Auction pricing, algorithmic feeds, skippable formats and the polish penalty. The campaign model wasn't wrong, its assumptions expired.
The campaign operating model is one of the great pieces of commercial machinery. Built for broadcast, refined over decades, it knew exactly what it was doing: concentrate the budget behind one big idea, pre-test it carefully because the bet is large, negotiate the media months ahead, flight it, measure it afterwards. Every step was rational, because every step rested on assumptions about how media worked. Media prices were fixed in negotiation, audiences were captive in an ad break, production was scarce and expensive, distribution was controlled by a handful of channels.
Paid social is now the largest media channel in the world, $306 billion in annual spend, more daily time than any other channel, and it kept the budget line while quietly breaking all four assumptions underneath the model, not weakening them but breaking them, at the level of mechanics. Each break carries its own operating implication, and together they describe why the inherited model underperforms on the channel, regardless of how talented the people running it are.
The first break is economic. Broadcast media was priced in advance, which meant creative quality could not change what an impression cost. On social, every impression is won in a real-time auction that weights your bid by how people are predicted to respond to the ad itself. Work that people watch wins impressions cheaply, work that people skip loses auctions it would otherwise have won and pays more for what it still gets. The price of media is no longer fixed before the creative exists, the creative moves it, in both directions.
The operating implication: creative is working media. Decisions about the quality, variety and freshness of the work are media-efficiency decisions, with effects measured in CPMs and cost per completed view, not just in brand terms. A model that treats creative as a fixed cost to be minimised is optimising the wrong line.
The second break is competitive. In an ad break, your ad competed with other ads, and exposure was guaranteed by the schedule. In a feed, an algorithm assembles every person's experience individually, and your ad is ranked against everything, the group chat screenshots, the creator they love, the news, the friend's holiday. Nothing guarantees you the slot. The feed gives attention to whatever earns it, and it knows with unsentimental precision what each person tends to watch.
The operating implication: the competitive set is not your category's other ads, it is the best thing in each person's feed. Work built to win an ad break, polished, formal, announcing itself as advertising, is competing in a contest that no longer exists. The bar is interestingness, and it is set by content, not by competitors.
The third break is behavioural. Broadcast audiences were captive enough that planning could treat exposure and attention as the same thing. Social formats are skippable after a second or two, and the swipe is the default behaviour, not the exception. Attention went from something media bought in bulk to something creative earns one viewer at a time, second by second.
The operating implication: the served impression is no longer the unit that matters, the attended one is. Structure changes, the hook comes first, the brand arrives early, the payoff lands fast. Measurement changes with it, completion and view-through become the numbers that tell you whether anything was actually communicated, the delivered count never did. Media creates the opportunity to see your ads. Creative determines whether they work.
The fourth break is cultural, and it is the one the industry has been slowest to absorb. For most of advertising's history, production value carried a signal. Only a real company could afford a film crew, so polish itself communicated scale and trustworthiness. Then the phone in everyone's pocket became a production studio, and billions of people became fluent in made-for-feed content. In that world, polish stopped signalling quality and started signalling ad, and the audience's learned response to ad is avoidance. The skip happens on the production style alone, before the idea, the offer or the brand has a chance to land.
The operating implication runs in two directions. Creatively, the work needs the grammar of the medium, real people, real environments, native pacing, or it triggers the immune response no message survives. Operationally, the same democratisation that broke the signal also collapsed the cost of making work, which is what makes a portfolio of twenty to thirty distinct creative routes affordable at all. The break and the opportunity are the same event.
It is worth being precise about what this adds up to, because the conclusion is not that the people running campaigns made a mistake. The campaign model wasn't wrong. It was a correct answer to a set of conditions, fixed prices, captive audiences, scarce production, controlled distribution, and it was refined into something close to optimal for those conditions. The conditions changed, the model kept running, because operating models outlive the assumptions they were built on, especially when they are embedded in org charts, agency rosters and approval chains.
What the new conditions require is not a better campaign but a different shape of operation entirely, a portfolio of distinct creative hypotheses rather than one pre-tested bet, production native to the medium rather than adapted to it, and a weekly loop of pause, remix and replace decisions rather than a flight plan, because the channel now produces data daily that the quarterly cycle was never built to use.
Each of the four breaks points to the same place. When the work sets the media price, earns the slot, wins the attention and carries the targeting, creative stops being an input to the plan and becomes the thing the plan is built around. The brands winning on the channel are not the ones that fixed the old model, they are the ones that noticed it was answering a question the medium no longer asks.