Marketing Operations: The Work You Already Have

You Probably Already Have This Problem

The monthly report gets rebuilt by hand, again, because nobody wrote down how it was built the first time. The weekly marketing slot moves three weeks running and then quietly stops being a slot. A campaign ships four days late because two people were each waiting for the other to approve the copy. A tool nobody has opened since spring is still billing. Nobody is doing a bad job. There is simply a set of work that has never been given to anyone. That work has a name, and the name is the problem. Marketing operations sounds like something a company with a marketing department has. The phrase arrives attached to enterprise software, job titles, and conference tracks, so an owner-operator reads it once and correctly concludes it is not addressed to them. What the phrase actually describes is much smaller and much more familiar: the work of making marketing happen on purpose rather than by memory. You do not decide whether to have this function. You decide whether anyone owns it. Left unowned it does not disappear — it gets absorbed, unevenly and at the worst moments, by whoever has capacity that week. Usually that is you, between customer calls and payroll. This page is for the reader who has never had this function and is starting to feel its absence. It defines the work, explains why it accumulates whether or not anyone is assigned to it, shows one artifact from our own operation, and points at a first move.

What Marketing Operations Actually Is

Marketing operations is the set of recurring, unglamorous jobs that keep marketing running: getting work out the door on a cadence, measuring what happened, keeping the data and tooling from rotting, and deciding what not to do. That is the whole definition. It is deliberately boring, and the boringness is the point. Every one of those jobs is invisible when it is done and expensive when it is not. The clearest outside description of this work sits in the federal occupational data. The task inventory for marketing managers on O*NET reads less like a creative brief than like a maintenance schedule: coordinating activities, evaluating results against objectives, directing the people who produce the material. Read it as a description of the work rather than of a hire — that inventory covers marketing managers specifically, a narrower population than the one doing this work inside a small business, where it is usually spread across two or three people who each own a fragment.

The distinction that matters is between the marketing and the operating of it. Writing the email is marketing. Knowing which list it goes to, that it goes out on the second Tuesday, that the link is tagged so the result is attributable, and that somebody reads the result afterwards — that is the operating layer. Businesses of every size do the first well and the second by accident. Two failure modes follow, and you will recognize both. The first is activity without a record: work happens, nobody can say why it happened or what it produced, and the same decision gets re-made every quarter. The second is a record nobody uses: a spreadsheet, faithfully maintained, that no decision has ever consulted. The function is working only when the record and the decisions are connected.

The Four Jobs, and Where They Are Covered in Detail

The four jobs above are the whole surface: cadence, measurement, hygiene, and the decision about what to stop. At a small size, all four are usually covered by one or two people who do not have “operations” anywhere in their title, and the useful question is not who to hire but which of the four is currently uncovered. We have written the long version of that elsewhere, and it does not need repeating here. If you want the four jobs broken out with the smallest structure that covers each of them at three different business sizes, read what a marketing operations team actually looks like. It carries the staffing detail, the size thresholds, and our own seat-by-seat roster. This page is the entry point; that one is the blueprint.

What It Costs to Leave the Work Unowned

The cost is not a line item, which is exactly why it survives so long. It arrives as three quieter charges. The first is rework. Anything not written down gets rebuilt from memory by whoever is available, at full price, every time. The monthly report is the usual example because it is the most visible, but the same tax sits on every recurring thing: the launch checklist that exists only in one person’s head, the approval path that gets renegotiated per campaign, the brief that starts from a blank page because last quarter’s is unfindable. The second is decisions that never get made. Stopping something requires a moment where somebody looks at what is running and asks whether it still earns its place. Without an owner, that moment never arrives on its own. Spend does not usually leak through bad decisions. It leaks through decisions nobody scheduled. The third is the one owners feel and rarely name: the work lands on the person least able to refuse it. Marketing operations, unowned, defaults upward. It reaches the founder, who has the most context and the least time, and who then becomes the bottleneck for a set of tasks that were never supposed to require judgment at all. These three compound, and the compounding is what makes the timing counterintuitive. Each individual instance is small enough to absorb, which is precisely why none of them ever triggers a decision. The rework is an afternoon. The unmade decision is one line of spend. The escalation to the founder is one interruption. No single instance is worth restructuring anything for, and the sum of them is a marketing function that runs at roughly half the speed the business is paying for. None of this shows up as a failure. It shows up as a business that is busy, and slower than it should be, with no single thing to point at. That is the honest reason this function is worth naming before you are big enough to staff it.

