Distribution Resource Planning: Beyond the Forecast

What Distribution Resource Planning Is (and What It Isn't)

Distribution resource planning (DRP) is the system that decides how much product leaves your distribution center (DC) toward each store, when it leaves, and in what quantity. It is not a forecast model. It is not a purchase order (PO) system. It is the mechanism that translates a demand signal into real inventory movement across a network of points of sale.

The distinction matters. MRP (Material Requirements Planning) was designed for manufacturing: it calculates input requirements from a production plan. DRP does something different — it coordinates inventory flows between nodes of a distribution network based on the expected demand at each point of consumption. In retail, that point of consumption is the shelf, not the DC.

DRP doesn't tell you what your store is going to sell. It tells you what you need to move — and when — so the product is there when the customer arrives.

A well-executed DRP connects three variables in real time: available inventory at each store, in-transit inventory, and projected demand by point of sale. When those three variables are aligned, the system replenishes correctly. When they aren't — and in most Latin American retailers they aren't — the system replenishes with noise.

DRP vs. Forecast: The Confusion That Costs Retailers Millions

73% of retailers in Latin America have overstock and stockouts at the same time, in the same week. That is not a forecast problem — it is the exact signature of a DRP that is moving numbers instead of moving product.

The most costly misconception in the industry is this: improving the forecast does not improve replenishment if the DRP that translates that forecast into orders is broken. You can have a predictive model with an 8% MAPE and still have 4.1 days of monthly stockouts on your top 200 SKUs — because the problem is not in the prediction, it's in the execution of the flow.

The forecast says how much will be sold. DRP says what needs to move. They are separate layers. Treating DRP as a forecast problem is like changing the map when what's broken is the engine.

How DRP Works in a Retail Distribution Network

DRP operates in a cascade. Demand at the store generates a replenishment need. That need flows up to the DC. The DC calculates whether it has enough stock to cover all stores within the planning horizon. If it doesn't, it generates a purchase order (PO) to the supplier.

In theory, the flow is clean. In practice, each link in that chain introduces distortion: lead times that don't reflect reality, safety stock parameters calculated three years ago, system inventory that doesn't match what's physically on the shelf. When those distortions accumulate, the result is not an imprecise forecast — it's a DC full of product that isn't reaching where it needs to go.

---

Why Traditional DRP Fails in Modern Retail

The classic DRP model was designed for stable distribution chains: predictable lead times, low-variance demand, reliable suppliers. That environment does not exist in Latin American retail. And that incompatibility has concrete costs.

There are three structural reasons why traditional DRP generates overstock in some stores and stockouts in others simultaneously:

1. Parameters are static in environments where demand changes every week, causing the system to over-order for stores where turnover has dropped and under-order for stores where it has risen.

2. Store-level inventory visibility is partial or delayed, so the DC makes dispatch decisions based on data that is 24 to 72 hours old — enough for a store to experience two stockouts before receiving replenishment.

3. The model does not differentiate SKUs by demand behavior, treating a high-turnover product with stable demand the same as a seasonal product with sales concentrated in three weeks of the year.

The Problem of Static Parameters in Volatile Demand Environments

The safety stock at most retailers is calculated with a formula that assumes stable variance. In Latin America, that variance is not stable. Currency volatility affects prices and demand. Atypical seasonal events — Buen Fin, 11/11, payday cycles, local patron-saint festivals — generate spikes that a static parameter cannot capture.

When the parameter is not updated, the system compensates by over-ordering in aggregate. The result: excess inventory on average, stockouts at the stores with the highest variance.

Centralizing Without Visibility: The DC-as-Bottleneck Trap

A DC that cannot see real-time inventory across its 80 stores is not coordinating a network. It is managing a queue of requests. The difference is critical: in queue mode, the DC responds to what stores report. In network mode, the DC anticipates what stores are going to need.

Retailers operating with inventory reports that are 24 hours or more delayed are, by definition, always replenishing late. DRP becomes a reactive system disguised as planning.

---

The Real Cost of Poorly Executed DRP: Calculating in Dollars

The damage from dysfunctional DRP does not appear on a single P&L line. It is distributed across tied-up capital, lost sales, and margin erosion. To quantify it, you need three calculations, not one.

Formula 1: Cost of Overstock as a Percentage of Working Capital

The capital tied up in inventory carries a real financial cost — not just an accounting one. It is the rate you charge that money for not moving.

Financial cost of overstock =


Excess inventory ($) × Annual cost of capital rate × (Days of excess / 365)

If you have USD 8M in average inventory and 22% is in excess (above your target coverage), that is USD 1.76M tied up. At an annual cost of capital rate of 14%, that excess is costing you USD 246,400 per year in financial cost alone — before counting obsolescence, shrink, and warehouse space.

