What did the media buyer’s work process look like back in 2016? They could spend three straight days building a single spreadsheet — pulling numbers from six ad platforms by hand, matching campaign IDs, checking for typos in UTM tags. Friday came, and half the data was already dead. That was just the deal back then. Nobody questioned it — not like we had a choice anyway.
Now? That whole thing takes about twenty minutes, tops. Half of that is just babysitting the automation to make sure it didn’t do anything stupid. What changed isn’t just “we got better tools.” The whole logic of how advertising gets planned, built, and optimized has shifted underneath everyone’s feet, and a lot of people in the industry are still catching up to that fact.
The old setup was never meant to scale
The old way was a classic assembly line: strategy → creative → media → and finally, two weeks too late, a report from analytics. Each handoff added lag. And each platform — Meta, Google, TikTok, programmatic DSPs — had its own dashboard, its own naming conventions, its own quirks that someone had to memorize.
What’s interesting is that the bottleneck wasn’t really about talent. Agencies had smart people. The issue was that a human brain, no matter how sharp, can’t process bid adjustments across forty ad sets in real time while also reading creative performance signals and adjusting budget pacing. AI systems can, and they don’t get tired doing it at 2 a.m. on a Sunday when a campaign suddenly spikes.
The part where ML finally pulls its weight
You hear “AI in advertising” so often now it’s lost all meaning. Here are a few areas where the tech actually changed results:
- Predictive bidding — platforms like Google’s Performance Max and Meta’s Advantage+ now adjust bids per auction based on hundreds of signals, something no human trafficker could replicate manually even with unlimited time.
- Creative testing at scale — instead of running two ad variants for two weeks, systems can now test dozens of headline-image combinations simultaneously and kill underperformers within hours.
- Audience modeling — lookalike and predictive LTV models built on machine learning outperform manual segment-building in most controlled tests agencies have published, particularly for e-commerce verticals.
- Anomaly detection — catching a tracking pixel failure or a sudden CPA spike within minutes rather than discovering it in a Monday report, by which point the budget’s already gone.
None of this is science fiction. It’s mostly pattern recognition applied at a speed and volume that human teams physically cannot match. The uncomfortable truth for a lot of agencies is that the “gut feel” media buyer — the one who could “just tell” which creative would work — is competing against systems that process more signals in a minute than that person sees in a year.
The workflow layer nobody talks about enough
Everyone loves discussing the flashy stuff — generative creative, AI copywriting, automated video ads. The middle — campaign management, cross-platform reporting, briefing — is less glamorous, but it’s where things actually get done.
The unsung hero here isn’t the algorithm—it’s the orchestration layer that connects brief to launch to report, so you don’t have to be the human bridge between six different screens. Some vendors — AdFactory included — position workflow automation as a way to boost campaign throughput and handle the tedious cross-platform syncing that used to eat entire afternoons. Sure, it’s not as exciting as “AI does your creative for you.” But ask anyone who actually runs campaigns for a living, and they’ll tell you — this middle layer is where the real time gets freed up.
Agencies don’t lose clients because their creative was marginally worse. They lose clients because reporting was late, or a budget pacing error went unnoticed for three days, or nobody caught that a campaign was still running after the promo ended. Boring failures, expensive consequences. That’s why platforms like AdFactory are of paramount importance.
What this means for the people doing the work
Naturally, this raises the question everyone’s actually thinking about: does this mean fewer jobs? The honest answer is nuanced. Junior roles built entirely around manual reporting and campaign setup are shrinking — that part’s real, and pretending otherwise doesn’t help anyone plan their career. The roles that actually require human judgment — brand positioning, creative direction, client relationships — are trending up, not down. The machine still needs someone to tell it what winning looks like.
There’s also a subtler shift worth mentioning — trust calibration. Early on, a lot of media buyers didn’t trust automated bidding at all, and honestly, some of that skepticism was earned; early algorithms weren’t great. Now the skill isn’t “should I trust the AI,” it’s “under what conditions should I override it, and how quickly.” That’s a genuinely different competency than the one media buying required a decade ago.
Where this probably goes next
Predicting the exact shape of things in five years is a fool’s errand; this industry moves too fast for confident forecasts. A few things you can actually bank on: attribution modeling is going to keep moving away from cookies and toward modeled data; creative is going to get faster, start to finish; and the agencies that stick around will be the ones that treated AI not as a feature to add, but as the foundation to build on.
If you take one thing away from this, let it be this: don’t run after every shiny new AI product that claims it’ll transform your ad business overnight. Most won’t. Instead, look hard at where the actual time drains are in a given workflow — the boring, repetitive, error-prone parts — and start there. That’s usually where automation pays off fastest, and it’s rarely the part anyone posts about on LinkedIn.
