Schema mapping
Compare the meaning of fields, types, categories, null values and units before defining a shared data model.
02 / Data integration & consolidation
Bring separate sources
into a shared picture.
Information is often spread across databases, supplier exports, collections and spreadsheets. We help align fields, identify corresponding records and build a coherent output without losing the context that makes the records meaningful.
Data integration that preserves context
Combining datasets involves more than placing columns beside one another. Different sources may use distinct identifiers, dates, classifications, units and definitions for similar concepts. Inconsistent values can hide genuine connections, while superficial similarities can create incorrect matches.
The starting point is a common purpose: what will the resulting dataset enable? That determines the fields to align, entities to resolve, conflicts to review, and level of provenance needed for future users.
Record and entity matching
Use identifiers and agreed comparison rules to propose connections; retain uncertain candidates for review.
Traceable consolidation
Preserve original keys or maintain a crosswalk, then document sources, transformations and retained exceptions.
From separate systems to usable records
- 01
Profile each source
Inspect structures, key coverage, duplicates, missing fields, formats and restrictions before joining records.
- 02
Design the target
Agree common fields, entity definitions, data types, reference vocabularies and required output formats.
- 03
Match carefully
Use deterministic keys where reliable; compare additional fields for ambiguous matches and record reasons for review.
- 04
Validate and deliver
Test totals, field constraints, source traceability and unmatched records; provide mappings and exceptions alongside exports.
A practical example: supplier catalogues
Suppose one catalogue stores product identifiers as SK-104 and weights as 0.8 kg, while another uses sk104 and 800 g. A shared model can align keys and units, but a label such as “Sand” versus “Beige” may need business review rather than automatic rewriting.
The output can include the proposed shared product record, a source-to-target field map and a separate exception report. This is an illustrative scenario, not a claim about a customer project.
What a data integration engagement can produce
Depending on scope, the handover can include a consolidated dataset, schema definitions, field mapping tables, matching rules, identifier crosswalks, reconciliation checks, lineage notes and unresolved exceptions. Outputs should match the consuming application or analysis, not a generic predefined file.
Questions that require judgement—such as whether two similar people, objects or companies are truly the same entity—must be surfaced. Integrating sources does not guarantee that every record can be matched or that all source inconsistencies disappear.
Where the main problem is inaccurate labels, duplicates or missing values within one dataset, data cleaning and preparation may be the better starting point.
Common questions
Can you merge files with different column names?
Yes, provided there is enough information to determine corresponding meanings. Field mapping and definitions are part of scoping.
What if the datasets have no common ID?
Potential matches can be evaluated using agreed attributes and review thresholds. Some may remain unresolved; matches are not assumed from name similarity alone.
Will original identifiers be kept?
The project can retain source keys and a crosswalk where that is needed to trace merged records back to their origin.
Do you replace our databases or run ongoing integrations?
A defined consolidation or integration deliverable is the starting proposition. Continuous operation, system integration and maintenance require a separately agreed scope.
How do we begin?
Describe your sources, approximate sizes, formats, how they overlap and the output you need. Do not upload confidential records in the initial enquiry.
Start with a description
Bring your sources together.
Tell us how many sources you have, the formats involved, whether records overlap and what a successful combined output should support.
No confidential dataset upload needed.
Scope and next steps come first.
Enquiry submitted
Thank you for getting in touch.
Your enquiry has been accepted for sending. We will review your message and get back to you.