There's a story the big digital agencies sell, and it goes like this: ad spend at scale is complicated, complication needs a team, and a team needs an agency. Spend enough and you'll get a "pod" — a strategist, a media buyer, an account manager, maybe a junior who actually touches the account. It sounds reassuring. It's also where the money quietly leaks out.
I took over a Meta account running north of $2 million a year from exactly that kind of setup. Not a side experiment — a primary lead engine the business depended on. The brief was simple and a little terrifying: don't break it, then make it better. Here's what actually happened.
The numbers, not the narrative
In a single 30-day window the account moved $171,150 in spend and returned 1,149 leads at roughly $149 each, across 18 live ad sets. Annualised, that's the $2M+ figure. But the headline number was never the point — the per-region number was. The best-performing markets were landing leads at $141; the worst was bleeding at $275 for a fraction of the volume. Same campaign, same creative, nearly double the cost depending on the postcode.
That spread is where every dollar of improvement lives. And it's exactly the thing an agency pod is structurally bad at catching — because the person staring at that table on a Tuesday afternoon is rarely the person allowed to change anything.
Why "smaller but expert" wins
At an agency, insight travels through a relay. The junior spots the $275 region, flags it to the buyer, who raises it with the strategist, who frames it for the account manager, who puts it in next week's deck. By the time a decision comes back down the chain, the auction has moved on and you've spent another $20K learning nothing.
In an expert-run setup there is no relay. The person reading the data is the person changing the bids — the same afternoon. A region drifting past frequency 7 (people seeing the same ad seven-plus times) gets caught and capped before it burns budget on fatigue. A region outperforming at $141 gets fed. Underperformers get cut without a meeting about whether cutting them makes the report look bad.
The advantage was never working harder than the agency. It was the distance between seeing a problem and fixing it dropping from two weeks to twenty minutes.
What the big agency was actually doing wrong
Three things, and they're depressingly common. First, cookie-cutter structure: one big national ad set instead of regional granularity, which hides the $275 outliers inside a tidy blended average. Second, frequency neglect — letting saturated markets run hot because nobody owned the daily check. Third, and most expensive, slow iteration: a reporting cadence built for client comfort, not auction speed.
None of that is incompetence. It's the math of headcount. When five people share an account, nobody owns it. When one expert owns it, every decision has a name attached and a same-day consequence.
The catch
This isn't an argument that ad management is easy, or that you should fire everyone and wing it. Running $2M responsibly takes ruthless discipline — daily eyes on cost-per-lead by region, hard frequency caps, fast creative refreshes, and the nerve to kill a $275 ad set instead of nursing it. The point isn't fewer people. It's the right person, close to the numbers, allowed to act.
The verdict
Scale doesn't reward the biggest team. It rewards the shortest distance between data and decision. A $2M account managed by one expert who reads the table and moves the bids the same hour will beat a five-person pod passing a deck around — every quarter, on the metric that actually matters: cost per lead. Headcount is not a strategy. Speed and ownership are.
If you're spending serious money on Meta and your cost per lead has quietly crept up while your agency's invoice hasn't — that gap is the tell. It's almost always fixable, and usually within the first 30 days.