I've just been catching up on the debate about ERP systems (CW March 1, p1).
I've been in IT for the past 22 years and have implemented commercial ERP systems (Symix and MFG/Pro) and have also been in charge of, or involved with a number of custom in-house and hybrid efforts - successful projects both here and overseas.
I've seen it from both sides and have concluded that for smaller companies the commercial ERP systems are the way to go; however, for larger companies, significant cost savings can be achieved by going it alone.
The benefits include: far greater degree of fit and customization (competitive advantage); faster response time to changing business requirements (competitive advantage); more immediate support response from in-house people who know the business intimately and actually care; greater opportunity to make legacy system interfaces and add-ons seamless; ongoing upgrade and maintenance program instead of infrequent painful release and upgrade cycle at the vendor's whim; lower cost of ownership - initial development and installation (about the same or a bit more), ongoing maintenance (lower, including staff costs), cost of additional users (negligible); implementation can often be tackled in more manageable chunks at a pace that suits the availability of internal resources.
The main benefit of a packaged system is that you can implement (somewhat) earlier, but coming to terms with all the additional complexities, inflexibilities and compromises can be as time consuming as getting started with the basics and doing a few rounds of enhancements.
I'm postulating that the break-even point is probably at around $80 million to $120 million revenue. Anyone who spends $10 million-plus on a commercial ERP system is getting extremely poor value for money compared to a well-written custom system. The trick is getting the right people on staff who have the experience and ability to pull it off.