Questions to ask before hiring an ads agency.
A list of questions is not the hard part. Knowing which answers should end the conversation is — and most published question lists can be aced by an agency that will underperform for two years.
The problem with choosing an ads agency is that the sales process tests the wrong skill. A polished pitch measures how well an agency pitches, which is close to unrelated to how well it runs an account through a bad quarter. Everyone knows this, which is why question lists exist. What the lists mostly miss is that a question only helps if a weak agency cannot answer it well.
That is the test worth applying before you ask anything: could a smooth, underqualified agency give a strong-sounding answer to this? If yes, the question is theater. "What is your process?" fails immediately — every agency has a process slide. "Show me the last report you sent a client whose performance was down" cannot be faked, because either the artifact exists or it does not.
Two properties make a question useful. It asks for an artifact rather than a description — a report, a number, a name. And it has a wrong answer you can recognize without being an expert. Everything below satisfies both, and each comes with the answer that should worry you, because a question you cannot score is only marginally better than no question.
Measurement: how will we know if this is working?
Ask before any contract and ask for it in writing. This is the highest-signal question available, because measurement is where competence and honesty overlap.
- Reassuring: a governing metric, an explanation of how they separate brand from non-brand demand, a method for reconciling platform-reported results against your revenue, and a stated evaluation window with a reason behind its length.
- Worrying: a ROAS figure from someone else's account. It answers a different question — what happened elsewhere — and it means the measurement plan is being invented after the contract.
- Disqualifying: any version of "we use the platform's reporting." Google and Meta will each claim the same conversions, neither nets out demand that would have converted anyway, and an agency taking that at face value will report success your bank balance disagrees with. This is the problem in MER vs ROAS and in brand vs non-brand.
Incentives: when is the right move to spend less?
Ask it in exactly that form, then ask the follow-up: would your fee ever recommend it? Pricing is an incentive system and the question is whether they have thought about theirs.
- Reassuring: a named condition — a marginal return threshold, a saturation signal — plus an acknowledgment that a percentage-of-spend fee has a conflict there, and ideally a suggestion for governing it.
- Worrying: "we always act in the client's interest." Sincere, probably true, and unfalsifiable. It tells you nothing about what happens when the decision is genuinely marginal.
- Disqualifying: the claim that no conflict exists. A percentage-of-spend fee at 12% pays them $1,200 for a $10,000 spend increase whether it lands above or below your breakeven — an agency that will not concede the arithmetic either has not run it or is hoping you have not. The sizes are worked in PPC management pricing models compared.
Ownership: what happens to my accounts and data if we part ways?
There is only one acceptable answer here, which makes it a fast filter.
- Reassuring: you own the ad accounts, the pixel and Conversions API setup, the audiences and the full historical data; they hold access rather than ownership; offboarding is documented.
- Worrying: "we can transfer everything when the time comes." Transfers that were not architected in advance lose conversion history and audience seed data — the two assets that take longest to rebuild.
- Disqualifying: agency-owned ad accounts, a pixel living on their Business Manager, or data you cannot export yourself. That is a hostage arrangement with a monthly retainer, and it is independent of how good they are.
The artifact questions
These ask for something that exists rather than something they can describe, which is what makes them hard to fake.
- "Show me the last report you sent a client whose performance was down." The single most revealing artifact an agency owns. Reassuring: it leads with the decline, names a cause, and states what changed as a result. Disqualifying: every sample is a victory lap — you are looking at a marketing function, not a reporting one.
- "Walk me through a client you lost or fired, and why." Everyone has them. Reassuring: a specific account with a specific cause, including their own contribution to it. Worrying: "they were not a good fit." Disqualifying: no losses in five years, which means either invention or a client base too small to have a track record.
- "Who exactly will work on my account, and how many accounts do they carry?" Reassuring: a name, a load figure, and that person present in the meeting. Worrying: a team structure diagram instead of a name. Disqualifying: a load number they cannot state — at high spend, an operator carrying fifteen accounts cannot do the work, and the arithmetic does not care how good they are.
- "On Meta, how many net-new concepts per month does this include, and who owns them?" Reassuring: a number you can check against your own requirement, plus their historical hit rate. Disqualifying: "as many as needed" — the derivation in how much a Facebook Ads agency costs shows why that phrase cannot survive contact with a production budget.
