Evidence services
Evaluation archives and institutional knowledge
Turn evaluations and organisational records into searchable, reusable institutional knowledge, with source context, human review, and clear limits.
The archive problem
The reports remain. The learning can be hard to find.
Evaluations, project reports, and learning briefs often sit across disconnected folders and repositories. Staff changes can separate documents from the people who know their context. Teams need a way to find prior learning without losing where it came from or what it can tell them.
Who the service is for
Evaluation and learning teams
Explore findings and lessons across evaluations while keeping methods, dates, and evidence gaps available for review.
Programme and strategy teams
Bring relevant experience into programme design and planning, with enough context to judge whether it applies.
Knowledge managers and evidence owners
Agree how material is organised, who may use it, and who keeps the collection current.
Beyond a summary
Build knowledge that can answer the next question.
A one-time document summary
A summary answers a defined question about selected material at a point in time. It is an output to read and review.
Reusable institutional knowledge
A maintained collection supports different questions over time. Documents, structured records, source references, and curator decisions remain available for retrieval, review, and updates.
How information is organised
From scattered documents to reviewable records.
Ingestion, retrieval, and maintenance are scoped around your collection. Start with the source boundaries and the questions people need to ask.
Agree the collection
Identify the evaluations, project records, research, and learning material in scope. Confirm permissions, intended users, and exclusions before ingestion.
Prepare and describe
Review document quality and record titles, dates, versions, and source locations. Flag unreadable files, duplicates, and missing metadata for a curator to resolve.
Structure for retrieval
Agree a taxonomy for themes, sectors, places, interventions, and outcomes. Connect relevant passages to those labels and their originating documents so users can search and compare in context.
Review and maintain
Check retrieval against real user questions and original sources. Agree how corrections, new material, outdated records, and access changes will be handled.
Expected outputs
A collection your team can inspect and reuse.
Agree the deliverables for your collection and intended users. Search and structured records should carry the source context and review guidance with them.
Organised evidence collection
An agreed document inventory and metadata structure, with gaps and exclusions recorded.
Searchable, source-linked records
Relevant passages organised for retrieval and comparison, with references back to the source material.
Curation and review guidance
A record of the taxonomy, review decisions, responsibilities, and agreed maintenance process.
Source traceability
Follow the record back to its context.
A useful archive connects the material users find to the documents behind it. Reviewers need to see the source and decide what it supports.
Keep document and passage references
Retain the source location and passage context so a reviewer can inspect the original material. A retrieved excerpt is a starting point for checking, not proof of a conclusion.
Keep the setting in view
Record dates, geography, programme context, and methods where available. Make missing context visible rather than treating findings from different settings as interchangeable.
Separate records from interpretation
Distinguish what a document reports from an AI-generated answer or a team's interpretation. Check citations and make disagreement or uncertainty explicit.
Human governance
People govern what enters, changes, and gets used.
Agree ownership and review responsibilities before the collection becomes part of everyday work.
Evidence owners set the boundaries
Agree which documents may be ingested, who can access them, and how confidential or restricted material should be handled. Publication requires its own permission.
Curators review the collection
Named reviewers check metadata, labels, source references, and proposed corrections. Maintenance and access arrangements are scoped with the organisation.
People approve consequential use
Subject and context specialists review findings before they inform a decision or publication. Accountability stays with the responsible team.
An example from our work
DTLLM and Evidence Portal
The existing DT Global case study describes an evidence portal connecting internal records, academic research, and public evaluations for search and synthesis. It illustrates how a reusable evidence resource can support different learning and project questions.
Limitations
Know the boundaries of the collection.
- The collection reflects the material and permissions agreed for the engagement. Missing reports, poor scans, incomplete metadata, and language coverage can limit retrieval.
- Search can miss relevant passages, and AI can misclassify material or produce unsupported answers. Source-level checks and specialist review remain necessary.
- A source reference does not establish that a finding is reliable or transferable. Methods, bias, timing, and local context still need to be assessed.
- A knowledge collection does not replace a formal evaluation, guarantee a complete organisational record, or grant permission to share restricted information.
Work with DevelopMetrics
Start with the knowledge your team needs to reuse.
Tell us about the collection, its users, and the questions they need to answer. We can discuss the scope, source permissions, and review responsibilities together.