Most agency reviews stall on questions an account team can answer straight from the pitch deck: is the reporting on time, is the team responsive, does everyone like the calls. None of that tells an operator running five to a hundred locations whether the work underneath the reporting is any good. The ten questions below are built to be answered from records, not impressions — and a vague answer to any one of them is information in its own right.
1. What is the cost per acquisition at each location, not blended across the system?
A blended number hides the only thing worth knowing in a multi-location account: which locations the spend is actually working for. Ten strong locations can carry ten weak ones inside a single system-wide average, and that average will look fine right up until the weak locations close.
A good answer arrives as a table, broken out by location, on a cadence the operator can already name without asking. A bad answer restates the system-wide figure in different units, or explains that location-level tracking has not been set up yet — months into a live retainer.
2. Who is actually working this account day to day, by name and role?
Every pitch shows a strategist, an account lead, and a bench of specialists. The invoice charges a blended rate that does not disclose which of them actually shows up week to week, and nothing stops the account from being run almost entirely by the newest person on the team.
A good answer names names, states how long each person has been on the account, and states roughly how their time is split. A bad answer stays at the level of "our team," or answers with the org chart from the pitch rather than the one running the account today.
3. What happens to creative once it goes stale in one market but is still converting in another?
Creative fatigues on its own schedule in each market, not on a shared calendar. A single national refresh cycle treats a market that is still converting the same as one that has been staring at the same ad for months, which wastes spend in both directions.
A good answer names the specific signal that triggers a swap — frequency, click-through decay, a set threshold — and describes a process that runs per market. A bad answer describes a quarterly refresh calendar and nothing that would trigger an earlier one.
4. How is the split between national or brand-level spend and local, location-level spend decided — and by whom?
The ratio matters less than whether anyone can explain it. A split with a stated logic, tied to what each market needs right now, is a different thing from a split that was set at signing and never revisited.
A good answer gives a rationale tied to funnel stage or market maturity and names when it was last reviewed. A bad answer describes a ratio that has "always been" that number, with no one able to say why or when it changed last.
5. Who owns the ad accounts and the creative if this relationship ends tomorrow?
This is a contract question with an operational answer. An agency that runs campaigns inside its own ad account structure controls the audience data, the pixel history, and the account itself — all of which stay behind if the relationship ends.
A good answer states plainly that the accounts sit under the client's own business identifiers, with the agency holding access rather than ownership, and that the contract says so. A bad answer treats the question as unusual, or reveals that the campaigns live inside the agency's own infrastructure.
6. What is the notice period in the contract, and does it auto-renew?
An account team that runs the relationship day to day should know this number without opening the file. If they cannot state it, that is itself the finding.
A good answer states the period from memory and describes what happens to work already in motion at that date. A bad answer requires a pause to go check, or turns out to be a long lock-in with a short and easily missed cancellation window.
7. If the budget dropped 20% next quarter, what would you cut first?
This question tests whether anyone on the account has a ranked view of what is actually working. An account that has one should answer immediately.
A good answer names specific line items and the reasoning behind the order. A bad answer treats every line as equally essential — the answer of a team that has never had to build a model of what is carrying the results and what is simply running.
8. How does a lead or a booking get attributed back to a specific channel and a specific location?
"Attribution" is a word every agency will use. The mechanism behind it — call tracking numbers, per-location UTM discipline, form routing that preserves the source — is what actually determines whether the number in the report can be trusted down to the location level.
A good answer walks through the specific mechanism, channel by channel. A bad answer is a dashboard total with no description of how any single lead was traced back to the spend that produced it.
9. What happens when one location underperforms for two straight months?
This is the question that reveals whether a location in trouble gets noticed on its own, or gets averaged into a system-wide number until someone on the client side happens to catch it.
A good answer names a specific threshold and a documented next step once it is crossed. A bad answer is some version of "we keep an eye on it," with no threshold and no record of anyone having looked.
10. What does the reporting tell us that we could not calculate ourselves by logging into the ad platforms directly?
This is the honest test of what the retainer is actually buying — judgment applied to the numbers, or the numbers themselves in a nicer layout.
A good answer gives a specific instance of a decision the agency made that was not visible from the raw platform data alone. A bad answer is a report that restates platform metrics the client already has access to, dressed in the agency's own template.
What the pattern tells you
No single answer above proves an agency is underperforming — there is a defensible version of almost every weak answer, taken on its own. What matters is the pattern across all ten. An agency that cannot name who is on the account, cannot explain the budget split, and treats a 20% cut as a threat to its own story is describing the same underlying condition three different ways.
An agency that has done the work can describe it in specifics. An agency that has only reported on the work can only describe it in totals.
Run these ten before a renewal, not after one. The answers are available at any point in the contract — the only variable is whether anyone has asked.