Most problems in data exchange between institutions are not technical.
They are two organisations meaning different things by the same field name — which produces data that transfers perfectly and means something different at each end.
Agree definitions first
- What counts as an active researcher.
- What counts as a publication of this institution.
- Which affiliation applies, and at what date.
- How joint appointments are represented.
- What the reporting period covers, and by which date field.
- Agreeing these takes a meeting and prevents most subsequent problems.
Identifiers before anything else
- Without shared identifiers, matching is guesswork.
- Agree identifiers for scholars and institutions before exchanging records.
- Handle records without identifiers explicitly rather than matching them by name.
- Retain your internal identifiers alongside shared ones.
Technical arrangements
- Agree the format and its version.
- Agree the frequency and mechanism.
- Support incremental updates, not only full replacement.
- Include provenance and last-verified dates in the exchange.
- Agree how errors are reported back.
Governance
- Name who is responsible at each side.
- Agree what may be done with the data received.
- Agree onward sharing rules explicitly.
- Agree retention and deletion.
- Document the legal basis where personal data is involved.
Quality expectations
- State expected completeness by field.
- Agree what happens when quality drops.
- Agree correction responsibility and turnaround.
- Monitor volumes and rates — a change usually means a format change upstream.
- Review periodically rather than only when something breaks.
Common failure points
- Definitions agreed verbally and remembered differently.
- A format change on one side breaking the other silently.
- Staff change and nobody remaining who understands the arrangement.
- Corrections at one end never propagating to the other.
- Document everything — arrangements outlive the people who set them up.
One thing worth remembering
Write down the definitions before exchanging any data.
Two institutions can exchange records flawlessly for years and produce incompatible figures throughout — because active researcher meant one thing at one end and something else at the other, and nobody wrote it down.
Câu hỏi thường gặp
What causes most data exchange problems?
Not technical issues but two organisations meaning different things by the same field name — producing data that transfers perfectly and means something different at each end.
Which definitions must be agreed?
What counts as an active researcher, what counts as a publication of the institution, which affiliation applies and at what date, how joint appointments are represented, and the reporting period.
Why do identifiers come first?
Because without shared identifiers matching is guesswork — records without identifiers should be handled explicitly rather than matched by name.
What governance is needed?
Named responsibility at each side, agreement on what may be done with data received, onward sharing rules, retention and deletion, and legal basis where personal data is involved.
What should be done before exchanging any data?
Write down the definitions — two institutions can exchange records flawlessly for years and produce incompatible figures throughout.
