At real spend levels, a Meta ads partner should be accountable for more than the account. Expect documented budget logic, a measurement setup you can audit yourself, a creative pipeline built on distinct ideas rather than variations, and reporting you see daily. Anything less is a vendor relationship described in performance language.
What does a Meta ads agency actually own?
A media buying agency is scoped to the account, and that scope is legitimate. It covers campaign structure, audience and placement settings, bid and budget mechanics inside the platform, and the pace at which tests run. It usually does not cover the offer, the landing experience, the merchandising calendar, the retention program, or the decision about how much the business can afford to pay for a customer. Those sit with the brand, or with whoever holds the strategic layer above the account. When the account improves and the business does not, that gap is almost always the explanation. Write down, before you sign anything, which of those items you are handing over and which you are keeping. The answer changes what you should reasonably expect, and it changes who is accountable when growth stalls while the in-platform numbers look fine.
How should budget decisions be made and explained?
Budget logic should be something a partner can say out loud without opening a dashboard. The useful version sounds like a rule: spend rises on this campaign while incremental orders hold above the threshold the business set, and it comes back down when they do not. The weaker version sounds like a reaction: results dipped, so campaigns were paused. Ask how the threshold was derived, who signed off on it, and what happens to it during a promotional period when discounting changes contribution on every order. Ask what the partner does in the opening days of a scale-up, when the platform is relearning and results usually look worse before they improve. Senior operators answer those questions from memory. The answers also tell you whether budget is being managed against the business or against the comfort of a weekly report.
What does honest measurement look like before you scale?
Measurement is the part most buyers underprice. Platform reported results, the store’s own revenue, and a third party tool such as Triple Whale or Northbeam will disagree with each other, and they are supposed to. What matters is whether your partner can explain the disagreement, name which source each decision is made on, and say what must be true for that choice to hold. Before spend rises, server side events should be verified against browser events, the purchase event deduplicated, consent handling understood, and the store treated as the scoreboard. That sequencing is a large part of how Plaid Testing runs Meta buying, and it happens before any campaign gets rebuilt. Scaling on a broken signal does not create a scaling problem. It creates an argument months later that nobody has the data to settle.
What should the rhythm look like once spend is live?
Once spend is live, the rhythm splits: a daily view for money and a weekly one for judgement. Daily visibility is not about watching, it is about catching a misallocation on the day it happens rather than at the end of a month, and about owners never learning of a structural change from a summary. The weekly pass is for judgement: what moved, what was changed deliberately, what was left alone on purpose, and why. Creative runs on the same clock, with a small number of genuinely different concepts entering each month and iterations built beneath whichever ones survive contact with the audience. One warning sign is worth naming. Reporting that grows more elaborate as results decline usually means the account has run out of things to say about the business.
With a heritage fashion brand, new customer cost per acquisition fell from $33.08 to $31.06, down 6.1%, while new customer revenue rose 589.0%. The efficiency gain was small. Holding that efficiency while the count of new customers grew is the part that changed the business.
If Meta is the majority of your spend, the first thing worth auditing is what the account is being measured on.
Jason Lu is the founder of Plaid Testing, where he works on growth strategy, budget allocation and measurement for ecommerce brands rather than on day to day account management.
