Every agency has rehearsed answers for the standard questions. How long have you been in business. Who else do you work with. What is your process. Can we see case studies. None of these separate a good agency from a well-practiced one, because all of them can be answered impressively without having thought about your business for a single minute.
The questions below are harder to survive. Not because they are aggressive, but because a real answer requires work the agency has to have done before the call.
section 01"what do you think is currently limiting our growth?"
Ask this first, before they pitch, and let the silence do the work.
An agency that has looked will describe a mechanism: your service pages are not indexed, your form drops mobile submissions, you rank for your brand name and nothing else, your best-converting page has no internal links pointing at it. Specific, checkable, and occasionally wrong, which is fine. Being wrong about a specific thing is a sign of engagement. Being vague is a sign of absence.
An agency that has not looked will describe a category: you need more visibility, your funnel has gaps, your messaging could be tighter. Those sentences fit any business on earth, which is what makes them safe and useless.
section 02"who will actually do the work, and how many hours a month?"
Agencies sell with senior people and deliver with junior ones. This is not a scandal, it is the business model, and it is fine as long as you know the ratio you are buying.
The answer you want names a person, describes what they own, and gives you a number of hours. The answer that should worry you routes to a team, a pod, or a process. Those words exist to avoid the hours question, and the hours question is the one that determines whether you get the middle tier or the cheap tier at middle-tier prices.
Follow up with: will the person on this call be the person I talk to in month four? Watch what happens.
section 03"how will we know in ninety days whether this is working?"
Not whether it worked. Whether it is working, which is a different and much harder question, because at ninety days the outcome metrics have not moved yet on most channels.
A real answer names leading indicators. For search, that is indexed page count climbing toward sitemap count, impressions rising even while clicks stay flat, and the number of distinct queries you appear for getting wider. For paid, it is cost per qualified lead rather than cost per click. For any channel, it is a number that moves before revenue does.
An agency that answers this with "we will send monthly reports" has described a delivery format, not a measurement plan. The reporting version of this failure is common enough that we wrote it up separately in your agency says it's working.
section 04"what would make you tell us to stop?"
This is the question almost nobody asks and almost nobody has an answer for.
Every engagement has a failure mode. Sometimes the market is too competitive for the budget. Sometimes the offer is the problem and no amount of traffic fixes it. Sometimes the right advice is to spend the money on the product instead. An agency that has thought seriously about your business has thought about the version where it does not work.
The answer you are listening for is a threshold and a date. "If impressions have not moved by month three, something upstream is broken and we would stop and re-diagnose rather than keep publishing." That is an agency with a spine. "We are confident in our process" is an agency with a script.
section 05"what are you not going to do?"
Scope is defined by exclusions, and most proposals list only inclusions, which is how month five becomes an argument.
Ask directly what falls outside the retainer. Ask who writes the content, who implements the technical fixes, and who is responsible if the developer and the agency disagree. Ask whether you own the accounts, the data, and the content when the relationship ends. That last one is worth asking early, because the answer is occasionally no and you would rather learn it now.
An agency that answers all five of these specifically has done work before you paid them, which is the only reliable signal that they will do work after.
Bring the questions to the first call, not the third. The answers get better the later you ask and less honest at exactly the same rate.