- "What would you not do for us?" Reassuring: a refusal with a reason — a channel they think is wrong for you, a spend level they will not endorse, a tactic they have stopped using. Disqualifying: no refusals. An agency that will do anything is selling capacity, not judgment, and judgment is the entire product.
The questions that reveal nothing
These appear on almost every published checklist and they fail the test at the top of this article — a weak agency answers them as well as a strong one, so asking them costs you the meeting time and buys nothing.
- "What is your process?" Every agency has a process slide, and the slide is written by marketing rather than by the operator. Ask for the bad-news report instead; it shows the process actually running.
- "Can you show me case studies in my industry?" Selection bias by construction — you are shown the wins, and vertical experience matters far less than whether they understand your unit economics. Ask what they would need to know about your margins before proposing a budget.
- "How long until we see results?" Any confident number here is a red flag rather than a reassurance, since the honest answer depends on your conversion volume and how broken the account currently is. What you actually want is the evaluation window and why it is that length — which is the measurement question.
- "Are you a Google Premier Partner?" Partner status is largely a function of spend under management and certification completion. It is not a quality signal, and treating it as one selects for size.
Run all of this against us too. An agency that flinches when you apply its own recommended questions has just answered the most important one.
Where this approach has limits
- It tests transparency and self-awareness, which correlate with competence but are not identical to it. A candid operator can still be wrong about your account, and a guarded one can be excellent — you are reducing risk, not eliminating it.
- It favors agencies practiced at consultative selling. Some genuinely good small teams answer badly under interrogation and do the work well; weight the artifacts (the report, the named operator, the numbers) above the conversational polish.
- A prepared agency can rehearse these answers, particularly if a list like this is in circulation. The artifact questions resist rehearsal because a report either exists or it does not; the descriptive ones do not, which is a reason to lean on the former.
- None of it substitutes for a paid trial period or a reference call with a client who left. Those two remain the strongest available signals, and no question list beats them.
For the broader selection framework, see how to choose a performance marketing agency and the in-house vs agency comparison. To pressure-test an incumbent before you switch, the free audit shows what a genuinely independent read of your account looks like.
What should I ask a Google Ads agency before hiring them?
Ask how they will measure success in writing before any contract; when the right move is to spend less and whether their fee would ever recommend it; what happens to your accounts and data if you leave; who exactly will work on the account and how many accounts that person carries; and to see the last report they sent a client whose performance was down. The common thread is that each asks for an artifact or a number rather than a description, which is what stops an articulate but weak agency from answering well.
What answers should make me walk away from an agency?
Three are disqualifying regardless of everything else. That they rely on the platform's own reporting, which will report success your revenue disagrees with. That their fee structure creates no conflict at all, which is arithmetically false for percentage-of-spend pricing. And that they own the ad account or the pixel sits on their Business Manager, which makes leaving costly no matter how they perform. A fourth is softer but telling: no refusals — an agency that will do anything is selling capacity rather than judgment.
Which agency questions are a waste of time?
Ones a weak agency answers as well as a strong one. "What is your process" gets you a slide written by their marketing team. "Case studies in my industry" is selection bias by construction. "How long until we see results" invites a confident number that should worry you, since the honest answer depends on your conversion volume. And Premier Partner status is largely a function of spend under management, so it selects for size rather than quality.
Should I ask about account and data ownership?
Yes, and it is one of the fastest filters available because there is only one acceptable answer: you own the ad accounts, pixel and Conversions API setup, audiences and full historical data, with offboarding documented. Be wary of "we can transfer everything when the time comes" — transfers not architected in advance tend to lose conversion history and audience seed data, the two assets that take longest to rebuild.
How is hiring a Meta agency different from a Google agency?
On Meta you must ask about creative in units, because targeting has consolidated into the algorithm and creative is the lever: how many net-new concepts per month the fee includes, what share of them historically became winners, and who owns the assets. "As many as needed" is disqualifying, since creative production is a fixed monthly cost that can be derived rather than hand-waved. On Google search the equivalent questions center on brand/non-brand separation, account structure and bidding philosophy. Measurement, incentive and ownership questions apply identically to both.
Written by The ADSRUNNER team. If this resonated and you want to apply it to your own account, you can book a strategy call or run a free audit.
How we research, source figures, and handle corrections: editorial policy.