The proposal stage is the only moment in the entire relationship when you hold the leverage. Afterwards you are a client with switching costs. Most buyers spend that moment asking questions that have no wrong answer. Do you have experience in our category. Who else do you work with. How do you report. Every agency alive has a prepared, pleasant response to all three, and hearing it tells you nothing.
A question is worth asking if one of two things is true: you can verify the answer later, or a bad answer disqualifies them on the spot. Apply that test and most of a standard vendor questionnaire falls away.
Four questions about the work
These are short because the evidence question is covered elsewhere on this site. Ask them anyway, and ask the first one before you have handed over any data.
- What do you think is wrong with our business right now, and what would you need to confirm it? Asked before they have account access, this separates an agency with a hypothesis from one with a template. A confident diagnosis with no stated means of checking it is worse than an honest I do not know yet.
- What would make you tell us to spend less? If there is no answer, there is no mechanism in the relationship by which that advice ever arrives.
- Walk us through a live account, screen shared, not a case study. What you are looking for is how they talk when the numbers are not curated.
- What happens in the first three weeks? You are checking whether the answer is a diagnostic or a kickoff call and a dashboard build.
The paper is where the real answers are
Everything above is a conversation. The contract is the part that still exists in month nine when everyone who was in the pitch has moved on, and it is the part almost nobody reads closely enough.
How you pay, and what that pays for
Fee structure is not an administrative detail. It is the incentive the agency operates under for the whole engagement, and it predicts their behaviour better than anything they say about values.
A percentage of media spend has a genuine case behind it: workload broadly tracks spend, so capacity and fee move together. The problem is structural rather than moral. Under that model, every recommendation to reduce spend is a recommendation to cut their own revenue. You are relying on individual integrity to overcome a standing financial incentive, every month, forever.
A performance fee sounds like the fix and usually is not. It pays for attributed results, and attribution is exactly the thing the agency controls the reporting of. You end up negotiating measurement methodology with the party being measured, which is a worse conversation than negotiating a fee.
A flat retainer removes the spend incentive, and we charge one, so treat this paragraph as interested testimony and check the reasoning rather than the source. It has a real failure mode of its own: there is no upside for growth, and as the account gets larger the quiet temptation is to service it with less senior time. The mitigations are a defined scope, a named staffing level, and the willingness to ask what has changed about the team on the account since month one.
Then ask the question that gets skipped. Are media, tools, software licences, production and freelance costs billed at cost or with a margin, and what is the margin? Get the number. A retainer that looks competitive can be rebuilt entirely out of a fifteen percent uplift on everything that passes through.
Term and notice, which are one question not two
What you keep
Account ownership is covered elsewhere on this site, so the contract-specific additions. Who owns the creative, and what are your usage rights after termination? Work made for hire and work licensed to you are very different positions when the relationship ends and you want to keep running the ads that work. Who owns the dashboards, scripts and models built during the engagement? In what format can your data be exported, and how long do you have to do it? And is transition assistance an obligation or a separately billed service, because the moment you give notice is the moment that distinction becomes expensive.
Scope, and what triggers a new invoice
Ask what is included, then ask for three examples of work that would require a change order. The second question is the informative one. An agency that cannot name the boundary either has no boundary, which means scope will be argued later under pressure, or has a very specific boundary it would rather discuss after signature. Settle revision counts and response-time commitments here too, in the dull language of the contract rather than the warm language of the pitch.
Exclusivity, in both directions
Do they currently work with a direct competitor, and will they take one on during your term? If they offer exclusivity, pin down what it means: a defined category, a geographic radius, a named list. Vague exclusivity is worth nothing. Then listen for the mirror question, because an agency asking you for exclusivity, meaning they want to be your only partner in a discipline, is telling you something reasonable about accountability, and you should know whether that is on the table.
Who is actually doing the work
Ask for the delivery team to be named in the contract, with the lead's name on it, and ask how many other accounts that lead carries. Then ask what happens if they leave. The gap between the pitch team and the delivery team is the single most common disappointment in this industry, and the only reliable protection is having the names written down before money moves.
Two questions for their references
Any reference you are given will say positive things, so stop asking whether they were happy. Ask these instead.
- What did they get wrong, and how did you find out about it? Every engagement has something. The informative part is the second half: whether the client discovered it themselves or was told.
- What do you end up doing yourselves that you had assumed they would handle? This surfaces the real scope boundary faster than the contract does, and references answer it freely because it does not feel like criticism.
Four questions for yourself
Some engagements are doomed by the buyer, and these are answerable before anyone is blamed for it.
- 01Who decides? If a change of direction needs four people to agree and none of them owns the outcome, the first quarter will be spent manufacturing consensus rather than doing work.
- 02What is the real contribution margin? Not the number you would put in a deck. An agency optimising toward a margin figure you softened will optimise toward the wrong thing, confidently, for a year.
- 03What access can we actually grant in week one? Ad accounts, analytics, CRM, call recordings, finance data. If the honest answer is six weeks, say so now and move the timeline rather than letting the diagnostic be guessed at.
- 04What will we do if they tell us to stop spending? Decide the answer before you are in the room. Plenty of buyers ask for candour and then punish the first instance of it.
Red flags in the sales process itself
- A free audit that could have been produced from public data alone. If nothing in it required access to your accounts, it is a template with your domain pasted in.
- A guarantee of a specific result. In any category with a sales cycle longer than the guarantee window, the arithmetic does not permit it.
- Urgency manufactured around capacity. The last remaining slot has been remarkably available across this industry for about fifteen years.
- A proposal with no stated constraint. If it does not say what is holding the business back, it is a menu rather than a plan, and you are being asked to do the diagnosis yourself.
- Reluctance to put the named team in the contract. This is the cheapest concession an agency can make, so refusing it is informative.
- An annual auto-renewal mentioned for the first time during redlines.
The asymmetry you are working against
You will evaluate perhaps three agencies, over a few hours each, once every few years. They have been through this conversation hundreds of times and know which answers land. That gap does not close by being more sceptical in the meeting.
It closes by asking questions whose answers you can check against reality later, writing the answers down, and then actually checking. The diagnosis they offered before seeing the data, the date they said was the longest you could be locked in, the names they put on the team, the margin they said they bill on pass-through costs. An agency that is comfortable with you keeping that list is usually the one worth signing.