The observation
A few weeks ago, the company I work at kicked off a campaign cycle. The marketing team has a hard rule for its designers: two image fixes per week, every week, no exceptions.
I do not think the rule was set with bad intent. The rule was set by a manager who wanted to make sure the designers were producing. The manager was, in the manager's own reading of the work, doing what a manager is supposed to do. The rule was clean. The rule was measurable. The rule looked like accountability.
I have been thinking about it, on and off, ever since. The rule is the textbook example of a failure I have been trying to name on this site for a year. It is failure mode B of the KPI problem — the one with teeth — applied to a creative team. It is also, in the same breath, the textbook example of the dirty-work problem — the one that happens when the things that cannot be counted get squeezed out by the things that can.
Let me write down what the rule actually does.
What the rule actually does
A quota of "two image fixes per week" looks, on paper, like clarity. The designer knows what to do. The manager knows what to check. The form can be filled. The number can be reported. Everyone can be held to it.
What the rule actually does is encourage four things at once. None of them, on its own, is the designer's fault. All of them are downstream of the rule itself.
It encourages Goodhart's law to close. The line I keep quoting is the standard one: when a measure becomes a target, it ceases to be a good measure. The "two image fixes" stops being a proxy for "the team is producing design work" the moment it becomes the goal. The designer, who needs to ship two, picks the two that are safest to ship. The designer picks the two that take the least time. The designer, in the worst case, picks the two that are easiest to defend in the weekly review, even when neither of them is the thing the campaign actually needs. A campaign that needed one breakthrough piece and one polish ends up with two safe pieces. The campaign, in the worst case, lands flat. The number says the campaign is on track.
It encourages the squeezing-out of dirty work. A designer at the company has, on top of the fix list, a pile of support work that does not show up in any number. Re-exporting an asset at three aspect ratios. Adapting a layout for a screen size that no one planned for. Sitting in a cross-department call that goes nowhere for an hour. These tasks are real. These tasks are necessary. These tasks do not count. The designer, who has to hit two, has an incentive to push the un-counted work to the end of the day, and an incentive, in private, to be quietly relieved when the un-counted work gets canceled. The work that does not get counted is the work that does not get done.
It decouples design from the outcome. Two high-quality campaign images that lift engagement and two "I-just-need-to-get-this-done" fixes look, in the quota, identical. The quota counts the cards. The quota does not count the campaign. The number is, in the language of the KPI piece I wrote earlier, the wrong proxy for the right thing.
It turns the designer into a defender of the job boundary. "Two fixes" becomes the unit. Anything outside the unit is, by construction, extra. The designer, who wants to be seen as hitting the unit, stops volunteering for the things that are not the unit. The team, in the worst version, ends up with everyone hitting their own number and the overall campaign looking like a five-person job done by five people who never talked.
I have been hearing versions of this story from people I work with. A version of it was told to me this week, in a way I will not repeat, because the person who told it is not me to quote. The pattern, in the version I was told, was the same pattern I have been seeing for a year: a hard number, a soft real-world result, and a quiet erosion of the work that the number was supposed to be a stand-in for.
The principle, in one sentence
Creative work is not piecework. A small creative team is not a factory floor. The job of the manager is to set the outcome, draw the boundary, and trust the team inside the boundary.
I have been holding this principle for a few months. The hard part is not the principle. The hard part is the discipline to not reach for the number, in the moment the number feels like accountability.
What I have come to do instead
Here are the four moves I keep coming back to, in the team I actually run. None of them is original. All of them are the only moves I have found that survive contact with a real creative team for more than a quarter.
One. Project or goal orientation, not piecework. I do not tell my overseas team to ship a number of posts, or a number of designs, or a number of fixes per week. I tell them what the outcome looks like — engagement rate, qualified inquiries, growth on a specific market — and I let them decide the cadence. The number, when it is set, is the number of the thing we are trying to move. It is not the number of the artifacts we are trying to produce. The two are, more often than not, inversely related. The artifact number goes up; the outcome number goes down. The KPI piece is the longer version of this argument; it is older than this one and has the structural case.
Two. Reserve a buffer for the work that does not count. Some of the most important work in a social team has no number. Replying to the cold lead in a comment. Curating the next meme. Doing the copyright sweep. Tagging the assets in the DAM. The work is the work. The work is, by construction, the part no proxy will ever reward. The team that has to hit a quota will, in private, hate this work. The team that has been given a goal will, in my experience, do this work in the cracks, because the cracks have room for it. I have stopped asking the team to log the un-counted work. I have started asking the team to log only the counted work, and to give the un-counted work the room it needs.
Three. Blur the job boundary on purpose. One of the more interesting things I have come to think about, over the last year, is the value of a generalist — not in the sense of "knows a little about a lot of things," but in the sense of "can ship competently across a wide span of disciplines." The social team I work with is expected to know copywriting, light video, light design, community management, and trend reading — not at the level of a specialist in any of them, but at the level of a person who can ship a finished post in any of them, with the help of a few good tools (Canva, CapCut, Figma). The generalist, in my experience, ships better work on any one of those tasks than a specialist who has been cut off from the rest of the workflow, because the generalist can see how the post will land. The specialist, in the worst case, ships their two assets and walks away. The generalist ships the same two assets, in the same week, and they land differently. The reason is not that the generalist is more talented. The reason is that the generalist has the context.
