Warehouse management systems and omnichannel ERP software are sold as alternatives, which is strange, because in a working omnichannel operation they do different jobs. The confusion costs retailers real money. I have seen a business buy a WMS it did not need because the ERP demo made warehouse functions look weak, and another run three years without a WMS while its picking accuracy collapsed. Both mistakes come from not understanding the boundary.

Why the two systems overlap

Every ERP has a warehouse module. Every WMS has inventory and order functions. The marketing material on both sides claims the other is unnecessary, and the overlap is real. The ERP's warehouse module can handle receipts, putaway, picking lists and dispatch. The WMS can talk to channels and feed stock levels out. So where is the line?

The line is speed and granularity. The ERP thinks in stock lines and financial postings. The WMS thinks in bins, pallets, waves and scans. When you run a few hundred orders a day, the ERP's module is fine. When the warehouse becomes a machine, processing thousands of lines with tight service levels, the ERP becomes the bottleneck. The UK Warehousing Association publishes exactly this kind of operational reality from its member network: the businesses that push high throughput all converge on the same conclusion, which is that warehouse execution is a specialist problem.

What the ERP should own

The ERP should own the single version of stock truth. That means the sellable quantity per SKU per channel, the financial value of inventory, purchasing, receipts against purchase orders, stock adjustments and the cost of goods sold. When the finance team asks what stock is worth, the ERP answers. When the buying team asks what to reorder, the ERP answers.

This is the record that keeps stock levels honest across every channel, and it is why the ERP, not the WMS, should sit at the centre of your omnichannel architecture. If the WMS becomes the stock authority and the ERP merely receives summaries, you have two truths and a reconciliation problem. Decide which system is the authority before you connect anything.

What the WMS should own

The WMS should own everything that happens physically inside the four walls: where each unit is located, how it is picked, how waves are planned, how labour is assigned and how dispatch is sequenced. It tracks bins and scans, not just SKUs. Its output to the ERP is simple and powerful: stock moved, stock received, stock counted.

That division sounds obvious but it is violated constantly. Retailers ask the WMS to hold the sellable stock figure and publish it to channels directly, cutting the ERP out of the loop for speed. The result is that purchase orders, margins and finance reports drift away from reality. The single stock record guide article shows where that drift shows up: in margin, because margin is calculated from a stock record that no longer matches the warehouse.

When you genuinely need a WMS

Use the volume and complexity test rather than the turnover test:

  • Order lines per day. Above roughly 1,500 to 2,500 lines, ERP picking screens start to slow teams down and scan based workflows become essential for accuracy.
  • Picking complexity. Mixed SKU cartons, batch and expiry tracking for food, or picking from multiple zones push the ERP module past its design.
  • Multiple locations. Several warehouses, or shop stock as fulfilment inventory, need location level control the ERP cannot give you cleanly.
  • Labour management. If you manage picking performance, wave planning or shift allocation, that is WMS territory.

The UK Warehousing Association's member directory is a useful way to see how professional 3PL and distribution operations are structured, and every serious one runs a WMS. That is not a fashion statement. It is because their service levels depend on per location accuracy that an ERP cannot deliver.

Making them work as one

When you run both, the integration quality decides whether you get the best of both or the worst. The ERP sends orders and purchase receipts. The WMS sends back confirmations, stock levels and adjustments. The integration must be near real time and bidirectional, and the failure handling must be explicit. A stock adjustment made in the WMS must reach the ERP the same day, or the financial stock record decays. This is the seam that causes the overselling described in the overselling problem, where a delayed sync sells stock that left the building an hour ago.

One rule has served every client well: the ERP publishes the sellable quantity to channels, and the WMS reports physical reality back to the ERP. Nobody else publishes stock numbers. The unified operations model depends on that single direction of truth, and it keeps the integration debate simple enough to test.

The decision in practice

If you are evaluating, ask the ERP vendor what happens at your projected peak volume. Take your busiest day, double it, and ask for a walkthrough. If the demo wobbles at that volume, you need a WMS. If it handles it comfortably, save the WMS licence and the integration for later.

The sequencing matters too. Most retailers should implement the ERP first, get the stock record clean, and add the WMS when volume demands it. That order keeps the authority question settled. Jumping straight to a WMS alongside an ERP go live doubles the risk and the number of things that can break on day one. The marketplace data flows guide covers the same sequencing logic for channel integrations, and the stockout pricing guide shows what is at stake when the warehouse and the stock record disagree during peak.