The agency retainer is a bet on availability. You are not really buying a campaign; you are buying a claim on a few people's attention, metered in hours, most of which get spent on coordination rather than the work you hired for.
That model held up when the work genuinely required a person at every step. It does not hold up as well when the same steps can run continuously.
What the retainer is actually paying for
Open any agency invoice and the line items are mostly overhead: status calls, reporting, account management, the internal handoffs between the person who plans and the person who executes. The strategy is real. The execution is real. The connective tissue between them is what you are overpaying for.
Ask a simpler question: between your monthly calls, how many hours is anyone actually looking at your account? For most small and mid-sized advertisers, the honest answer is a few hours a month, concentrated right before the report is due.
The shape of the alternative
An agent does not need a status call to know what happened. It works from the same data your reports are built from, and it works on the days between the meetings. That produces a different cadence:
- PPC — bids, budgets and creative adjusted against conversion data continuously, instead of in a Monday batch.
- SEO and GEO — technical fixes, content and local signals maintained on a loop, plus visibility in the answer engines buyers now start from.
- Web — pages and funnels built and tuned by the same system running traffic to them, so the test and the fix are not two vendors.
- Social — production and scheduling that matches what each platform rewards, without a content calendar meeting to approve it.
Where people still win
This is not an argument that judgment is obsolete. Positioning, offer, brand voice, the decision about which market to go after — those are human calls and they are the ones that decide whether any of the execution matters.
What changes is the ratio. You stop paying people to do the parts that are mechanical, and you keep them for the parts that are actually a decision.
The question is not whether an agent is better than a strategist. It is whether you should be paying strategist rates for work that runs itself.
How to test it without betting the budget
Pick one channel. Run it alongside what you have for a full cycle and compare on the only two numbers that settle it: cost per qualified outcome, and how much of your own time it consumed. If the agents lose, you have learned something cheap. If they win, you have a number to make the rest of the decision with.