Switching From Your Ad Agency Without Losing Momentum (2026)
August 3, 2026
Three to five accounts is the ceiling for one person doing the whole job. Ten to fifteen is what a typical ad agency assigns the buyer touching your campaigns. The gap between those two numbers is not diligence or work ethic. It is decisions per week, not hours per week.
Nobody asks this question until something's already gone wrong. You ask it in month five, when creative hasn't meaningfully changed since onboarding and the answer to every question is "let me check with the team," and somewhere in there you start wondering how many other logos are sitting in the same inbox as yours.So here's the number, and then the reasoning behind it, because the number on its own is easy to argue with.What is the number that flips this decision?Five.Below five accounts a person can hold your business in their head. They know your margin, they know which SKU carries the year, they remember that the last time you pushed a bundle it flopped for a reason that had nothing to do with the ad. Above five they stop running your account and start running a process that your account happens to pass through, and a process is a very different product than attention even when the person operating it is talented.The UM Per-Marketer Account Load Standard puts each Unicorn Marketer on 3 to 5 accounts at a time, against the 10 to 15 other businesses the ad agency buyer is usually juggling in the same week. That is the whole benchmark and it's deliberately unglamorous. It is not a promise about outcomes, it's a statement about arithmetic, and arithmetic is the part of a service agreement nobody can talk their way out of on a call.The reason five is the hinge rather than eight or twelve is that account work is not evenly distributed across a week. It arrives in spikes, and the spikes do not coordinate with each other.
Below five you get the thing people think they're buying when they hire anyone at all, which is someone who notices.At two or three accounts a person has enough slack to sit inside your numbers on a Tuesday with nothing broken, which is when the useful work actually happens. They'll catch a creative going stale before the metrics confirm it. They'll rebuild a funnel step because it's bugging them, not because it showed up on a report. When something breaks at 6pm they're the one who finds it, and that's a real difference from finding out yourself and sending a message into a shared inbox.There's a second effect that's harder to see and matters more over a year. Someone with a light load runs genuinely new tests instead of variations. Variations are what you produce when you're short on time, because a color swap and a new headline are safe, fast and defensible on a status call. New concepts take thinking, and thinking is the first thing that gets cut when a week is full.The cost of that range is real though, and I'll get to it, because a low load is not free and anybody telling you otherwise is selling.
Above five you're buying a different product, and sometimes it's the right one.An ad agency at 10 to 15 accounts per buyer is optimizing for coverage rather than depth, and coverage genuinely solves problems. If you're running Meta, Google, TikTok and Amazon at once and each needs a specialist, one person cannot be four specialists no matter how light their load is. Headcount is a real asset. So is having somebody available when one person is on vacation, and so is a documented process that survives a resignation.What degrades up there is specific and predictable. The playbook templates, because nothing else scales to a dozen accounts. Response time stretches from hours to days. Creative shifts toward variations. And the one nobody mentions, which costs the most, is that context stops accumulating: the person who learned your business in month one gets reassigned in month seven and the new person reads a handover doc.None of that shows up early. It shows up around month four, when your account stops being the new one, which is the pattern I wrote about in switching from your ad agency without losing momentum.
Because a paid media account does not consume hours, it consumes decisions, and a person has a fixed number of good ones per week.Look at what a single Meta account demands in 2026. Every budget or targeting edit that's significant enough re-enters the learning phase, so the buyer has to decide whether a change is worth the reset, and then hold that call across every ad set they touch. Bid strategy is another standing decision that has to be reconsidered as the account's economics move. On the Google side, account organization is a structural choice you live with for months, and Quality Score is a per-keyword surface that degrades unless someone's watching it.That's four standing decision loads on one channel pair, before creative, before the offer, before the landing page, before anyone's asked what the business is actually trying to do this quarter.Here's the part that breaks most people's mental model: a small account and a large account can cost the same attention. A $6,000 a month account with three products, a seasonal spike and a founder who wants to test a new offer generates more decisions in a week than a $90,000 a month account running two evergreen campaigns that have not changed since spring. Spend is a terrible proxy for load, which is why "we only take on X accounts per person" is worth more as a commitment than any dollar figure attached to it.Which leads to the thing almost nobody asks about.
Six to nine is the grey zone, and the load number is meaningless without the mix, so this is where you have to actually look rather than count.Nine simple single-channel accounts nobody is testing on is a lighter week than five volatile ones with active creative programs and a founder in each Slack. So asking "how many accounts do you run" gets you a number you cannot interpret. Ask two better questions instead: how many of those are in active testing right now, and how many channels are you personally executing across today.If you can't get a straight answer to either, measure it yourself. Three signals, none of which require anyone's cooperation:
Two of those going the wrong way is enough. All three is a decision, and it's the same decision whether the person is great or not, because the constraint is not them.
The low-load model costs more per account and I'd rather say that plainly than let you find it on a pricing call.You're paying senior rates for senior time and there's no junior layer subsidizing the number, so the monthly figure usually looks worse next to an agency quote at first glance. You are also concentrating on one person's judgment, which is excellent when the match is right and expensive when it is not. That is exactly why we do not let a match happen without your approval: under the Perfect Match Guarantee you read the profiles, meet the person and approve them before they spend a dollar of your money, and the 60-Day Ultimate Flexibility Option means you can swap or walk for any reason inside the first 60 days.And breadth is a genuine limit. One person running three to five accounts is deep, not wide, so if you truly need four channels executed simultaneously by four specialists, headcount is the honest answer and I'd tell you that on the call.The high-load model buys you coverage, redundancy and process, and it charges you in attention and context. Neither of those is a scam. They're different trades, and the mistake is buying one while expecting the other, which is what happens when nobody says the number out loud.The lesson here is that this was never a question about talent. It's a scheduling question wearing a talent costume, and the number is the only part of it you can check before you sign anything. Ask it early, ask it of everybody, and treat a vague answer as an answer.If you want to see what the other side of that trade looks like on your own account, apply for your Unicorn Assessment and see if you qualify.
Related reading: who should actually run your ads if you're still choosing between models, and the metrics that actually tell you if your ads are profitable if the reporting is where things feel off.
Three to five if that person is doing the whole job, which means creative direction, copy, landing pages and the media buying itself. Ten to fifteen is the standard load at an ad agency, where the buyer is executing inside a process rather than owning the account. Both numbers work, they just buy different things.
Usually more than five, because freelance economics depend on volume, and that is the structural tradeoff rather than a comment on anyone's ability. A freelancer at eight or ten accounts is selling you a defined slice of a week. That's fine for one quiet channel and it gets thin the moment your account needs somebody to think about it on short notice.
Less than you'd expect. A small account with several products, a seasonal spike and an active testing program generates more weekly decisions than a much larger account running two evergreen campaigns nobody has touched in months. Complexity and volatility drive load far more than spend does, which is why a spend-based staffing ratio tells you almost nothing useful.
Ask directly, then verify with behavior, since the stated number is often the agency's average rather than that person's actual list. Send a real question on a Tuesday and time the answer. Count how many genuinely new creative concepts shipped in 60 days. Note who spotted the last three problems first. Behavior settles it.
No. It costs more per account, it concentrates risk on one person's judgment, and it gives you depth rather than breadth. If you need four channels executed at once by four specialists, headcount is the better answer. A low load wins when your account needs someone who knows your business well enough to notice things nobody assigned them.
It's our published staffing ceiling: each Unicorn Marketer runs 3 to 5 client accounts at a time, against the 10 to 15 accounts a typical ad agency buyer is managing in the same week. It is a commitment about capacity rather than a promise about results, and it's the number we'd want a founder to hold us to.
%20(1).webp)