Datasets
Datasets are where every answer — Human Value and Digital Twin Value alike — lives as a clean, queryable, exportable row. One source for both flavors of truth, with origin always tagged, lineage always intact, and exports always reproducible.
How it earns its place
Most teams stitch insights together by hand: pull from one tool, clean in another, join in a spreadsheet, hope nothing changed underneath. Datasets remove the seams. Synthetic and human data sit in the same row format. A filter you build once travels with you across surveys, across quarters, across teammates. Numbers shared today still pull the same way tomorrow.
What used to be a data engineering ask becomes a slice you do yourself in seconds. What used to be a fragile spreadsheet becomes a versioned, audit-ready record.
When you’d reach for it
Export a clean, typed, labeled dataset and let modeling start where it should — in modeling tools, not in cleanup.
Build a slice once — “Promoters in DACH, last 90 days” — save it, share it, apply it everywhere. No SQL, no engineering ticket.
Cite a number in a quarterly readout, then re-pull six months later and get the same number. Versioning makes sure stories stay consistent.
Combine Human Value and Digital Twin Value rows in one view — origin tagged, never blended away — and compare what each says about the same question.
One-click export to SPSS .sav, Excel, or CSV. Variable types, value labels, and metadata travel with the file. Nothing gets lost in translation.
Every row points back to the answer it came from and the run it came from. When someone asks “where did this number come from?” the trail is right there.
What good looks like
- Stakeholders pull their own slices instead of queueing through a data team.
- Numbers shared in PDFs in March match numbers re-pulled in September.
- The data science team starts in modeling, not in cleanup.
- Origin is never lost — Human Value and Digital Twin Value are clearly tagged in every row.