Analyzing Higher Education’s Return on Investment 

The challenge

​

Comparing what people studied against what they went on to earn means combining education and workforce data that live in separate state systems, arrive in different formats, and share no clean common key. Doing this once is straightforward — doing it again for every new source and every refresh is where the cost piles up

​

• Five sources, five formats — each state's records needed standardising before anything else.

• Sensitive data everywhere — SSNs and PII had to be anonymised at every step, as a compliance requirement.

• No clean common key — matching education and workforce records required careful, repeatable linkage.

• Manual, so slow and costly — ad-hoc cleansing, validation, anonymisation, and matching for every drop.

 

The approach: build once, run automatically

​

The platform replaces hand-redone data prep with templates defined once, assigned to a flow, and turned into a pipeline that runs on its own for every future drop.

​

Five capabilities do the work

​

• Data Anonymiser — protects SSNs/PII automatically, so raw identifiers never flow downstream.

• Data Cleansing — standardises multi-source inputs into one consistent shape.

• Data Validation — catches bad or non-conforming records before they spread into results.

• Data Matching — links education and workforce records for the same person, across sources.

• Data Collection — consolidates finished output, feeding Databricks directly so dashboards stay current.

​​

The outcome

​

• The platform took ~1.2M records from five states to a working ROI dashboard, with sensitive data protected throughout:

• Set up once, reused across all five sources and every refresh — manual re-work disappears.

• Privacy by default — anonymisation runs as a pipeline step on every execution.

• Repeatable matching at scale — a predictable, reviewable match rate across a million-plus records.

• Always-current insight — collection feeds Databricks directly, no manual export or hand-off.

• Auditable by design — every flow logs each step, so results trace back to how they were produced.

 

Illustrative use case demonstrating platform capabilities; not a specific named organisation. Datasets are synthetic and figures are not a client's reported results.