The most common mistake here is treating the decision as a shopping comparison: line up a few agencies, put their packages, channel lists and monthly retainers next to each other, and pick whoever sounds most confident about your industry. It feels rigorous. It isn't, because every deck in that stack was built for exactly that comparison. They will differ on price and vocabulary and differ almost not at all on the thing that decides whether the money works — whether anyone, including you, will be able to tell afterward what the spending did.
What works is inverting the order. You settle what a good outcome looks like on your side of the transaction, and how it would show up in your books, before you evaluate anybody. Do that and the shortlist mostly sorts itself. Agencies that engage with your numbers behave like a different species from agencies that route around them, and the difference is visible in the opening conversation — long before a contract, a strategy deck or an onboarding call.
Sort candidates by what they ask, not what they offer
Everyone offers roughly the same menu: paid search, paid social, SEO, content, reporting. The menu is not the product. The product is judgment applied to your account every week, and judgment is impossible to sell in a deck, which is why decks skip it. So stop reading what they offer and start scoring what they ask. Do they want to know what a customer is worth to you? Whether leads arrive by form or phone? How many of last quarter's inquiries turned into paid work, and how you'd know? Where the good customers come from now, and which ones you'd rather not have more of?
The candidate who asks nothing and quotes immediately is not being efficient. They are telling you their process doesn't depend on the answers — which means their optimization won't either. That's the difference between a shop that manages your money and a shop that manages a checklist. It also cuts across the agency-versus-freelancer-versus-in-house question, which matters far less than owners expect. A curious freelancer beats an incurious agency; a specialist who lives in one channel usually beats a generalist who touches five.
The search itself has a season
This question is asked more than people assume and less than the marketing industry pretends: roughly 1,030 searches a month across the whole cluster, split across 24 distinct phrasings of the same worry. It is also violently seasonal. The main phrase pulled 480 searches this past July and just 40 in August a year earlier — the same question, an order of magnitude apart, depending on when the budget cycle or the bad quarter lands. Meanwhile agencies pay up to $12.79 for a top-of-page click on these terms, which tells you how hard they compete for an owner in exactly your state of mind.
Bring the answer sheet to the meeting
| Ask this | What a weak answer sounds like |
|---|---|
| How will we know a lead actually came from this channel? | "You'll get a monthly report with impressions, reach and clicks." |
| What's a customer worth to us, and where would you get that number? | "We'll optimize for conversions" — with no mention of your books. |
| Who touches the account day to day, and can I meet them? | The person answering has never logged into an ad account. |
| What would make you tell us to spend less? | Nothing would; more budget is always the recommendation. |
| If we part ways, what do we keep? | "That's our proprietary setup and it stays with us." |
That last row is the one owners skip and regret. Your ad accounts, your analytics property, your tag container and your domain should be created under your own credentials, with the agency granted access as a user. Anything built inside their account walks out the door with them, taking the conversion history that made the account work. This is not a trust issue and it isn't rude to raise it — good operators expect the question and often bring it up unprompted.
The tell in the early weeks
You can judge the hire long before you can judge the results. A serious partner spends the early weeks fixing measurement before scaling anything: confirming that a form submit or a phone call actually registers, that offline sales can be tied back, that the conversion counted in the dashboard is a real event and not a page view. They show you raw search terms, not a curated highlight reel. They report in your language — jobs booked, quotes sent, revenue — and translate their own metrics into it. A partner who quietly triples spend in week one while the reporting stays vague has told you everything you need.
So the practical next step isn't a shortlist. It's an afternoon with your own records: what a customer is worth, what you can afford to pay for one, and where the last batch of good ones came from. Take that page into every conversation and the sales pressure mostly evaporates, because you're now checking answers rather than absorbing pitches. If the measurement side is the part you can't verify alone, that's a small, bounded piece of work — someone can check whether the tracking and the ad account tell the truth before you commit to anyone for a year. Hiring for that narrow job first is often cheaper than hiring for the broad one blind.