Finding out what nobody uses
Invert the default: migrate what the evidence shows is used, not what nobody will admit is dead.
On any platform more than a few years old, a substantial share of the objects and reports are dead. Nobody will tell you which — asking owners produces a list on which everything is critical, because the cost of saying “retire it” is borne by the person who says it.
Get the evidence instead:
- Query and workload telemetry from the source. Which objects have been read, by whom, and how recently. Six to twelve months of history is enough to be decisive; a full annual cycle is better, because quarter-end and year-end reporting will otherwise look dead for eleven months.
- BI platform usage statistics. Report execution counts and distinct viewers. Most BI tools carry this and almost nobody looks at it.
- Scheduler history. Jobs that have been disabled, failing silently or producing output nothing consumes.
- Downstream extract consumption. Files produced and never collected are surprisingly common.
Then invert the default. Instead of “we migrate everything unless someone proves it is dead”, adopt “we migrate what the evidence shows is used, and anything else moves only if an owner claims it within a stated period.” Publish the list, set a deadline, honour claims, and archive the rest against the source retention policy rather than migrating it.
This single exercise typically removes more effort from a migration than any technical optimisation available to you. It also needs sponsorship above the delivery team, because it is a political exercise dressed as a technical one.