■ HIGH RISK ■ Sales & Marketing
For digital inventory, it largely already has — programmatic platforms and AI bidding do the buying. What's left of the human job is strategy, negotiation for premium placements, and explaining to clients why the algorithm spent their budget the way it did.
“Programmatic advertising buys in milliseconds. Your three-week negotiation is adorable.”
Our AI replacement risk score — how we score jobs
The classic media buyer negotiated rates with TV stations, radio networks, and publishers, built insertion orders, haggled over added value, and reconciled billing when the spots didn't run as booked. The modern version spends most of the day inside demand-side platforms — The Trade Desk, DV360, Meta and Google's ad managers — setting audiences, budgets, and bid strategies, then watching dashboards to see what the machines did overnight.
Here's the uncomfortable part: the buying itself was automated years ago. Real-time bidding executes millions of auctions per second, and the platforms' own AI — Google's Performance Max, Meta's Advantage+ — now increasingly automates the targeting and creative rotation decisions that used to justify a buyer's salary. Campaign setup that once took a week of trafficking is becoming a prompt and a budget cap. The platforms are openly building toward advertisers stating a goal and letting the black box handle the rest, which disintermediates exactly the hands-on-keyboard work most buyers do. That trajectory is why our risk score sits at 66 despite the field feeling busy right now.
The defensible ground is judgment and accountability. Someone has to decide which channels deserve budget at all, negotiate sponsorships and premium video that never touch an open exchange, catch the platform grading its own homework with inflated attribution, and take the client call when performance craters. Brand-safety disasters, made-for-advertising site sludge, and measurement disputes all require a skeptical human who understands the plumbing well enough to know when it's lying. Buyers who become cross-channel strategists and auditors of the machines will do fine; buyers whose job is toggling settings inside a platform are training their replacement with every campaign.
Automatability: our editorial assessment of current and near-term AI capability
Pressure is here now and compounding. Platform automation like Performance Max and Advantage+ is absorbing targeting and optimization decisions today, and agencies are consolidating buying teams accordingly. By around 2030, expect hands-on campaign management roles to be sharply reduced, with remaining jobs concentrated in strategy, premium deal-making, and auditing automated systems. Traditional linear TV buying declines with its medium on roughly the same schedule.
The execution layer is dying; the judgment layer isn't. Platforms have automated bidding and are now automating targeting and optimization, so the number of people needed to run a given ad budget keeps shrinking. But brands still need humans to set strategy, negotiate premium deals, and audit the platforms' self-serving reporting. Fewer seats, more strategic ones.
The mechanical buying is already algorithmic, and platform AI is absorbing campaign optimization right now. By around 2030, goal-based automation will likely handle most standard digital campaigns end to end. What resists longer: premium negotiated inventory, cross-channel budget strategy, and accountability when things go wrong — the platforms very much do not want that last job.
Measurement science first — incrementality, media-mix modeling, attribution's failure modes — because auditing automated buying is the durable job. Second, negotiation and relationships for inventory that trades on deals, not auctions: sponsorships, retail media, premium CTV. Third, the AI tools themselves, deeply enough to explain to a client what the black box is actually doing with their money.
Because someone has to be accountable. Automated platforms optimize toward their own metrics, grade their own homework, and happily spend budget on junk inventory if unsupervised. Clients pay agencies for strategy, oversight, and a human to blame. The role is shifting from operating the machines to supervising and second-guessing them — a real job, but one needing fewer people.