Closing the funding year properly

At the final return, three separate things become permanent: what you were paid, what your results say, and what an auditor will look at.

That combination is what makes the close different from every other return.

An error left in an in-year return can be fixed next month. An error left at the close is permanent in three dimensions simultaneously.

The most damaging errors at the close are omissions rather than mistakes: an outcome never recorded, a learner never withdrawn.

After the final return the year's earnings, the achievement rate population and the assurance sampling frame are all drawn from the closed data.

Plan the close backwards from the final deadline, with the dry run at the preceding return rather than the final one.

Start in the spring, not in the autumn.

• Around R09: start the cleanse and produce the issues list with owners.
• R11 and R12: record every known completion, withdrawal and outcome.
• R13: run the full final-return check set as a dry run.
• R14: confirm, submit, archive and document what is known to be wrong.

The dry run at R13 is the step colleges most often skip and most often regret. If the first time you run the final checks is at the final return, you will find work you cannot complete.

Documenting what is known to be wrong, and why it could not be corrected, materially changes how a later audit conversation goes.

Maintain the issues list with owner and date from R09 onwards, and review it at every return. Escalate anything that will not be resolved in time before the close rather than after.

Archive the submitted file, validation output and funding reports together as the record of the closed year.

The new funding year starts on 1 August, before the old one has closed. For a few months you are working in two years at once.

That is normal, and it is why the first returns of a year need as much care as the last returns of the previous one.

Configuration for the new year has to be right before the first enrolments, not before the first return. An aim configured under last year's rules produces errors from day one.

Review the specification changes, the validation rule changes and the devolved position before the year starts.

Run a validation pass on a test extract under the new specification before the first live generation, so specification changes are found in July rather than in September.

Keep the two years visibly separate in reporting. Mixing them is one of the more common causes of an unexplained funding movement in the first quarter.