At our Best Practices in Distribution Center Design, Operations and Management workshop in Atlanta this week, featuring Ken Miesemer (former Director of Distribution and International Logistics at Hershey Foods), I spoke with one rather large company in attendance that is in the process of developing its own Warehouse Management System (WMS).

That approach used to be somewhat common going back to the 1990s, but is a real rarity these days. Why? Well, “packaged” WMS software keeps getting better, and few companies have the IT staff availability or stomach for such a major (and high risk) development effort these days.

I don’t know if this was really a smart decision (I might have looked at more options), but the root of the issue was this: the finalist WMS vendor quoted something like 4000 hours of modifications for the project. That is a lot of time – and money. The company’s reaction to that quote: if we are going to invest that kind of cash, we might as well do it ourselves, get exactly what we need, and not be dependent on anyone but ourselves.

My experience is that getting to a successful end using a “do it ourselves” mode is harder and more expensive than most companies realize, and that the risk is very high. But my real issue/question was around the estimate of 4000 hours of customization. I mean, you're kidding me, right?

I know this company pretty well, and I don’t believe its distribution needs are that unique.

My suspicion is this – there wasn’t enough real vetting of the requirements, or dialog with the customer about areas where the perceived need for customization could have been mitigated with other approaches. Frankly, there are often ways of making things happen in the WMS or reducing the amount of mods needed that the front-line sales team doesn’t know.

So, while I am not sure if building a WMS today makes sense, given the long-term burden of such an effort (one large consumer goods company recently went the other way, giving up on a homegrown system due to the costs of the support/development effort and the lack of some key features available in a modern WMS), I find it hard to believe that an innovative look at what was really required couldn’t have led to a more reasonable approach and estimate.

Most new WMS systems go in with much less customization, so this was an outlier. But think the end result wasn’t good for the vendor or the customer.

As a final thought (will comment more on this on some future post), what usually gets companies that try to build their own WMS or any major supply chain application is this: they don’t know what they don’t know. Things bite them in the *ss that did the same thing to the packaged vendors years before, which the vendors eventually overcome.

There are lots of “gotchas” out there that are just very hard for first-time developers to get right when trying to mimic large commercial applications.

I’d love your thoughts on this.