Look for the complete shape of the example
A useful case study explains the starting problem, audience, context, agency responsibility, work performed, timeframe, evidence and limitations. Without that structure, a polished story can be difficult to compare or apply to your brief. Ask for missing context rather than assuming it was favourable.
Separate observable facts from interpretation. The agency can explain why it believes the work contributed, but the result may also reflect product changes, pricing, sales activity, market conditions or other partners. The aim is not to disprove success; it is to understand what the example can support.
Put this into practice
- Starting problem and baseline
- Priority audience and relevant context
- Agency scope and named responsibility
- Method, timeframe and result source
- Limitations and other contributing factors
Establish the agency's actual contribution
Ask which parts of strategy, execution, production, technology and measurement the agency performed. Confirm whether another agency, the client team or a platform partner owned important work. A result from a multi-party program should not be attributed entirely to one supplier without explanation.
Then ask who from the case-study team is proposed for your engagement. Organisational capability can remain relevant even when individuals differ, but direct experience should not be implied if the named delivery team was not involved.
Put this into practice
- Work completed directly by the agency
- Client and partner contributions
- Decisions the agency controlled
- People involved in the example
- Overlap with the proposed team
Check how the result was measured
Ask for the baseline, measurement period, data source and definition of the reported outcome. Clarify whether the figure is absolute or relative, modelled or observed, and whether it covers all activity or a selected segment. You do not need confidential raw data to ask for a method that can be understood.
Look for measures connected to the stated problem. Impressions, reach, traffic or engagement can be useful diagnostic signals, but they do not automatically demonstrate a commercial or behavioural outcome. The case study should distinguish outputs, audience response and business effects.
Put this into practice
- Baseline and comparison period
- Data owner and measurement method
- Meaning of percentages and selected segments
- Connection between measure and stated outcome
- Known tracking or attribution limitations
Judge relevance by problem and conditions
An example from your industry can be useful, but relevance can also come from a similar audience decision, regulatory condition, technology constraint, operating scale or production challenge. Ask what the agency learned that would transfer and what is materially different in your context.
Avoid using brand fame as a proxy for difficulty or quality. A smaller engagement can provide strong evidence when its problem and delivery conditions resemble yours. Conversely, a prominent client name may reveal little if the agency's responsibility was narrow or the proposed team was not involved.
Put this into practice
- Comparable problem and audience behaviour
- Similar complexity, constraints or channels
- Relevant agency responsibility
- Transferable lesson explained
- Important differences acknowledged
Ask questions that reveal decisions and learning
Invite the agency to describe a difficult trade-off, an assumption that proved wrong and a change made after evidence arrived. This shows more about judgement than a smooth chronology in which every decision was correct from the start. Ask what it would do differently now.
Where confidentiality limits detail, ask for an anonymised explanation of the process, roles and method. Respect legitimate restrictions while keeping the evidence standard consistent. Confidentiality should not require you to accept an unsupported outcome claim.
Put this into practice
- What was the hardest decision?
- Which assumption changed during the work?
- What evidence caused a change in direction?
- What would the team do differently now?
- What can be explained despite confidentiality?
Record supported conclusions and unknowns separately
Use a consistent case-study review template across shortlisted agencies. Record what the example supports about capability, process or team experience and which questions remain open. Do not convert missing public evidence into a quality judgement; request clarification when it matters to the decision.
Score relevance and evidence quality against your criteria rather than ranking case studies by the size of the displayed result. A credible example with clear limitations may support a decision better than a dramatic claim that cannot be interpreted.
Put this into practice
- Claim supported by the information provided
- Relevance to the brief explained
- Proposed team connection recorded
- Limitations and unknowns visible
- Follow-up or reference check identified
Common questions
How recent should an agency case study be?
Recency matters when platforms, regulation, technology or the proposed team have changed, but an older example can still show useful judgement or delivery experience. Ask what remains transferable to your current context.
What if the agency cannot name the client?
Ask for anonymised context, scope, roles, method, timeframe and limitations. Legitimate confidentiality can restrict identifiers without preventing a useful explanation of what the evidence supports.
Should case studies include quantified results?
Use quantified results when they are relevant and can be interpreted with a baseline, timeframe, method and limitations. Qualitative evidence can also be useful for process, governance, research or complex change where a single number would be incomplete.