Research evidence workshop / 5 MIN READ
Keep the proxy’s knowledge attached to its point of view
Use proxy participants without presenting their observations or predictions as direct user testimony.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-06
A colleague, adviser or support worker may know a great deal about a user’s situation. Their account can inform research, but it has a different relationship to the experience from the user’s own explanation, and that difference needs to survive the report.
What you will leave with: A proxy-evidence record with relationship, knowledge basis and claim limits.
Use this when
- Direct participation is unavailable, inappropriate or limited for a study.
- Stakeholders offer to describe what customers think or need on their behalf.
This is not a substitute for consent, permission to obtain private case details or a claim that all proxy accounts are either equivalent to users or useless.
Explain why a proxy is involved
DWP researchers describe using proxy participants when direct user participation is difficult. Record the reason in this study and what the proxy can realistically help the team understand.
Distinguish an access limitation, an ethical concern and simple convenience. Consider appropriate ways to include direct user perspectives without pressuring people or bypassing safeguards. A proxy route should not become a permanent excuse to exclude users who could participate with suitable support. Keep the research question narrow enough that the proxy’s actual knowledge can contribute to it.
Describe the relationship and basis of knowledge
Record how the proxy encounters the relevant users and which part of their experience they can observe. A support worker may see difficult cases, while an account manager may mostly hear from organizational buyers. Neither view necessarily represents every user or every moment.
Ask for examples of what the proxy has directly observed, what users have told them and what they infer. Avoid requesting identifiable client records when a de-identified description is sufficient and permitted. A confident professional judgment remains a judgment; label it accordingly. Do not attribute a quote about users’ feelings to the users themselves when it came from somebody speaking about them.
Probe the limits without dismissing useful expertise
Ask which experiences the proxy does not see and where another perspective might differ. Explore whether their role shapes the account, such as receiving only escalated problems or being responsible for a policy they are describing. Keep the discussion respectful rather than treating potential bias as dishonesty.
Use the proxy’s account to identify questions, constraints or safeguards for further work. If they predict how a user would respond to a concept, record it as a prediction. Do not treat a proxy’s successful completion of a task as evidence that the intended user can complete it under different conditions. Expertise about a population does not recreate every member’s experience.
Report claims with their provenance intact
Keep proxy observations, reported user statements and researcher interpretations distinguishable in the findings. Explain the relationship and coverage limitations at the point where a claim is used. Avoid combining all participants into a single “users said” summary that erases who supplied the evidence.
Identify which decisions the available material can reasonably inform and which need direct evidence or another qualified source. If later user research differs, examine the context rather than forcing agreement or discarding the earlier work automatically. The proxy record is useful when it makes a particular perspective available without quietly turning it into the voice of everyone the service is meant to serve.
Make the reasoning concrete.
A fictional agency interviews client account managers about a reporting portal. They say customers want more detailed dashboards, but their conversations are mostly with senior sponsors rather than the staff who prepare reports.
Present the statement as a user requirement
Opening: Report that portal users need more detail.
Hypothesis: The recommendation sounds decisive.
Risk: It generalizes beyond the people and tasks the proxies actually know.
Preserve the perspective
Opening: Report the sponsor-facing account and identify the staff workflow as an unanswered question.
Hypothesis: The team can use the insight within its real scope.
Risk: A broader design decision may need additional direct research.
The researcher keeps the account managers’ observations linked to their role and proposes direct work with report-preparation staff where appropriate. The report does not pretend that a proxy prediction is an observed user need.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Reason
Why is proxy participation appropriate for this question?
Relationship
Which users and situations does the proxy actually encounter?
Basis
Is the claim observed, reported by others or inferred?
Limit
What perspective or experience is missing?
Use
Which decision can this evidence support without overreach?
Watch for the wrong conclusion.
- Calling every proxy statement direct user testimony.
- Requesting private case records to make an anecdote more convincing.
- Treating a proxy’s task performance as equivalent to the intended user’s.
Your next action
Take one proxy-derived finding and add its relationship, knowledge basis and exact limit before it enters a decision document.
Sources & scope
The workflow, worksheets and hypothetical examples are original editorial analysis. Sources support only the identified principles. No measured business results or named human reviewer are claimed.
- DWP Digital: How we do user research when we cannot speak to real users →Primary government research-team account · Supports: The researchers describe using proxy participants when direct participation is difficult; the claim-boundary record is an original adaptation.