How to Evaluate a Paid Media Agency for a Service Business

Two people reviewing a proposal together at an office desk

Over the past decade or so managing millions in ad spend in Marketing lead roles, I’ve developed some strong opinions on how to vet paid media vendors.

These are the checks I run before signing anyone to run Google or Meta ads for a company that sells a service, whether that is a home services company spending $4,000 a month or a B2B firm selling $40,000 engagements. Most of it comes down to who is actually in the account, what they have run before, and which number they are willing to be judged on.

Key Takeaways

  • Default to a founder-led shop under ten people. A freelancer can be the best value you will find, or the worst, so only go solo if you can check the work. Once you are handed to a sub-team, size is working against you.
  • Category experience is the deciding factor under $10K a month.
  • Paid media must be the firm's main business. Bolt-on paid from an SEO or web shop is a junior person and a template.
  • Agree on your outcome metric before launch, share the internal numbers under NDA, and connect them to your CRM. Judge the work on your dashboard, never the platform's.

Freelancer or agency

My default for hiring out paid media is the founder-led shop: five to ten people, with the founder still in the accounts. It is small enough that a change happens this week instead of in next month’s deck, and big enough that work is reviewed by multiple people and otherwise insured against major errors. (Here are some example PPC agencies and Meta ads agencies that fit the bill.)

Freelancers can work, but the range of outcomes is much wider. The best solo operators are very good and very good value. They make the change on the call, and nobody is nudging your budget toward a partner tier. The floor is a different story. A bad freelancer has no one checking the account, and you find out when the money is gone. So the question is really about you: if you or someone on your team can read a Google Ads account well enough to spot a problem in month one, a freelancer is a fine bet. If nobody can, the shop’s process is part of what you are paying for.

Either way, ask about side deals. Partner tiers, rebates and referral arrangements with platforms and vendors are never in the proposal, and all of them tilt your budget toward whatever the firm is rewarded for. The more people in the firm, the more of these there tend to be. The software side of this is in a separate guide.

Where it goes wrong is one step larger than the founder-led shop. Once the person who sold you hands you to a sub-team for day-to-day account management, you are in trouble, unless you spend enough that the big agency’s scale is something you need. At $100,000 a month, multi-market coverage and a data team are worth paying for. At $18,000 a month you are the account the new hire learns on.

They must have done it in your category

This matters more than the freelancer-or-agency question, and most where budgets are smallest. Under $10,000 a month there is no room for wasted spend. A generalist spends your first three months learning which keywords are tire-kickers and what a lead is worth in your business. That is tuition. Hire someone who already paid it on a previous client’s budget.

High-ticket B2B with a long sales cycle is the clearest case. It is a different game from ecommerce, from local lead gen, and from most other B2B. Someone whose track record is DTC will over-index you on the wrong platforms, optimize toward cheap form fills that never become meetings, and report a great cost per lead while your sales team says nothing is qualified.

Ask directly: show me three accounts you have run in a business like mine, the budget, and what happened. A logo wall is not an answer.

The SEO firm that “also handles Google Ads.” The web shop that added Meta management to the retainer. Avoid these. Bolted-on paid media is almost always a junior person and a template, and the team is rewarded for keeping you on the retainer, not for the ads working.

The reverse is fine. A paid-first shop that also builds your landing pages, wires tracking into your CRM, or produces creative is handling the parts of the funnel that decide whether the ads work. The test: what does their homepage lead with, and what did the founder do before starting it?

Align them to your outcome, not ROAS

The usual pattern: the agency reports return on ad spend and hits its number. Meanwhile you wanted visibility in a new market, faster first meetings for your sales team, or a second channel so you depend less on referrals. ROAS measures none of that. It is the platform’s metric, built from the platform’s own attribution, and it usually looks good.

Before launch, decide what you are buying and write it down. Booked jobs per month. Qualified first meetings. Cost per customer, not cost per lead. That number goes first in every report.

This means letting them in. Put them under NDA and share close rates, average job value, and what sales says about lead quality. An agency that knows a Google lead is worth three times a Meta lead for you will spend differently than one optimizing both to the same cost per lead.

It also lets you cut quickly. If you agreed on cost per booked job and 90 days in the report still leads with impressions and click-through rate, the conversation is short.

Make them live in your data

Connect them to your CRM and to the dashboard you actually run the business from, and say it out loud: I do not care what the Meta dashboard looks like. I care what this dashboard looks like.

  • Read access to the CRM with lead source and stage visible. If your CRM cannot show lead source by campaign, fix that before you spend on ads. Our cleaning company and gym CRM roundups cover tools that do this out of the box.
  • Call tracking with recordings. See our call tracking roundup. Ten recorded calls tell you more about lead quality than any platform metric.
  • One shared report on your numbers. Not their deck and your spreadsheet.

A lot of paid media teams are weak on attribution and would rather not be held to it. A team that resists connecting to your CRM is telling you which number they trust. And if the weekly report leads with impressions, click-through rate, cost per click or platform ROAS, those are diagnostics, not outcomes. Ask what happened to the number you agreed on.

Questions to ask before signing

  1. Who is in my account every week, by name, and how many other accounts do they run? More than ten is about an hour a week.
  2. Show me three accounts in a business like mine. Budget, duration, result, and whether I can talk to the client.
  3. What share of your revenue is paid media management? Under half and it is a bolt-on.
  4. Do you have partner status, reseller agreements or referral arrangements with anything you will recommend to me? You want the list, not a reassurance.
  5. What would you cut from my account on day one? A good operator can tell you. One who only adds is building a retainer.
  6. What is the notice period? Thirty days. A twelve-month lock-in plus a platform you do not own is a red flag.

The first 90 days

Outcome metric and target in writing before launch. CRM access and internal numbers under NDA in week one. At 30 days, tracking works from click to CRM record. At 60 days, cost per outcome is trending. At 90 days it is at target, on a credible path with specific changes behind it, or you are having the exit conversation with a clean metric to point at.

Patrick Ward
Written by Patrick Ward
Hi, I'm Patrick. I'm a marketing executive and the founder of NanoGlobals, where I help lean marketing teams do more through automation and offshore staffing. Past work with Nike, Walmart and MasterClass; featured in Forbes and Ad Age. Published Nov 15, 2026 View all posts by Patrick Ward →
Table of Contents