What We Write Down, Every Run

We run this function on ourselves, so the example here is our own rather than an illustration of somebody else’s. One artifact carries the argument better than a description would. Every time any part of our marketing system runs, it writes exactly one row into a log. The row has the same fields every time: which agent ran, the date, an outcome, a link to whatever was produced, a note carrying measured units and cost, and a summary of what happened and what needed a human eye. Nothing about that is sophisticated. A spreadsheet with those columns would do the same job. What makes it useful is that the row is written by the same act that does the work, rather than by a person remembering to record it afterwards. A record that depends on somebody circling back is a record that is accurate for about three weeks. The outcome field is the part worth stealing. It has five values, and only one of them is “success”. The others are partial, failed, skipped, and the one that changed how we think about this: shortage logged.

Why an Empty Queue Is a Logged Outcome

“Shortage logged” means the run happened, found nothing to do, and stopped. It is not a failure and it is not a gap in the record. It is the system saying, on the record and with a date, that there was no work waiting. That sounds like bookkeeping pedantry until the first time you need it. An unowned function produces silence in exactly two situations: when nothing needed doing, and when something needed doing and nobody did it. Those two look identical from outside, and telling them apart after the fact is usually impossible. A logged shortage separates them permanently. It also removes the incentive that quietly ruins measurement. If the only recordable outcome is success, then every run is under pressure to produce something, and a system under that pressure will manufacture work to have something to report. Making “there was nothing to do” a first-class, logged result is what stops a cadence from turning into activity for its own sake. Our log carries less flattering entries too, including a batch of generated images that were rejected wholesale with the spend recorded beside them. That entry, and what it cost, is written up on the team structure page rather than repeated here. The general point stands without it: a function that only records its wins is not measured, it is advertised.

Where to Start If Nobody Owns This

One move, not a maturity model. Pick the recurring marketing thing that costs you the most hours and produces the least argument, and write down how it currently happens. Not how it should happen. How it does. Who starts it, what they need before they can start, who approves it, where the output goes, and what tells you afterwards whether it worked. That document will be shorter than you expect and will have at least one hole in it that surprises you. The hole is the finding. In most small businesses it is the last item — nobody has ever defined what “it worked” means for that piece of work, which is why the question of whether to keep doing it has never been answerable. Then give that one written process an owner and a date. Not the whole function. One process. The reason to start this small is that the function is not a thing you install; it is a habit of writing down how work happens, and habits start on one thing. The owner does not have to be a new person, and at your size it usually is not. It has to be a named person rather than a role, because roles do not notice when something slips. Give them the authority to change the process they own, not just to execute it — an owner who has to escalate every adjustment is a router, and the work will route around them within a month. You will know the habit has taken when somebody asks how a recurring piece of work happens and the answer is a document rather than a person’s recollection. That is the whole of it. Everything else in this category is an elaboration on writing down how the work gets done and then reading it back.

Do It Yourself, Automate It, or Hand It Out

Once the work is named, the question of who covers it becomes answerable, and it splits three ways. Some of it stays with you, because it is judgment and it needs your context. Some of it is mechanical enough to automate, and the line between those two is the subject of how to use AI in marketing — the short version being that machines are safe on work where being wrong is cheap and unsafe on work where being wrong reaches a customer. Some of it is better bought from outside, and if that is where you land, the questions worth asking are in do I need a marketing agency and the comparison in marketing agency vs freelancer. All three answers are fine. What does not work is leaving the question unasked, because the default answer is the one you never chose: it lands on whoever has time, in whatever order it arrives. If you want a sense of how businesses your size are actually operating and outsourcing this work, the Census Bureau publishes the underlying data in its Business Trends and Outlook Survey. The Bureau labels that survey an experimental data product, so read it as directional rather than settled.