Four. Protect the white space. A version of this is in the piece on the 16:00 break, and I am going to keep re-stating it here. A creative team that has to sit at the desk for the whole afternoon produces less than a creative team that has been given permission to be off the desk for a part of the afternoon. The permission is not a perk. The permission is the part of the workflow where the work actually gets done. The team that has been told to "look busy" produces the form of busy — the hours in the chair, the status pings, the visible artifacts — and not the work. The team that has been trusted, by contrast, takes the walk, comes back, and ships. I have watched this play out, in different shapes, for two years. The shape that works is the one where the manager has decided to trust the team, and the team has decided to take the trust seriously.
What I am not saying
I am not saying metrics are wrong. I am not saying quotas are always wrong. I am saying that for a small creative team, the proxy is the wrong tool for the wrong job. The proxy is the workaround for the manager who cannot see the work. In a small team, the manager can see the work. The proxy, in that case, is a tax on the manager's afternoon and a tax on the team's afternoon, and the work that the proxy was supposed to measure is the part that gets squeezed into the cracks.
I am not saying this is easy. The easy version — set a number, check the number, hold the team to the number — is the version I have watched fail in three different teams, in three different verticals, over the last two years. The hard version — set a goal, give the team room, watch the work directly, take the call when the call needs to be taken — is the version I have watched produce. The hard version requires a manager who is, in fact, watching.
I am also not saying this is original. The instinct against piecework on creative teams is older than I am. The hard part is the moment a manager is asked, by their own manager, to "show me the number," and the manager has to decide what to do in that moment.
Where this connects
The piece on KPI is the direct precondition. The piece argued that KPI is a workaround for the manager who cannot see the work, and that the workarounds tend to fail in two predictable ways. The two-image rule is a third way the workaround fails — a way I did not quite name in that piece, because I had not yet watched it play out in a creative team. The way it fails is the way the workarounds always fail: by counting the wrong thing harder.
The piece on dirty work is the closest cousin. The argument there was that dirty work belongs to the system, not the person. The argument here is the same argument, applied to the unscheduled work — the work that a creative person is supposed to do, that does not appear in the count, and that gets pushed aside the moment the count gets tight. The unscheduled work is the most important work, and the count is the thing that pushes it out.
The piece on the 16:00 break is the postscript. The argument there was that the creative afternoon is the most reliable production block of the day, and that the half hour in the middle of it is the cheapest way to make that block produce. The argument here is the precondition for that argument: a team that is being held to a hard count does not have the white space the 16:00 break needs. The 16:00 break is a creative-person policy. The two-image rule is a piecework-person policy. The two cannot both be true.
The paved-paths piece is the systems-level version of the same argument. The platform engineer's job is to make the right thing the path of least resistance. The manager's job, in a small creative team, is the same job: to set up the system — the goal, the boundary, the buffer, the white space — so that the right kind of work is the easy kind of work. The team, in both cases, is not being told to ship more. The team is being told to ship the right thing, and to be given the conditions to do it.
What the literature already says
I did not invent any of this. The framing is borrowed, openly, from work that has been around much longer than I have.
Goodhart's law, in the form I keep quoting — when a measure becomes a target, it ceases to be a good measure. The line is older than the AI era, and the pattern is older than the line. The pattern shows up every time a number gets welded to a workflow, and the workflow adapts to the number. The two-image rule is the pattern in the small.
Michael Polanyi, The Tacit Dimension. Polanyi's argument that we know more than we can tell is the deeper layer. The creative team holds things in their gut that no proxy can capture. The taste the designer brings to the campaign. The instinct the social lead has for the comment thread. The pattern the editor sees in the data. The proxy cannot measure any of these things. The proxy can only measure the things that can be measured. The manager who has decided to trust the proxy has, in Polanyi's reading, given up on the tacit layer. The manager who has decided to trust the team has, in Polanyi's reading, kept it.
Mihaly Csikszentmihalyi, Flow and Creativity. Flow requires a clear goal, immediate feedback, a balance between challenge and skill, and freedom from interruption. The two-image rule provides none of these. The rule is, in Csikszentmihalyi's reading, hostile to flow — and the team that produces the form of busy, instead of the work, is, in Csikszentmihalyi's reading, the team that has been put out of the flow state, and that has been given a number to hit instead of a problem to solve.
Donald Schön, The Reflective Practitioner. Schön's argument is that the work of the professional is the recognition of the situation, the framing of the problem, the naming of the trade-offs. The two-image rule removes the recognition. The rule replaces the framing with a count. The rule, in Schön's reading, is the version of the system that has decided the professional is interchangeable with the next professional, and that the situation is interchangeable with the next situation. The professional, in Schön's reading, knows the situation is not interchangeable. The count, by construction, does not.
I am not going to re-explain them here. The point is the same point.
A postscript, on managing a small team
I wrote most of this sitting in the back of a meeting that I should probably have been paying more attention to. The meeting was, by the standard of the room, productive. The meeting was, by the standard I am trying to articulate here, mostly theatre.
The hardest part of managing a small creative team, in 2026, is not the work. The work is mostly learnable. The hardest part is the moment the manager is asked, by the system, to produce a number, and the manager has to decide, in that moment, whether the number is the work or a workaround for the work. The default answer, in the systems I have worked in, has been the number. The answer I am trying to hold, more and more, is the work.
I am not always sure I hold it. I am getting better at it. The improvement is the only thing I have to point to, at the end of the year, as the work I have actually done.
The hardest part is not the principle. The hardest part is the moment the system asks, and the manager has to say no, and the manager has to be ready to defend the no, in the language the system understands, in the meeting the system has scheduled, in the form the system has decided is the form.