Form: Interface Data Mapping Verification and Reconciliation Record

This is the record used to verify a data interface between two systems, the highest data-integrity risk point in automation. It captures the field-by-field mapping with units and transforms, the error-handling and transaction-integrity behavior, and the source-to-target reconciliation. Replace every <<FILL: ...>> placeholder and keep the field definitions with the record. A filled specimen follows. This content is educational and general; adapt it to your own quality system.

Section A: interface identity

Field	Entry	Definition
Interface ID	`<<FILL>>`	Unique identifier
Source system	`<<FILL: e.g. DCS FERM01>>`	Where the data originates
Target system	`<<FILL: e.g. MES/EBR>>`	Where the data lands
Direction	One-way / Bi-directional	Bi-directional needs transaction-integrity testing
Transport	`<<FILL: API / file drop / middleware>>`	The mechanism
GMP impact	`<<FILL>>`	Does the target field feed the batch record

Section B: field mapping verification

Verify each mapped field. The transform column is where unit-conversion defects hide.

#	Source field	Target field	Source unit	Target unit	Transform / rounding	Test input	Expected
1	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`
2	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`
3	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`	`<<FILL>>`

Section C: error handling and transaction integrity

#	Test	Expected behavior
E1	Interface fails mid-transfer	Target shows a gap or raises an alert; defined manual fallback invoked, no silent loss
E2	Bi-directional partial write	All-or-nothing, or a documented reconciliation step; no inconsistent state
E3	Out-of-sequence or duplicate message	Handled per design; no corrupted or duplicated record
E4	Peak transaction load	No dropped or delayed records beyond the defined limit
E5	Timestamp alignment	Source and target timestamps consistent within the defined tolerance

Section D: reconciliation

Field	Entry
Records sent by source (period)	`<<FILL>>`
Records received by target	`<<FILL>>`
Discrepancy	`<<FILL>>`
Each discrepancy explained	`<<FILL: Yes, with detail / No>>`

Section E: sign-off

Role	Name	Signature	Date
Executor	`<<FILL>>`
Process SME	`<<FILL>>`
QA	`<<FILL>>`

References

21 CFR 211.68; 21 CFR Part 11. EU GMP Annex 11 (Computerised systems), including data accuracy checks and electronic data controls. ISPE GAMP 5 (Second Edition, 2022).

Filled specimen

#	Source field	Target field	Source unit	Target unit	Transform	Test input	Expected	Actual	P/F
1	FERM01.TEMP	Step3.Temperature	C	C	none, 1 dp	37.0	37.0	37.0	Pass
2	FERM01.PH	Step3.pH	pH	pH	none, 2 dp	7.05	7.05	7.05	Pass
3	FERM01.PRESS	Step3.Pressure	kPa	bar	divide by 100	101	1.01	1.01	Pass

Error test E1 specimen: interface link pulled during transfer; EBR showed a flagged gap and raised alert INT-ALM-07; the defined manual entry fallback was used and reconciled. Pass.

The pressure row is the point of the whole form. Without the divide-by-100 transform the EBR would read 101 instead of 1.01, and a reviewer might not catch it. The explicit transform column makes the conversion testable.

Common inspection findings this form prevents

A unit-conversion or offset defect ships because only the target value was ever reviewed, not the source.
Interface failure loses data silently, with no gap, alert, or fallback.
Bi-directional partial writes leave the two systems inconsistent.
Source and target timestamps drift, so the same event reads as manipulated.
No reconciliation, so nobody can show the target received everything the source sent.

How to adapt this form

List every mapped field, not a sample, in Section B.
Add the specific failure modes your transport can produce to Section C.
Define your timestamp tolerance and your peak-load target from real production volumes.
Reference the parent interface validation protocol and the two systems’ validation records.
Keep the reconciliation evidence (source and target extracts) as attachments.