An agency pitch is a prepared sample under favourable conditions. Relevant cases are selected, persuasive presenters attend and the proposed plan is shaped to the brief. That is normal professional behaviour. It also means presentation quality is incomplete evidence of delivery quality.

The selection process should test what month four might feel like: how the team handles ambiguity, communicates risk, makes forecasts, exposes working evidence and behaves when an assumption fails.

Distinction pitches for work, so this argument includes our own commercial behaviour. Buyers should apply the same tests to us.

Treat the pitch as one evidence source

Use the pitch to test understanding, reasoning, team chemistry and communication. Do not treat it as a live simulation of delivery.

Ask the agency to explain:

  • the problem it believes the organisation is solving;
  • assumptions and missing evidence;
  • approach and alternatives considered;
  • client inputs and decision dependencies;
  • named team and expected allocation;
  • quality, security and accessibility practices;
  • forecast range and major uncertainty;
  • what would cause it to stop or change course;
  • ownership, operation and exit.

Compare what the agency says with the proposal, contract and later mobilisation plan. A persuasive commitment that never reaches the agreement is fragile.

Meet the people expected to deliver

Ask which individuals will lead strategy, delivery and technical decisions, what each will do and how much capacity is reserved. Provisional allocation may be reasonable before signature; the uncertainty should be explicit.

Meet enough of the working team to understand their questions and judgement. Avoid assuming seniority guarantees quality or that junior colleagues are unsuitable. The issue is whether the evaluated capability and oversight match the team assigned.

Define substitution rules, notification and equivalent competence. People can become unavailable in any firm. A responsible plan handles change rather than promising that named individuals can never move.

Ask who reviews the work when senior presenters are not in routine meetings. Senior involvement should be a specific role and cadence, not an implication created by attendance at pitch.

Watch behaviour between formal sessions

Procurement itself provides evidence. Does the agency meet deadlines, ask useful questions, protect confidential information, correct errors and distinguish facts from assumptions?

Do not overinterpret response speed. A considered answer may be better than an immediate one. Agree communication expectations and observe whether the agency manages them.

Give bidders equivalent access to clarification. A firm that asks deeper questions should not be disadvantaged by information later shared only with others. Fair procurement helps behaviour become comparable.

Request a short example of normal delivery artefacts: a real redacted status report, decision paper, risk record, quality plan or forecast. Polished methodology slides show what a firm believes; working records show how it operates.

Create a respectful disagreement

A capable partner should challenge an unsupported assumption. Test this by asking about a deliberately uncertain part of the brief or questioning its recommendation.

Look for evidence, listening and clarity about who decides. Automatic agreement can indicate weak analysis or commercial caution. Performative pushback can be equally unhelpful.

The agency should explain consequences and respect a client decision made within authority. It must also have a route to refuse unsafe, unlawful or professionally unacceptable work.

Ask what it would remove or sequence differently if budget or time reduced. Trade-off quality is more revealing than a promise to retain everything.

Test how trouble is handled

Every substantial programme faces new evidence and difficulty. Ask for a reference from an engagement where delivery became challenging and the relationship survived.

Explore:

  • how early the issue was raised;
  • whether impact and confidence were clear;
  • which options and recommendation were provided;
  • how forecast and contract changed;
  • what the client had to contribute;
  • how the team learned and prevented recurrence;
  • whether transition remained possible.

An agency that solves every problem silently may appear proactive while making material choices on the client’s behalf. Define which incidents it can handle operationally and which require immediate client decision or specialist escalation.

References are selected evidence. Ask specific questions and, where possible, corroborate with artefacts or several roles from the client organisation.

Investigate outliers rather than applying slogans

A materially shorter timeline or lower price is a question. It may reflect reusable capability, a narrower interpretation, innovation, omitted work or unrealistic optimism.

Ask the bidder to reconcile the difference against scope, team, assumptions, quality, client dependencies and comparable proposals. Avoid declaring the outlier wrong merely because it is different.

Examine total cost: discovery, content, data, integration, licences, testing, change, support and operation. A low initial figure may be valid for a bounded phase and misleading for the full outcome.

Check how change is priced and approved. Do not assume every later change request proves deliberate underbidding; genuine discovery changes scope. The contract should preserve transparency and options.

Use paid discovery where it reduces real uncertainty

A bounded discovery or assessment can test the relationship before a larger commitment. It should have its own outcome, scope, evidence, deliverables, ownership and decision gate.

Do not use paid discovery as an automatic pre-contract ritual or a way to fragment an otherwise well-understood procurement. It is valuable when important uncertainty remains and its output can stand alone or transfer.

Define who owns research, architecture, content and other outputs if another supplier delivers the next phase. Avoid creating a discovery that only the incumbent can interpret.

Assess the working experience: access to evidence, collaboration, challenge, decision quality, forecast, documentation and response to setbacks. A pleasant team is useful; the gate concerns capability and fit.

Contract for the reality promised

Record named key roles, governance, quality standards, asset access, IP, security, data, subcontracting, reporting, acceptance, change, termination and transition with legal and specialist review.

Protect client-owned administration, repositories and working files. Visibility should exist throughout delivery, subject to appropriate access controls.

Align payments with a fair commercial model and meaningful evidence. Milestones should not encourage premature acceptance, while excessive withholding can make the relationship adversarial.

Define forecast-to-complete and escalation. The client also commits to decisions, access, content, people and review; supplier performance cannot be judged without those dependencies.

Evaluate the ordinary Tuesday

The strongest signals are consistent working behaviours: material risk raised early, choices made visible, the promised team present, evidence available and the client treated as an accountable partner rather than an audience.

The pitch begins the evaluation. Questions, references, working artefacts, a bounded discovery where justified and a specific contract make it more representative.

If an existing relationship is already failing, the companion article on what to do when your agency relationship is not working addresses recovery. Selection discipline still matters because replacing one agency without changing the operating conditions can repeat the problem.

Do not choose the least rehearsed agency. Choose the one whose evidence suggests that its prepared claims, assigned people and day-to-day delivery are the same organisation.