seen.What changed with Kyros
When Rangmanch moved their allocation to Kyros ahead of the Winter 2024 season, the first significant change was the introduction of planogram data as a hard constraint — not a soft input, not a reference number, but a ceiling the allocation engine cannot exceed.
Each store's display capacity was entered by category: how many style slots were available for kurtas, for dupattas, for fusion western, for occasion wear. The numbers came from the store ops team and took about two weeks to compile across the network. Once in the system, they became inviolable.
The allocation engine now works like this:Before any quantity is assigned to a store, the engine checks how many styles are already planned for that store in that category. If the planned allocation would push the store past its display capacity, the excess is not sent. It is redistributed — either to stores that have capacity, or flagged for a second allocation wave once the first wave has cleared floor space.
For Rangmanch, this meant that a store with 40 kurta slots received at most 40 styles. Not 58. Not 62. Forty — and the engine determined which 40 based on that store's historical attribute performance, its cluster's regional preferences, and its grade.
This last part mattered as much as the capacity ceiling.
Store grades and regional suitability
Rangmanch's stores had been informally graded for years — everyone knew which were the flagship locations and which were the smaller-format stores. But that grading had never been formalised into the allocation logic. High-grade stores got more inventory because planners knew they should. Low-grade stores got less. The reasoning was sound. The consistency was not.
Kyros formalised the grade structure into three tiers (A, B, and C) calculated quarterly from actual performance data — not planner memory. The allocation split across the network was set at 50% of available units going to A-grade stores, 35% to B, and 15% to C. These percentages were adjustable by season, but they were explicit and documented.
The second layer was regional suitability. Rangmanch had been sending a largely uniform range across geographies, with minor adjustments made by planners on instinct. The Kyros allocation engine used cluster-level attribute performance data to adjust which styles went where. Embroidered occasion wear performed differently in the Delhi NCR cluster than in the Bengaluru cluster. Cotton everyday styles had a different velocity profile in Maharashtra than in Rajasthan. The engine weighted these signals and adjusted style selection by store accordingly — not overriding the planner's range decisions, but distributing the approved range in a way that reflected where each product was most likely to perform.
The numbers, one season later
The Winter 2024 season was the first full season running Kyros allocation across Rangmanch's network. The end-of-season review compared it against the equivalent period the previous year.
| Metric | Winter 2023 | Winter 2024 | Change |
|---|---|---|---|
| Styles never displayed (stockroom inventory) | 18% of allocated units | 6% of allocated units | −67% |
| First allocation accuracy (no transfer needed) | 61% | 79% | +18pp |
| Inter-store transfers in-season | 340 movements | 198 movements | −42% |
| End-of-season markdown spend | ₹1.4 crore | ₹91 lakh | −35% |
| Average sell-through across network | 58% | 71% | +13pp |
The transfer reduction was the second meaningful signal. Inter-store transfers are expensive — in logistics cost, in planning time, and in the weeks of lost selling opportunity while inventory is in transit. The drop from 340 to 198 movements was not because the business had fewer inventory imbalances. It was because those imbalances were being corrected at the allocation stage rather than discovered mid-season.
What the planning team said
The head of merchandising at Rangmanch made one observation that captured the shift better than the data:
"Before, we were allocating what we thought stores could sell. Now we're allocating what stores can actually show. Those are not the same number, and for four years we had no way to tell the difference."The second change she noted was in the quality of the post-season data. Because the phantom sell-through problem had been largely eliminated — fewer styles sitting in stockrooms and skewing the numbers — the performance signals from Winter 2024 were cleaner than any previous season's data. When the team started planning the Summer 2025 range, they were working from a more accurate picture of what had actually sold, not what had been sent.
That feedback loop — cleaner allocation producing cleaner data producing better planning — is the compounding effect the system is designed to create.
What this means for growing brands
The planogram constraint problem is not unique to Rangmanch. It is structural to any brand that has grown its store network faster than its planning infrastructure.
When a brand has ten stores, planners can account for each store's physical reality through direct knowledge. When it has a hundred, that knowledge becomes impossible to maintain at scale. The allocation system fills the gap — but only if it has access to the right constraints.
Most allocation systems do not. They optimise from sales data toward a target quantity, with no mechanism to enforce the physical limits of the stores they are distributing to. The result is the same pattern, repeated across brands of different sizes: overcrowded fixtures in some stores, mismatched inventory in others, and a layer of avoidable cost and complexity that gets treated as a normal cost of doing business.
It is not normal. It is a planning infrastructure problem — and it has a solution.
Kyros — Built for the brands that are ready to stop guessing.