Topic
Working data quality and funding monitoring reports
Reports that flag records for investigation are the most useful assurance tool available, and the most commonly misread. A flag is a question, not a finding.
Enough detail to do the job and hold a sensible conversation with MIS or Finance.
- Official requirement: Stated by the official funding document named in the sources for this page.
- FEFunding explanation: Our plain-English explanation. Helpful, but the official document is the rule.
- Worked example: An illustration of how the rule applies in one situation. Not a universal rule.
- Recommended good practice: Operational advice from FEFunding. Not itself a funding requirement.
A flag is a question, not a finding
These reports identify records that meet certain criteria, because those criteria are often associated with error.
Being on one means somebody should look. It does not mean anything is wrong.
Treating every flag as an error wastes time and damages trust in the reports. Treating every flag as noise means missing the ones that matter.
The discipline that works is three outcomes per record: wrong and corrected, right and evidenced, or right with the evidence missing. The third category is the dangerous one, because it looks fine internally and fails externally.
Appearing on a data-quality or funding-monitoring report usually means a record has been flagged for investigation, not that it is confirmed wrong. Some flagged records are entirely correct and simply need evidence.
Read the report's own published criteria rather than inferring them from its name. The criteria are specific and the name is not.
Report families, their identifiers and their criteria are published by the funding body and change between funding years, so always work from the current documentation for the year.
The method that works
Group, do not iterate.
- Read the criteria for the report.
- Group the flagged records by likely cause.
- Take each group and check the underlying data against the source evidence.
- Correct what is wrong; document what is right.
- Feed the causes back into the control that should have prevented them.
A report with two hundred rows usually has four or five causes. Working rows takes days and changes nothing; working causes takes hours and stops the report repeating.
Where the same records recur every period, the control has failed rather than the data. That is the finding worth escalating.
Maintain a standing exception log: record, cause, outcome, evidence location, date. A recurring legitimate case is then answered once rather than every period.
Track the number of distinct causes resolved per period as the control measure.
Run them yourself, first
The most valuable version of these reports is the one you run on your own data before anyone else sees it.
That way you find and fix the issues internally, while the year is still open.
Where a self-assessment toolkit or equivalent is available, running it each period is the single most effective audit preparation available, because it tests the same things in advance.
Where it is not, build equivalent queries from your own data. The questions are not secret: learners in learning too long, aims without outcomes, claims without evidence, patterns that sit outside the normal shape of the provision.
Appearing on a data-quality or funding-monitoring report usually means a record has been flagged for investigation, not that it is confirmed wrong. Some flagged records are entirely correct and simply need evidence.
The Academy's error lookup holds the symptom-led diagnostics for these patterns. Where the official catalogues of validation rules and monitoring reports have been imported, they are searchable here too; where they have not, the lookup says so rather than implying the index is complete.
Worked examples
Worked example
Two hundred rows, five causes
A monitoring report returns two hundred flagged records.
- Group. Sort by curriculum area and enrolment period. Five clusters emerge.
- Sample. Check three records in each cluster rather than all two hundred.
- Diagnose. Three clusters are a single enrolment configuration error. One is correct but unusual. One is a genuine mix.
- Act. Fix the configuration, document the unusual cluster, work the mixed one record by record.
- Prevent. Add a check for the configuration error to the monthly routine.
Half a day rather than a week, and the report does not repeat next period.
Illustrative method rather than a description of any particular report.
Common pitfalls
Treating a flag as a finding
What goes wrong: Flagged records are corrected without investigation, to clear the report.
Consequence: Correct records are changed to make a report empty, which is itself a data-integrity problem.
Prevention: Investigate before correcting. Document correct records rather than changing them.
What this means for your role
Funding or compliance officer
These reports are your audit rehearsal.
- Work them every period and keep the exception log.
See it play out
Scenario
A flagged record turns out to be entirely correct
A learner appears on a monitoring report every period. Each time, investigation shows the record is correct.
Work through itScenario
The same records appear on a report every single period
A monitoring report returns the same cluster of records month after month. Each time they are worked and each time they return.
Work through itSources
These are the official documents this page rests on. Where a figure, a deadline or an exact rule matters, the document is the authority and this page is the explanation.
Content reused from GOV.UK is Crown copyright, used under the Open Government Licence. FEFunding is not endorsed by the Department for Education.