How to choose a marketing agency without getting burned

Most agencies sell the same thing in different words. Here are the questions that separate the ones who will do the work from the ones who will send you a report.

August 20, 2026 · 4 min read · by Ido Katz

We are an agency, so read this knowing that. It is still the guide we would want a business owner to have before they talked to anyone, including us — partly because the questions below are the ones we would rather be asked.

Most owners choosing an agency have been burned once already. The pattern is familiar: an impressive pitch, three months of activity, a monthly report full of numbers that go up, and no visible change in how many customers walked through the door.

That pattern is avoidable, and it is avoidable at the choosing stage rather than afterwards.

What almost every agency is actually selling

Strip the language away and there are only three models.

They run your ads. You pay a fee or a percentage; they manage campaigns. Everything else — your offer, your page, what happens after somebody enquires — is your problem. Most agencies are this, whatever the deck says.

They build the machine. Offer, landing page, follow-up, tracking, then campaigns on top. Much more work, far fewer people doing it, and the only version where a bad offer gets fixed rather than advertised harder.

They sell you leads. You buy contacts, usually shared with your competitors, usually priced per lead. Sometimes fine, but you own nothing at the end, and the same person is being sold to three of you.

None of these is wrong. Buying the first while believing you are getting the second is where the money goes.

The questions worth asking

"What happens in the first month, specifically, and what will exist at the end of it?"

Listen for nouns. A page, a follow-up sequence, tracking, a written offer. If the answer is a process rather than a list of things that will exist — "we'll do discovery, then strategy, then optimisation" — you are buying activity, not assets.

"Who owns the ad account, the pixel and the data?"

It should be you, and it should be in writing. Agencies that keep the account keep your history, and leaving means starting from zero. This one question filters more than any other, and it costs nothing to ask.

"What are you optimising for?"

Clicks and impressions are the wrong answer. Leads is better. Booked appointments or closed jobs is the right one. Whatever they name is what you will get more of, so it had better be the thing you actually want.

"What is the smallest budget where this works, and what happens below it?"

An honest answer includes a floor and a reason. Under roughly $2,000 a month of media, the platforms never gather enough data to get efficient, and you pay for the expensive part of learning without reaching the part you were paying for. An agency that will take any budget is either not being straight with you or does not know.

"When would you tell me to stop?"

Everybody has an answer for how it goes well. Ask what failure looks like and what they would do about it. If there is no version where they say "this is not working, here is why", there is no version where they tell you.

"Who is actually doing the work?"

Not out of snobbery — out of arithmetic. If the person on the call is not the person building it, ask who is and how many other accounts they carry.

The answers that should worry you

"We guarantee first page of Google." Nobody controls that. Anybody promising it is either buying ads and calling it ranking, or targeting phrases nobody searches for.

"We'll get you X leads a month," with no questions about your close rate or what a customer is worth. Leads are not the product. Customers are. An agency incurious about what happens after the lead arrives is telling you where their responsibility ends.

A twelve-month contract before anything has been built. Some lock-in is reasonable once work is underway. Asked for up front, it is a way to be paid past the point where you would otherwise leave.

A report as the deliverable. Reports are evidence of work, not the work. If the monthly rhythm is a dashboard walkthrough, you are buying a dashboard.

What to look at instead of a portfolio

Case studies are selected by definition. More useful:

• Ask what they would do in your specific market, before you have paid anything. Vagueness here means vagueness later.
• Ask for the thing itself, not a screenshot of it. A page they built, a follow-up sequence they wrote.
• Ask about a client who left, and why. The answer tells you more than five successes.

The honest summary

If you can only remember three:

1. What will exist at the end of month one?
2. Who owns the account and the data?
3. What are you optimising for?

Those three separate the agencies who will build you something from the ones who will send you a report about activity.

And if the answers are good but the numbers do not work, here is the arithmetic that decides whether any of it is worth doing — and what you should actually be spending. Both are worth running before you sign anything, with anyone.