We are an agency, so treat this with appropriate suspicion. But we also lose pitches, and we sit across from brands who were badly served by whoever came before us. The pattern in those conversations is consistent enough to be useful.
Almost nobody asks the questions that would have revealed the problem in advance.
Here are ten that do. Ask them of us as readily as anyone else.
1. Who exactly will work on my account, and what else are they working on?
The most common failure in agency relationships is that the seniority in the room during the pitch is not the seniority on the account afterwards.
Ask for names, roles and current client load. A strong answer is specific and slightly uncomfortable, because it admits people are shared. A weak answer talks about "the team" without naming anyone.
2. What number will you optimise toward, and why that one?
If the answer is ROAS, ask which ROAS, and whether branded search is included. If they cannot immediately explain the difference between blended and non-brand return, they will manage your account to a flattering number.
The better answer involves contribution margin, MER, or some measure connected to profit rather than platform-reported revenue.
3. Show me an account where you reduced spend on a channel.
This is the question that separates advisors from vendors.
Every agency has examples of scaling. Very few have examples of recommending less spend, because agency revenue usually rises with client spend. An agency that has never advised a client to spend less has either been extraordinarily lucky or is not paying attention.
4. What would you need from us to do this well?
A weak answer is "just account access". A strong answer asks for cost of goods, shipping costs, return rates, repeat purchase behaviour and margin by category.
An agency that does not ask for margin data cannot optimise toward profit, because they do not know what your profit is.
5. What do the first 60 days look like?
Be suspicious of an answer where campaigns launch in week one. It sounds responsive and it usually means analysis is being skipped in favour of visible activity.
The honest version involves a period of audit and analysis that produces conclusions rather than campaigns, which is less satisfying and considerably more valuable.
6. Tell me about an engagement that did not work.
Everyone has them. The question is whether they will discuss one.
Listen for whether the failure is attributed entirely to the client. "They would not follow our recommendations" is sometimes true and always a warning sign when it is the only story available.
7. What happens to my accounts and data if we stop working together?
The answer should be immediate and unambiguous: you own everything, and there is a full handover.
Any hesitation, any mention of proprietary structures or agency-owned accounts, is a serious problem. Assets built with your money should be yours without discussion.
8. How do you charge, and does your revenue rise with my spend?
Percentage-of-spend models create a structural incentive to recommend more spend. That does not make them wrong, and plenty of good agencies use them, but you should know the incentive exists and watch for it.
Ask directly what happens to their fee if they recommend halving your budget. The answer tells you a lot.
9. What is the minimum commitment, and why that length?
Both extremes are informative.
No minimum often means the model depends on volume of clients rather than depth per client. A twelve-month lock with no rationale is just a lock.
A defensible answer explains why the work needs that long: what happens in the early phase, when compounding begins, and why a shorter term would push toward decisions that damage the account.
10. Why might we not be a good fit for you?
The single most revealing question in the list.
An agency that will take any client will tell you every client is a fit. An agency with a real point of view has criteria, declines work, and can describe exactly who they turn down and why.
If they cannot name a type of business they would refuse, they are describing a sales process rather than a practice.
What good looks like in the answers
Across all ten, the signal is the same: specificity, and willingness to say something against their own immediate interest.
A good agency will tell you when your budget is too small, when your margin cannot support paid acquisition, when the problem is your product rather than your ads, and when a freelancer would serve you better.
That willingness is not politeness. It is the same judgement they will apply to your account when a channel stops working, and you want to know it exists before you find out the expensive way.
