Six Questions to Ask a Marketing Agency Before You Sign

Most agency evaluations go badly for a structural reason: the buyer does not know the field well enough to tell a good answer from a confident one, and the pitch is built around that.

The fix is not learning marketing. It is asking questions where the quality of the answer is legible even if you cannot evaluate the underlying work. Below are six, with what a strong answer sounds like and what a weak one sounds like.

Use them in order. The first three will end most conversations.

One: What Will You Do In the First Thirty Days, Specifically?

Weak answer: onboarding, discovery, an audit, a strategy document, a content calendar.

Strong answer: a named diagnostic with a named output, and a decision that follows from it. Something you could verify happened.

This question works because the first thirty days reveal the operating model. An agency that opens with a content calendar has decided what to sell you before looking at your situation. An agency that opens with a diagnostic does not yet know what it will recommend, which is the correct state to be in on day one.

Watch whether the answer changes based on what you told them. If the thirty day plan would be identical for you and for any other business in your category, you are buying a template.

Two: What Would Make You Tell Me Not to Buy This Service?

Weak answer: deflection, or a version of "it works for everyone."

Strong answer: a specific circumstance, stated without hedging.

Every service has conditions under which it is the wrong purchase. Content and search work is a poor fit for a business that needs revenue in sixty days. Answer engine optimization is wasted on a business whose name the engines cannot yet recognize. More visibility is the wrong spend for a business whose offer does not convert the traffic it already has, or one without the capacity to serve more customers.

An agency that cannot name a disqualifying condition either does not understand the boundaries of its own service or is unwilling to tell you about them. Both are expensive.

This is also the fastest read on honesty, because the honest answer costs the person answering it something in the moment.

Three: How Will We Know In Ninety Days Whether This Is Working?

Weak answer: traffic, impressions, rankings, engagement, reach.

Strong answer: a metric that moves before revenue does but is causally connected to it, plus an honest statement of what will not have moved yet.

The distinction matters more than it used to. Impressions and rankings tell you your marketing exists, and they can rise while nothing happens to your pipeline. That gap has widened as answer engines have taken over the top of the results page: BrightEdge reported in February 2026 that only about 17 percent of AI Overview citations came from pages ranking in the organic top ten, down from roughly 76 percent in mid-2024. Ranking well and being recommended are no longer the same achievement, and a report built on rankings alone will not tell you which one you have.

Better leading indicators are inbound inquiries by channel, branded search volume, and citation presence for the specific questions your buyers ask.

The second half of the strong answer is the real tell. An agency that says "revenue will not move in ninety days and here is why, but this will" is describing a mechanism. An agency implying everything improves at once is describing a hope.

Ask a follow-up: what does month three look like if this is working but slowly, versus if it is not working at all? If those two pictures are identical, nobody will be able to tell the difference when you get there.

Four: Who Actually Does the Work?

Weak answer: "our team," followed by a vague process description, or a named senior person you never speak to again after the pitch.

Strong answer: a clear division between what is done by people, what is done by systems, which specialists touch which work, and who you talk to when something is wrong.

There is nothing wrong with a platform doing execution. Cadence is genuinely hard to sustain by hand, and the businesses that fall out of the answer are almost always the ones that published in bursts and went quiet. A system that holds a cadence is an advantage, not a shortcut.

What matters is that the division is stated. It determines what you are actually paying for and where quality will vary. An agency that conceals its automation is signaling it believes you would object, which is a bad sign about how it is being used. An agency concealing that the work is junior is signaling something worse.

The arrangement worth looking for is a platform handling production and consistency, specialists in each discipline handling the work that needs judgment, and one accountable person who knows your account and answers the phone. Ask which parts are which, and ask who you escalate to.

Five: What Happens to the Work If We Stop Working Together?

Weak answer: vagueness, or discovering that assets live in accounts you do not control.

Strong answer: immediate and specific, because it has been asked before.

