Case study: missing claims

How a testing observation became a 20-million-claim analysis, a $10M finding and a permanent daily audit.

Product decision
Prevent uncertain work from appearing complete.

Document type

Case study: missing claims

Lifecycle stage

Operate & learn · 26 / Operate & learn

Contents

3 sections · 1 table · about 1 min

Status

Sample format · headline results from my healthcare claims work

Product decision

Prevent uncertain work from appearing complete.

Portfolio sample from my healthcare product experience. Requirements, thresholds and review records are illustrative; they are not employer documents or historical sign-offs.

Issue and impact

While testing EDI 837 files I noticed that claims received and claims processed did not reconcile. A one-month sample confirmed the gap, and I traced it to specific inbound files. I proposed extending the check to a 20-million-claim, 50 GB analysis. It found 300,000 missing claims worth about $10M, and the business processed them. The $10M is the value of the claims found unprocessed. It is not a saving. The control I built: an audit table that recorded each inbound file's metadata, the file identifier stored on every adjudicated claim, a daily reconciliation of claims received against claims processed, and a Tableau dashboard that showed fallout as it happened. The report joined the weekly leadership readout.

1. While testing EDI 837 files claims received and claims processed did not reconcile
2. A one-month sample confirmed the gap traced it to specific inbound files
3. a 20-million-claim, 50 GB analysis
4. 300,000 missing claims worth about $10M the business processed them

1. an audit table that recorded each inbound file's metadata
2. the file identifier stored on every adjudicated claim
3. a daily reconciliation of claims received against claims processed
4. a Tableau dashboard that showed fallout as it happened The report joined the weekly leadership readout.

Learning

A success metric can conceal everything that never reached it. Completeness has to be defined against what was received, and the audit has to run every day, not once.

1. A success metric can conceal everything that never reached it
2. Completeness has to be defined against what was received
3. the audit has to run every day, not once