Formula 2: Lost Sales from Stockouts Converted to EBITDA

Annual lost sales =


Total sales ($) × Stockout % × % of demand not substituted internally


EBITDA impact =


Lost sales × Average gross margin

A retailer with USD 400M in annual sales operating with a 3.2% stockout rate across its top 500 SKUs is losing USD 12.8M in sales. If 60% of that demand is not recovered internally (the customer buys at another store or switches brands), the effectively lost sales total USD 7.68M. At a 28% gross margin, the direct EBITDA impact is USD 2.15M per year.

Numerical Example: 80-Store Retailer, 12,000 SKUs

| Parameter | Base Value | Annual Impact |

|---|---|---|

| Average excess inventory | USD 3.2M | Financial cost: USD 448K |

| Stockout on top SKUs | 2.9% on USD 280M | Lost sales: USD 4.85M |

| Unrecovered demand | 55% of stockout | Effective loss: USD 2.67M |

| Gross margin | 26% | EBITDA impact: USD 694K |

The combined damage — tied-up capital plus lost EBITDA — exceeds USD 1.1M annually in an operation that, from the outside, "works." Nobody accounts for it this way because overstock appears as an asset and stockouts appear as forecast variance.

---

The 4 Data Errors That Make Your DRP Lie

DRP is only as good as the data feeding it. And at most retailers in the region, that data contains systematic errors that distort every replenishment decision.

Fictitious Lead Times That Inflate Safety Stock

The lead time in your system was probably entered more than 18 months ago. Since then, your supplier changed its logistics operation, your distribution routes changed, and reality is 1.4 days longer — or shorter — than the parameter. That difference, multiplied by the average daily demand across 12,000 SKUs, generates safety stock errors that accumulate silently.

Smoothed Demand That Hides Real Store-Level Seasonality

Systems that average historical demand to calculate replenishment are making a design error. The D1 store in Chapinero does not have the same payday-cycle pattern as the D1 store in Suba. A model that averages those two signals replenishes incorrectly at both. Distribution resource planning at the individual store level — not by cluster or region — is the only way to capture that variability without over-ordering in aggregate.

Phantom SKUs: System Inventory That Doesn't Exist on the Shelf

In retailers with high turnover and manual store operations, the discrepancy between system inventory and physical on-shelf inventory ranges from 8% to 17% of active SKUs. DRP replenishes based on what the system reports. If the system says there are 12 units in the store and in reality there are 0, the system will not generate a replenishment order — and the stockout will persist until the next physical count.

The Fourth Error: Replenishment Frequencies That Don't Match SKU Turnover

A SKU with daily turnover that is replenished once a week will stockout regularly, regardless of how good the forecast is. Aligning replenishment frequency to the actual turnover of each SKU at each store can reduce stockouts by 1.8 to 2.6 percentage points without increasing average inventory.

---

How DRP Should Work in a Competitive Retailer

An execution-oriented DRP does not start from the central forecast. It starts from actual consumption at each point of sale and works backward.

From Push to Pull: Replenishment Based on Real Store Demand

The traditional push model centrally decides how much to dispatch to each store. The pull model lets the store — or more precisely, the store's actual consumption — generate the replenishment signal. The difference in outcome is concrete: pull models reduce store-level overstock by 15% to 23% while maintaining the same service level, because they don't send product the store hasn't demonstrated it needs.

Store-level demand planning is the starting point for this model. Without a granular demand signal at the point-of-sale level, any DRP will add noise as it moves up to the next level.

SKU Segmentation by Behavior, Not Just Category

A high-turnover SKU with stable demand requires a different replenishment policy than a mid-turnover SKU with high variance. Treating them the same generates overstock on the first and stockouts on the second — simultaneously, in the same store. Effective multi-store inventory management segments SKUs by observed behavior: sales velocity, variability, supplier lead time, and sensitivity to seasonality.

The DC's Role as an Enabler, Not an Accumulator

The DC should not be the place where inventory waits to be needed. It should be the most efficient transit node possible between the supplier and the shelf. When the DC operates as an accumulator — holding 21 days of coverage for SKUs that the store sells in 4 — the capital that should be turning over is sitting still, accruing interest.

---

DRP and Service Level: The Metric That Connects Logistics to Revenue

Retail service level has two faces. Fill rate measures what percentage of the units requested by stores arrive complete from the DC. On-shelf availability measures what percentage of the time the product is available on the shelf when the customer looks for it. These are distinct metrics and both matter.

The most common mistake is reporting the DC's fill rate as a proxy for actual service level. A DC with a 97.3% fill rate can be supplying stores with a 6.1% stockout rate if DRP is miscalibrated — because the problem is not whether the DC ships complete orders, but whether it ships the right product, to the right store, at the right time.

