Who should we trust to tell us which WMS we need?

The question assumes the answer is a party. It's a domain. Each group a buyer can ask is genuinely authoritative about something, and each is routinely asked about something else. The boundary is easy to cross without noticing, because from the buyer-only frame all of them look like experts on warehouse systems — which all of them are.

The WMS vendor is authoritative about its own product. Nobody knows it better, and a demo is a real demonstration of real capability. What the vendor cannot tell you is how that product compares to one it doesn't sell, or which of your requirements it would strain against. That silence isn't dishonesty. Strategic omission is what every sales motion produces by design: the information left out is the information the buyer didn't know to ask for, and no sales process is built to disqualify itself.

The ERP vendor and its implementation partner are authoritative about the integration and about the company's transactional data model. That knowledge is deep and it's earned. What they cannot tell you is how the floor actually runs, because the ERP side of an implementation has no occasion to watch it — and their read on the warehouse market is bounded by the products that have integrated cleanly for them before.

The systems integrator is authoritative about delivery risk: what it takes to stand a system up, where schedules slip, which configurations turn into expensive maintenance three years later. That expertise is real and badly underused at selection time. What the integrator cannot tell you is whether the system being stood up is the right one. They're usually engaged after that's been settled, and their read on any product is shaped by which ones they're staffed and certified to deliver.

An independent party has a boundary too, and it's worth stating plainly. It can document what the operation requires and evaluate the market against that without a stake in which vendor wins. It cannot tell you how a product will behave in year three of your particular configuration — the vendor's existing customers come closer to that than anyone else can. And it carries no delivery accountability, which is the price of having no revenue riding on the outcome. Independence buys an unconflicted recommendation. It does not buy an implementation.

Set side by side, the four have something in common: each is reliable about what its position lets it see, and three of them have revenue contingent on the answer to the one question the whole decision turns on. That doesn't make any of them unreliable — it makes them unreliable about that one thing.