How to choose a digital marketing agency (questions that expose weak ones)

Why agency hiring goes wrong

Most bad agency relationships were visible on the first call — the buyer just didn't know which answers were red flags. Marketing is easy to talk about and hard to verify, so the industry selects for confident presenters, not competent operators. Your defense is a short list of questions where vague answers are disqualifying. Here it is.

The ownership questions

"Who owns the ad accounts, pixels, and audiences?"

The only acceptable answer is you do, in your own Business Manager, with the agency working as a partner inside it. Agencies that run your ads from their own accounts are building an asset they keep when you leave — your campaign history, your audience data, your pixel's learning, gone. This one question eliminates more bad agencies than any other.

"If we part ways, what do I walk away with?"

You should keep everything: accounts, creative you approved, your website, your domain, your data, and admin access to all of it. Listen for hesitation, "proprietary systems," or website platforms you can't leave with. An agency confident in its work doesn't need hostages — the work is the retention mechanism.

The people questions

"Who actually does the work on my account?"

The person selling you is often not the person running your campaigns. Ask who builds the ads, who writes the copy, who you'll talk to monthly — and how many other accounts that person carries. There's nothing wrong with teams and specialists; there's a lot wrong with discovering post-signature that your "strategist" is a rotating account manager reading someone else's notes. (Our answer, for the record: the founder, in the account personally — it's the reason the agency stays deliberately small.)

"Walk me through a campaign that failed."

Every honest operator has failures and can tell you what they learned. An agency that claims everything works is either lying or too new to have data. The quality of this answer — specific, unprompted, with the diagnosis — tells you how they'll behave when your campaign has a bad month. Because it will have one.

The money questions

"What's your fee, and what happens to my ad spend?"

Fees and media spend should be visibly separate: spend goes on your card, in your account; fees pay for the work. Watch for undisclosed markups on media, "budget included" bundles that hide the split, or percentage-of-spend pricing that rewards the agency for spending more whether or not it works.

"What's the contract term, and why?"

Some minimum commitment is legitimate — real systems need a testing cycle to prove out, and month-one judgments punish honest operators. But a 12-month lock-in with heavy exit penalties is an admission: they expect you to want to leave. A fair structure gives the work time to work while keeping the agency permanently on the hook to earn the next month.

The measurement questions

"Show me an actual client report."

Not a template — a real (anonymized) report. You're checking one thing: could a business owner read this and know what was spent, what it produced, and what changes next? If it's a wall of impressions, reach, and engagement rate with no line to revenue, that's the reporting you'll get too. Vanity reporting isn't a style choice; it's where weak results go to hide. The mechanics an agency should have wired are covered in our guide to tracking leads from Meta ads.

"What numbers will we look at together, and how often?"

The answer should name business outcomes — calls, bookings, orders, qualified inquiries, cost per each — and a real cadence. Bonus points if they define the numbers before launch, so nobody gets to move the goalposts later. (This is stage two of our process, and we'd be suspicious of anyone skipping it.)

The red flags that end the conversation

• Guarantees. Guaranteed rankings, guaranteed leads, guaranteed ROAS. Nobody controls Google, Meta, or your sales follow-up. A guarantee is either a lie or a trap built into the fine print.
• Instant audits with terrifying findings. A "free audit" that returns in an hour with everything broken is a sales script, not an analysis.
• One-size proposals. If the pitch could be sent to a bakery or an elevator manufacturer unchanged, it will perform like it.
• Badge-dropping in place of proof. Partner badges and certifications mostly measure spend thresholds and multiple-choice tests, not judgment. Ask for work and references instead.
• Trash-talking your last agency before diagnosing anything. The next conversation like that will be about you.

How to compare finalists

Once two or three agencies survive the questions, stop comparing decks and compare evidence:

• Read their case studies like a skeptic. Named clients beat anonymous "results". Defined numbers beat decorative ones. An agency that publishes few case studies with receipts is usually more trustworthy than one with twelve pages of unattributed percentages. (Judge our case studies by exactly that standard.)
• Check their own marketing. An agency selling SEO whose site can't rank, or selling content while posting nothing, is telling you something.
• Ask for a reference you choose — "a client in month eight," not the hand-picked happy one.
• Weigh the diagnosis, not the enthusiasm. The best predictor we know: did they ask hard questions about your margins, capacity, and sales process before proposing anything? An agency that prescribes before diagnosing will spend your budget the same way.

The bottom line

You don't need to be a marketing expert to hire well. You need ten direct questions, the discipline to walk away from vague answers, and a bias toward operators who show their work. Any agency worth hiring will respect you more for asking — and the ones who get uncomfortable just saved you a contract's worth of regret.