Employer digital twin
Employers want to know whether their turnover is bad, what is coming, and what to do about it. The twin takes their raw HR export from upload to forecast and scenarios, and every number can be traced back to the raw rows.
The problem
The raw HR export is hard to use for any of this. It comes with hundreds of column variants, free-text titles, termination codes that differ by system, and transfers recorded as if people had quit. To convince a CFO or a board, the answers also have to be comparable to peers, linked to other datasets, and traceable back to the raw rows.
How it works
01Ingest anything
Headcount, requisitions, hires, terminations, and contingent workers arrive in whatever shape the employer's system exports (which can differ for the same employer over time). Columns map by fuzzy match with confidence scores. Formats, types, and outliers are cleaned and flagged.
02Standardize to a canonical panel
Titles resolve to federal occupation codes through the occupation resolver, termination reasons and requisition statuses map to controlled vocabularies, facilities geocode. Transfers are separated from real exits, with canonical reasons attributed. Contingent and full-time workers are kept separate.
03Status quo analysis
Survival curves and cohort retention for who stays and who leaves, seasonal decomposition to separate recurring patterns from real trends. Results are compared with anonymized peer benchmarks for context.
04Project, then intervene
Forecasting methods are chosen by data depth, and every projection carries its confidence band. Market conditions are decomposed from employer-specific effects, and each candidate intervention has a literature-derived range.
One employer's own data, taken from upload through to scenario.
A few pieces of the dashboard that the code produces, for an illustrative four-site Arizona employer. Selecting a facility refocuses the role table at the bottom to that site. Toggle the local supply layer to bring in the training and staffing inventories.
Supply is modeled as stocks and flows. Every projection has to satisfy this identity, so every term in it has to be measured.
+ inflow − outflow
Manager IDs in the raw export encode a reporting tree. The digital twin can reconstruct this into levels with headcount and spans of control, so turnover can also be read by level and along internal career paths.
Tools used
Final state
The pipeline runs in production on employer panels, with the same stages the console shows, on real headcount, requisition, hire and termination data behind row-level security. The run above uses synthetic data so it can be shared. Outputs are meant to be checked against published benchmarks. For example, the modeled time to close for high-volume clinical roles matches the national benchmark for comparable positions to the day.