Client references and anonymised cases: what evidence should you request from an agency?

Separate the client, the individual, the team and the outcome

A consultant may have relevant experience gained before joining the current firm. That can support selection if its context is clear and the work is not attributed to the wrong organisation. Ask when the assignment occurred and which proposed team members actually contributed.

Request geography, audiences, responsibilities and deliverables. A recognised global brand with a loosely described assignment can be less informative than a carefully documented case involving a smaller organisation. The similarity of the decision matters more than the prominence of the logo.

Build a chain of evidence

Evidence can be redacted. You do not need a former client’s confidential files to understand an agency’s method. An anonymised note, timeline or evaluation framework may be sufficient when its provenance and limitations are clear.

Accept anonymity without accepting vagueness

A case can describe an “energy-sector company” and its scope without identifying the client. Avoid unnecessary combinations of dates, places and unusual events that make identification possible. Request the evidence relevant to your decision rather than the largest possible quantity of detail.

A fictional or composite scenario must be labelled and cannot prove that an assignment took place. Likewise, an old proposal does not establish delivery of every proposed activity. For international work, distinguish a headquarters mandate from the actual work carried out in France.

Examine results and their limits

An employee population is a potential audience, not a measure of understanding. A target in a brief is not an achieved result. A change observed during the assignment does not automatically establish causation. Ask how the indicator was defined and what other factors may explain it.

For example, a fictional internal programme across eight countries demonstrates coordination scope and outputs. Additional data would be needed to establish its effects. An unmeasured impact does not make the case worthless; it requires a more precise description of what the evidence supports.

Arrange reference conversations responsibly

Where possible, request an authorised introduction to a former client. Ask about reliability, difficult moments, senior involvement and decision quality. Do not request sensitive information about a continuing crisis or confidential transaction.

Before selecting, record evidence obtained, important gaps and their implications. IPRA’s professional principles offer an ethical reference, while your decision requires account-specific facts. A credible agency can explain what it can demonstrate, what it cannot disclose and what it never measured.