I'd like to add one point to your great article. DW ROI is not proportionate to the development time invested.

Some business areas can provide high ROI, with a short development time. Having some idea of how the business operates, I like to approach these high ROI areas first - the outcome leads to a DW selling point to the wider company.

Also, for goodness sake, if the company decides to purchase a logical model like the IBM retail model, it's important that they understand that it isn't a functional DW. I'm not sure how IBM sell some of their models, but at least one company I've worked with has had the expectation that the retail model is a bolt-on solution.