Treat unsupported certainty as a question
Guarantees about rankings, sales, reach or return should be examined against what the agency can actually control. Before access to your evidence, systems and constraints, a precise outcome can rest on assumptions that have not been tested. Ask for the calculation, dependencies and conditions under which the claim would not hold.
Strong agencies can still be optimistic. The warning sign is not confidence itself; it is an unwillingness to explain uncertainty, trade-offs or evidence limits. Record important claims and make sure the proposal and agreement describe responsibility accurately.
Put this into practice
- Outcome promised before diagnosis
- Important assumptions left unstated
- Result claims without baseline, method or timeframe
- Agency responsibility confused with the whole business outcome
Look for gaps in team and scope
A senior pitch team with an unnamed delivery team creates a material unknown. Ask who will lead, perform and approve each workstream, how much access is included and where external partners are used. A change in people is not automatically a problem, but hidden staffing assumptions can change service quality and continuity.
Vague words such as strategy, optimisation, management or support can conceal very different levels of work. Request volumes, frequency, owners, response expectations, review cycles and exclusions. Be cautious when every request is said to be included but no process exists for priorities or scope change.
Put this into practice
- Delivery roles unnamed or frequently qualified
- Pitch participants presented as ongoing team without confirmation
- Activities listed without volume, owner or service level
- Unlimited revisions or support with no decision rules
- Important implementation work left unassigned
Check for commercial ambiguity
A bundled price can be workable, but you should understand the treatment of media, production, technology, travel, external specialists and optional work. Ask what is included, which costs are passed through, whether a fee or mark-up applies and who can authorise spend.
Review minimum commitments, fee reviews, pauses, cancellation and transition. A low entry price can carry assumptions that make the likely total less clear. Compare the same scenarios and client responsibilities across proposals before judging value.
Put this into practice
- Agency fee and pass-through costs cannot be separated
- Likely optional work has no rate or estimating method
- Approval authority for extra spend is unclear
- Minimum term or exit cost appears late
- Quote depends on significant unpriced client work
Protect account access, assets and rights
Confirm that your organisation receives appropriate access to advertising, analytics, domain, platform and data accounts. Ask where accounts will be created, who holds administrator rights and how access is transferred. Supplier-managed access may be convenient, but the control and exit consequences should be understood.
Clarify ownership and usage rights for strategy, copy, design, footage, photography, fonts, music, talent, software and source files. Check licence periods, channels, territories and handover. Seek suitable legal advice for rights or contractual issues that are material to your organisation.
Put this into practice
- Client administrator access restricted without a clear reason
- Account ownership changes after appointment
- Usage rights, licences or source files omitted
- Handover depends on an undefined future fee
- Data retention and deletion responsibilities unclear
Notice how difficult questions are handled
Selection is an early sample of the working relationship. Pay attention to whether the agency answers the question asked, acknowledges a gap and follows up when promised. Repeated deflection, pressure to decide before material information is available or resistance to written clarification deserves closer examination.
Do not confuse a respectful challenge with poor service. An agency should be able to question an impractical brief, explain why and offer alternatives. The useful distinction is whether the challenge is evidence-led and helps the decision, or merely avoids accountability.
Put this into practice
- Material questions repeatedly answered with general claims
- Pressure to skip reference, access or contract checks
- Verbal commitments resisted when put in writing
- Risks minimised without an alternative plan
- Disagreement becomes personal rather than problem-focused
Investigate, document and decide proportionately
A red flag is not proof that an agency is unsuitable. Ask a specific follow-up, request evidence and give the supplier a fair chance to resolve the concern. The answer may reveal a reasonable constraint, a difference in terminology or a practical mitigation.
Record the concern, response, remaining risk and decision. Some gaps can be managed through discovery, governance or contract terms; others affect eligibility or trust. Apply the same standard across shortlisted agencies and do not invent a negative score simply because information is not yet public.
Put this into practice
- Concern described as an observable fact
- Agency response and evidence recorded
- Impact and likelihood considered
- Mitigation or contract action identified
- Decision applied consistently across suppliers
Common questions
Is one red flag enough to reject an agency?
Not always. Consider what the issue affects, whether it can be clarified and how the agency responds. A confirmed failure against a mandatory requirement is different from a missing detail that can be resolved fairly.
Are awards and large client logos red flags?
No. They can provide useful context, but they should not replace relevant evidence about the proposed team, problem, contribution and result. Ask what the example demonstrates for your brief.
What should be put in writing before appointment?
Record the agreed scope, named roles, fees, assumptions, access, rights, service expectations, change process, third-party costs, reporting, handover and exit terms. Obtain appropriate advice for your contract and risk context.