At BETC in Paris, Valérie Maman runs traffic for around 200 creatives, with 18 traffic managers under her, and when AdForum asked her what the job is she put it in one sentence: "You need to put the right person, on the right brief, at the right time, to make sure budgets and timings are respected." Each of her traffic managers, she said, can be overseeing 300 or 400 projects in a typical week. That is two jobs under one title. The first is judgement, knowing who should be on what, and when. The second is counting, the timesheets and timelines agencies have needed somebody to keep ever since they started selling hours, and the counting is the part automation takes away. It's why, in our own sales conversations, that seat is where warm leads go cold, and why the people sitting in it, of all people, should be the ones who shape what replaces the counting.
That seat is older than most of the software that now claims to do it. The traffic manager has always opened the job numbers, built and rebuilt the timelines, chased the stalled work and, as one agency software maker lists the duties, helped management judge whether the agency could absorb the new business an account director had just promised.
Tim Williams, who ran agencies before he spent decades advising them at Ignition Consulting, put it plainly in a presentation on value pricing: "Agencies were not always paid based on hourly rates. We introduced this system ourselves in the mid-1980s, and our clients learned it so well that they now play it back to us relentlessly."
Once time was the thing being sold, somebody had to count it.
The people who lived through it don't remember it fondly. Rick Webb wrote a whole chapter on it for his book Agency, and his editor cut most of it for being, in his words, "too vitriolic". He published the outtake anyway, starting with the 1990s: "once a month or so, someone would come by my cubicle and nag the entire interactive department to fill out their time sheets. If you were good at avoiding this person, you could go 2–3 months without doing a time sheet." His verdict is harsher than anything I would write: "They are, first and foremost a massive fraud, contentedly performed and affirmed by all parties in the ecosystem — the agency, the client, and the employees. Finance knows it. Your client knows it. Your client's finance department knows it. Your boss knows it. His boss knows it. Everyone knows it, no one cares." Darryl Sparey, co-founder of the PR agency Hard Numbers, remembers his last agency the same way: "typically 50% of folks hadn't filled a time sheet out by the end of each month. There was one Account Director who worked on the company's biggest client who hadn't filled out a timesheet for ten years!"
The defence deserves a proper hearing, because it's a real one. Nick Hand, an agency finance director, made it on TrinityP3's Managing Marketing podcast: "There's always, in any business, demand for scarce resources and that's the same in an agency. And management needs to know, well, what are the people that we need for the tasks that we've got coming up? And the time sheets are critical to helping manage that process." He's right. The timesheet was never only an invoice, it was the closest thing an agency had to a map of who was doing what, and the resource manager was usually the one person who could read it.
Which is why the pattern in our own sales conversations at ARTISH doesn't surprise me. A CEO or senior manager sees the product, agrees it solves a real problem, and then the conversation moves one desk over to the resource manager or project lead and goes quiet. I've seen it in a small number of conversations, which makes it a pattern rather than a finding, and some of it is plainly timing, because budgets freeze and client emergencies eat months. But the stalls seem likelier where the daily work is exactly what automated workflows absorb.
None of that is a complaint about the people in those seats.
Consider what they're hearing. Earlier this year, WPP's chief financial officer, Joanne Wilson, told analysts that "a commercial model that is more closely linked to client outcomes will enable us, over time, to move away from time and materials and decouple revenue from headcount." If your job is scheduling the headcount, that sentence lands differently. Put it next to the two stories everyone has heard about AI, that it will take your job and that it will do everything, and slow-walking a workflow vendor is a reasonable call. It's just made on a bad map. In my experience, anyone who has pushed one of these models hard on real work learns fast that it needs someone standing over it who knows what good looks like.
So it helps to separate the role from the tax. Building the Gantt chart, rebuilding it when a client moves a date, chasing Rick Webb round the cubicles, being the only person who can reconcile the timesheets: that is the tax paid to hold the seat between clients, freelancers, finance and the CEO, and automation removes it. What's left is what Maman described, the right person on the right brief at the right time. Knowing which client is about to blow up, which freelancer is overloaded and won't say so, which deadline is real and which one exists because somebody needed a date on a slide. Hand's own defence makes the point. The data was always in service of a judgement about people, and the judgement never lived in the spreadsheet.
I should be honest about the trade. Part of what makes the role hard to replace is that nobody else can read the timesheets, and a well-built system dissolves some of that. I can't promise that nobody's headcount changes, and anyone who does is selling something, usually to the CEO. What I can point to is the other seat. Somebody has to set the rules the system runs on, the milestones it tracks and the triggers that pull a human back in, and whoever does that owns the system instead of operating it.
The status quo isn't on the table, though, because someone above them will eventually buy something. But is infinitely better to shape the next system, than inherit it.