This is a leverage question disguised as an administrative one. If your business profile, advertising accounts, domain, analytics, and published content sit under the agency's ownership, switching costs are artificially high and everyone in the relationship knows it.

Ask specifically: who is the owner on the business profile, who is the account owner on advertising platforms, where does the content live, and who holds the domain registration. Get the answer in writing before you sign rather than discovering it during an exit.

Six: What Does This Cost, and What Drives It Up?

Weak answer: a single number with no structure, or a number that only arrives after several more meetings.

Strong answer: a stated model, what is included at each level, and what specifically triggers additional cost.

The structure of the pricing tells you the structure of the incentives. Hourly billing rewards hours. Deliverable counts reward deliverables. Neither rewards outcomes, which is why so much agency work optimizes for volume nobody asked for. Fixed pricing per service, published rather than negotiated per prospect, aligns the incentive toward the result, because more hours no longer means more revenue.

Ask what happens when something takes longer than expected, whether revisions are limited, and what is explicitly out of scope. The specificity of that answer predicts the specificity of the relationship.

The Trade-Off Worth Knowing Before You Start

The agency that answers all six of these honestly will frequently quote higher than the one that does not.

That is not a coincidence. Naming disqualifying conditions, being specific about timelines, admitting what will not move in ninety days, and handing you full ownership of the assets all reduce the seller's leverage. A firm willing to give up that leverage generally has to charge more to be viable, because it will close fewer deals and will lose the ones where the fit was genuinely poor.

Which means the cheapest quote in your stack is often the one optimized to be chosen rather than to work. Not always. But if the cheapest proposal also had the vaguest answers to questions two, three, and five, you have learned something worth more than the difference in price.

How We Answer Them

We would rather you asked these of us than not, so here is the short version.

The first thirty days start with a visibility diagnostic, not a content calendar: where you show up today across search, answer engines, and locally, where you do not, and which of those gaps is costing you. The plan comes out of the diagnosis.

We will tell you not to buy if you need revenue in sixty days, if your offer does not convert the traffic you already have, or if you do not have the capacity to serve more customers. In those cases more visibility makes the problem worse rather than better.

At ninety days you should expect movement in citation presence, branded search volume, and inbound inquiries. You should not expect attributable revenue yet, and any agency telling you otherwise is guessing.

The work runs on a proprietary platform that handles production and holds the cadence, with specialists in their fields on the work that needs judgment, and one accountable person who knows your account.

Pricing is fixed per service and published on the site rather than quoted per prospect, with bundle discounts at three and five services.

Axia Atlas was founded to run marketing as a system rather than a set of deliverables. See how the services are structured at axiaatlas.com/services, or book a demo and put all six questions to us directly.

Frequently Asked Questions

How many agencies should I talk to?

Three is usually enough to calibrate, and more than five stops adding information. The purpose of multiple conversations is not to find the lowest price but to notice which answers vary. When two of three give the same answer to question three and one gives a very different one, that difference is the useful signal.

Is a long contract a red flag?

Not automatically. Some work genuinely does not show results inside three months and a longer commitment is honest about that. What matters is whether the length is justified by a stated mechanism or is simply the default. Ask what happens at month six if the leading indicators from question three have not moved.

What if I do not understand the answer?

Say so and ask them to explain it again without the terminology. How they handle that is itself diagnostic. An agency that can explain its own work in plain language understands it. An agency that retreats further into jargon is protecting something.

Should I ask about their own marketing?

It is a fair question and a limited one. An agency with modest visibility in its own category may simply be busy serving clients. More useful is asking what they measure about their own performance, because it reveals what they actually believe matters.

How much should I expect to pay?

Enough that the agency can afford to turn down poor-fit work. Published fixed pricing is easier to evaluate than a custom quote, because you can compare what is included rather than trying to compare two numbers arrived at differently. Whatever the figure, ask question six and get the escalation triggers in writing.

Sources

Citation overlap with the organic top ten, about 17 percent in early 2026 against roughly 76 percent in mid-2024: BrightEdge, February 2026.