- Result of the engagement
- Training and grants
Seven hours a week recovered through monitoring public notices
A training company cut the time spent on weekly monitoring from eight hours to one. An agent checked updates from 320 job centres, and an employee received a summary of the changes that needed attention. The seven recovered hours went into matching available funding options to client needs.
Problem
One person manually reviewed notices, checked rules and copied information into a spreadsheet. The task took an average of eight hours a week.
The change
An agent searches updates published by 320 job centres once a week and writes changes into a database. The employee receives a summary of the changes that need attention.
Effect
Human working time fell to one hour a week. The employee spent the seven recovered hours matching funding options to client needs.
- Key result
- Seven hours a week recovered in one role, that is 87.5% less human time on this task. The difference held for eight weeks after launch.
Measurement
| Metric | Before | After | Change |
|---|---|---|---|
| Human working time, weekly | 8 h | 1 h | −87.5% |
| Omissions in a sample of 320 updates | 16 | 2 | |
| Tool running time, weekly | — | 35 min |
Basis of measurement: Measured over eight consecutive weeks before the change and eight weeks after launch. Completeness checked on a sample of 320 updates against a separately prepared list of required information.
The challenge
Previously one person manually reviewed notices, checked rules and copied information into a spreadsheet. Over eight consecutive weeks the task took an average of eight hours a week, including checking and correcting the summary. A completeness check found 16 omissions in a sample of 320 updates that, under the agreed scope, should have reached the database.
What we changed
The person leading the project on behalf of dyrektor.ai agreed the list of sources, the scope of data collected and the way changes are presented. The delivery team built the monitoring and the database writes. The company’s employee was responsible for judging the usefulness of the result and handling exceptions.
Measurement and results
After launch the average human working time was one hour a week. That covers reviewing the summary, checking doubtful items and making corrections. The tool itself ran for an average of 35 minutes and did so without constant supervision.
For eight weeks after the deployment the difference held at seven hours a week, that is 87.5% less human working time on this task. In the next sample of 320 updates the check found two omissions. Quality was assessed against the same completeness criteria, using a separately prepared list of the information required in the summary.
What the owner gets
The owner gained regular monitoring and more of the employee’s time for client work. He also kept the information about limitations: the two omissions that were found fed further improvement of the solution. The result of the project covered both time saved and control over data quality.
- Conclusion
- The result covered both time saved and control over data quality. The two omissions that were found fed further improvement of the solution.