To understand the relationship between service level and lost sales, the metric you need is not the aggregate fill rate. It is on-shelf availability by SKU by store by day.

How to Measure Real Service Level (Not What the System Reports)

Real On-Shelf Availability =


1 - (Days with stockout × Estimated lost sales units) / Total potential sales

If your top 100 SKUs show an average stockout of 2.3 days per month per store, and each day of stockout represents 60% of that SKU's expected daily sales, your real on-shelf availability is significantly lower than what the DC's fill rate report shows.

The Direct Relationship Between Stockouts and Permanent Customer Loss

A stockout doesn't lose a sale. It loses a customer in 31% of cases — because that customer chooses a substitute brand and doesn't return to the original product. In categories with high brand loyalty (beverages, personal care, dairy), that percentage rises to 44%. The cost of a stockout is not the unit that wasn't sold. It is the lifetime value of the customer who stopped trusting your store to find what they're looking for.

---

Checklist: Signs Your DRP Is Destroying Capital Today

Answer these questions. Every "yes" is an active capital leak:

• Does your DC inventory cover more than 14 days for high-turnover SKUs? Anything above your lead time plus buffer is tied up for no reason.

• Have your safety stock parameters gone more than 6 months without an update? They are calculated for an operation that no longer exists.

• Is your DC fill rate above 95% but your stockout rate above 2.5%? The problem is not supply — it's distribution within the network.

• Does the discrepancy between your system inventory and physical inventory exceed 8%? Your DRP is replenishing based on inventory that doesn't reflect shelf reality.

• Do you have stores with overstock and stores with stockouts on the same SKU in the same week? That is the signature of a DRP that doesn't redistribute — it only accumulates.

• Are lead times in your system not automatically reconciled against actual lead times? Every day of difference is costing you unnecessary safety stock units.

• Does your replenishment use the same frequency for SKUs with different turnover rates? You are over-serving the slow movers and under-serving the fast ones.

If you answered "yes" to three or more of these questions, your DRP has a structural problem — not a data problem.

---

Is Your DRP Managing Inventory or Managing Excuses?

There is a question worth asking this week: when your team explains a stockout, do they explain why it happened — or why it was inevitable?

The difference is not semantic. A DRP designed to optimize execution produces fewer stockouts because it adjusts parameters before the problem occurs. A DRP designed to produce reports explains the problem after the customer has already left.

At Cruz Verde, after implementing a replenishment logic based on actual consumption by point of sale, GMROI rose 16.5%, average inventory dropped USD 2.1M (-12.3%), and sales grew 2.4%. That is not forecast optimization. That is a DRP that stopped managing excuses and started managing flow.

The question is not whether your forecast is precise enough. The question is whether your DRP is designed to move the right product to the right place before the customer looks for it — or to explain, with elegant numbers, why it didn't arrive in time.

---

Frequently Asked Questions

What is the difference between DRP and MRP in a retail context?

MRP (Material Requirements Planning) was designed for manufacturing: it calculates input requirements from a production plan. DRP (Distribution Resource Planning) coordinates inventory flows between distribution nodes and points of sale based on actual consumer demand. For a retailer, MRP does not apply directly — DRP is the layer that translates the in-store demand signal into replenishment orders from the DC and from suppliers.

How do I know if my problem is a DRP problem or a forecast problem?

If you have stockouts and overstock on the same SKU at the same time — across different stores or even in the same store — the problem is distribution, not prediction. A pure forecast problem generates uniform errors in the same direction. A DRP problem generates dispersion: excess at some nodes, shortage at others, with the same aggregate demand level.

How much capital can a well-executed DRP free up at a mid-sized retailer?

It depends on your starting point, but the formula is straightforward: take your average excess inventory (what exceeds your target coverage) and multiply it by your cost of capital rate. A retailer with USD 5M in excess inventory and a 13% annual rate is paying USD 650,000 per year to have product sitting idle. Reducing that excess by 30% frees up USD 1.5M in working capital and eliminates USD 195,000 in annual financial cost.

Does DRP replace the forecasting system or complement it?

It complements it — but not in the direction most people assume. The forecast feeds DRP with the expected demand signal. DRP converts that signal into inventory movement decisions. Improving the forecast without improving DRP produces more accurate plans that execute just as poorly. The correct sequence is: first ensure the DRP can execute well, then improve the quality of the signal feeding it.

What metrics should I use to measure whether my DRP is working correctly?

Three metrics, in order of priority: on-shelf availability by SKU by store (not the DC's fill rate), days of excess inventory by network node, and sales-weighted stockout rate (not by number of SKUs). The DC fill rate is a process metric. The three above are outcome metrics — and they are the ones that connect directly to revenue and working capital.