Why niche-matched proof matters more than ROAS numbers when evaluating a marketing agency

If you've already paid an agency that didn't deliver, you know what a polished pitch looks like. You've seen the ROAS screenshots, the before-and-after charts, the confident slide deck. You watched the numbers and nodded along. Then month one came and went with nothing to show for it.

The problem usually wasn't that the agency was lying about those numbers. It's that the numbers weren't about a business like yours.

What ROAS actually tells you

Return on ad spend is a ratio. Divide revenue by ad spend, get a multiplier. A 10x ROAS sounds like a proof of competence. And in some contexts, it is. But a 22x ROAS from a lighting company doesn't tell a painter in Ontario anything actionable. The product price, average job size, sales cycle, customer intent, and ad format that worked for that business may share nothing with what works in your trade.

This matters because agencies know ROAS is legible and impressive, so they lead with their best number regardless of where it came from. A roofing result gets shown to a junk removal company. A painting result gets shown to an HVAC contractor. The rationale is: "we understand paid ads, and results transfer." Sometimes that's true. More often, it's just the easiest proof they have on hand.

The contractor on the other end of that call can't easily fact-check the transfer. So they defer to the number, sign the contract, and find out three months later that their trade runs completely differently.

Why the trade mismatch matters more than people realize

Two businesses can both be "home services" and have almost nothing in common from an advertising standpoint.

Epoxy flooring jobs close over days or weeks of consideration. Auto detailing books same-day. Roofing is seasonal and heavily intent-driven around storm events. Pest control has recurring subscription logic built into the customer lifetime value. Junk removal skews toward immediate, local, high-frequency demand.

These differences determine everything: which campaign objective to run, what the ad creative needs to say, how the lead form is structured, what the follow-up sequence should do, and what a realistic cost per lead even looks like. An agency that ran a great roofing campaign doesn't automatically know how to run an epoxy flooring campaign. The platforms are the same. The audience psychology, the conversion window, and the right cost-per-lead benchmark are not.

When an agency shows you proof from the wrong trade, you can't use that proof to predict what will happen to your money. It's not dishonest. It's just not evidence.

What niche-matched proof actually looks like

Specific numbers from a business close to yours. Not a category match ("we've worked with home services") but a real result from someone in your actual trade, or one close enough that the economics translate.

The Safe Step result is a useful reference point for what this looks like in practice. Safe Step is a rubber resurfacing company, which is adjacent to epoxy flooring and concrete coating. The campaign produced 247 leads at $11 per lead on $2,800 in total spend. Those numbers are specific enough to be useful. A concrete coating contractor can look at $11 CPL and work backward through their own close rate and average job size to figure out whether that math makes their business grow or just break even. A painting contractor cannot do that same math from a lighting company's 22x ROAS.

Specificity is the tell. Not specificity as in a large number of digits, but specificity as in: cost per lead, total spend, lead volume, and the trade the campaign ran in. Any agency that has real results should be able to give you all four of those data points for at least one client in or near your trade. If they can't, what they have isn't proof. It's a claim.

How to vet proof when you're talking to an agency

Ask for a result that is close to your trade first. If they don't have one, notice that. It doesn't automatically disqualify them, but it changes what you're being asked to accept on faith.

If they do have a niche-adjacent result, ask for the cost per lead and total spend, not just ROAS. ROAS is a ratio that can look good on a small number. A $200 ad spend that returned $2,000 is technically 10x ROAS, but it's also a test that never scaled. You want to see what happened over a real spend level, over a real time period, in a trade with similar job sizes to yours.

Then run the math yourself. If they say $11 CPL and you close one in four leads, you're paying $44 per booked job. Is that viable given what you charge? What does it look like at $20 CPL? What's the scenario where this doesn't work financially? A real agency won't flinch at these questions. They'll help you run them.

The last thing to notice: what is the agency asking for before they've shown you anything close to your niche? A large setup fee and a multi-month contract asked for upfront means the risk sits entirely with you. If the proof isn't specific to your trade, and the commitment is front-loaded, you're being asked to bet on faith again. That's the same situation you were in before.

The model that removes that dynamic is one where the agency carries the risk until results show up, not you. No setup fee, no contract. That structure doesn't make an agency good, but it does mean they can't fall back on lock-in if the proof doesn't hold.

What to do with this the next time you're on a call

You don't need to be combative. You just need to ask two things before anything else: "Do you have a result from a business in my trade or close to it?" and "What was the cost per lead and total spend?"

If the answers are specific and trade-adjacent, that's worth paying attention to. If the answers are vague, or the result is from a category that doesn't match yours, you know what weight to give the rest of the pitch.

If you want to see what trade-specific proof looks like before getting on a call, ASN's results page has real numbers from real campaigns. Take a look, run the math for your trade, and if it makes sense, the contact page is the right next step.