The work, as it actually went

These are written so you can judge how decisions got made, which is the thing a case study is actually for. The awkward parts are left in: the systems that disagreed, the data that arrived incomplete, the exceptions automation handed back to people and the reconciliations that took longer than the build. Several include an option that was considered and dropped, because knowing when not to build something is part of the method too.

How to read these: Client names are substituted and commercially sensitive figures are replaced with representative ones, so treat the numbers as the shape of what happened rather than results you could audit. The situations, the sequence of decisions and the methods are real.