An honest breakdown of a rebuild budget: what is expensive, what is not, and the three items always underestimated.
Budget always comes up first, and it deserves better than a rough range. On a business-application rebuild the real split is fairly stable: around 15% discovery, 20% architecture and data migration, 45% development, 20% acceptance, deployment and support.
The most frequently forgotten item is data migration. Ten years of history in a spreadsheet contains duplicates, fields diverted from their intended use and implicit rules nobody can explain anymore. Cleaning that data regularly takes as long as building the screen that displays it.
The second underestimated item is acceptance testing. A business application replaces daily gestures: until the people performing them have validated every screen in real conditions, the rebuild is not finished. Plan for your teams' time, not only the supplier budget.
The third is the double-run period. For a few weeks the old and the new tool coexist. It is uncomfortable, but it is the only way to switch over safely on a system your invoicing depends on.