Re: tick, tick, tick ...
Not a lot to tell really. Each school maintains a database of their own pupils and there is a minimum mandatory subset of that data that DfE require. This contains a lot of sensitive information and changes slightly according to the whims of the current regime. This is uploaded to DfE three times a year and cross referenced against previous national submissions, with errors and changes flagged for clarification/rectification.
Each collection, known as 'school census' grabs a core dataset as well as collection-specific data. For example one collection a year focuses on what classes pupils are taking on a specific day, at a specific time to get a snapshot of how the curriculum is being applied nationally.
Crapita have around 80% of the market share for the database software (SIMS - 'database normalisation? We've heard of that' - not!). DfE provide a spec to the database providers for the upcoming census and they ensure that their database can extract the required data via a reporting routine. The reporting routine flags up missing data and errors or inconsistencies across submissions over time. These are corrected prior to the extraction and there is further work done with the extraction once it is on DfE's servers, possibly requiring re-extraction/submission. You could probably download this spec if you wanted to, I don't think it is controlled, just hidden in the usual government information jungle. Obviously, this being government, external suppliers and IT, the routine doesn't always work correctly.