Unbaking The Cake

For a good, solid sixty years, advertising was a lavishly funded film industry that also happened to sell washing detergent, cereal and insurance.

The luxuriant budgets reflected the prominence of the venue. A spot ran in cinemas, on prime-time television, in the slot before the news, three times an evening, across seven nights a week. Tens of millions of people watched it. Many of them watched it many dozens of times. They sat through it with cups of tea on their laps and their families by their sides and they demanded that it be entertaining. The alternative being that they had to sit through it anyway and feel annoyed at the interruption.

So the spots were suitably cinematic. Directors flew in from Sydney. Production companies built sets. Music was scored. Talent was cast. The numbers, looked at through a spreadsheet with media numbers on the other axis, made absolute sense. £500k for development for £500k of production for £9m of media buying is absolutely proportional.

But yeah, then the venue went away.

It is typical for Adland people to be doleful or in denial about the death of TV advertising, but the doleful lamentations are unhelpful. TV hasn’t really died, it’s still there but just in an enshittified form. Advertising can’t really broadcast itself to thirty million people in lockstep now. It most often targets two hundred thousand people, half of whom are bots, the other half of whom are scrolling past it with the volume off while their actual attention sits on a parallel device watching something else. The asset, by the time it gets delivered, is competing with a funny pet video, a freshly baked sourdough loaf and a porcine middle aged man falling off a paddleboard. So, even the real life humans that aren’t fake digital bots aren’t really watching.

Every CMO panel since approximately 2017 has said some riff on this. What hasn’t been illustrated and scrutinised is what this does to the economics of the asset itself. With good reason, as illustrating and scrutinising these economies is existentially troubling to anyone working in the advertising industry.

If your spot is going to be glanced at by four thousand humans, it cannot be a lavishly produced film. The maths absolutely forbids it. £500k of creative development against £500k of production cost against a media buy of £250k is not a proportional ratio, it is self evidently a huge waste of money. The spreadsheet, that same spreadsheet that used to bless the lavish detergent-film, now refuses to countenance anything more expensive than a man with a phone in his kitchen.

Agencies love to say “You have to meet the audience where they’re at”, and what they tend to mean by that is you can’t just shove product features down someone’s neck. And it’s true, you do have to meet the audience where they are at. But our audience are now scrolling past your ad while they sit on the toilet. So the asset has to come down to meet the audience where they’re at. I mean both literally and metaphorically on the toilet.

This is already happening. Production budgets are halving and halving again. Directors of considerable repute are now making things on iPhones with two-person crews, they are calling it “platform-native” through gritted teeth. Agencies are moving, guided by marketer demand, from delivering one polished spot to delivering thirty variants, all on the grounds that the platform’s recommendation algorithm wants volume, not craft. The model under which a suave Sydney director with considerable name recognition was worth flying in for is falling apart.

The fairly sane response, and the one that adland has fallen into, is to mourn the craft and try to make everything cheaper. Cheaper directors, cheaper edits, cheaper music, cheaper everything. AI arrived at the perfect moment for this. Advertising never says cheap though, we use the word ‘efficient’.

AI is in many ways a technology that empowers the forces of capital to make things shitter but to call that shittiness ‘efficiency’ in such a way that their share price increases dramatically. Ironically it doesn’t seem to be very good at making things cheaper, unfortunately it is very good at making things shittier.

I think the efficiency narrative is the wrong response. Partly because i don’t like doing a shitty job, not on purpose anyway. But more so because I think it misreads where the value was living the whole time in the process. So yeah, it’s both wrong and leads to shit. Not ideal.

Here is the contrarian thesis: The asset was never really the product.

The asset was the bit you could see, hold, screenshot, stick in the agency’s case-study reel and submit to Cannes. It was the bit that was legible. The part that a procurement team could write a line item for. And yeah, it was also the bit people saw in the real world too.

The value-generative work, the brief, the strategic thinking, the idea, the feedback, the third-round refinement of a single line in a script, the four routes that got killed so that the fifth could be born, the long nights of arguments that happened in a meeting room at 11pm two months before anyone shot anything, that was always where the value was being minted. The asset was just one very visible part of the outcome of that work. People mistook the most visible outcome for the whole shebang because that is just what people tend to do. Human beings, especially when arranged in matrixed organisational systems, are highly, highly biased to overvalue legibility.

There wasn’t any deception in that though. Nobody was deliberately lying about anything. The bundle just worked. Media spend was concentrated enough to justify paying for both the hard work and the final asset as a single invoice, and the fortunate consequence was that nobody, neither agencies, clients, nor procurement departments, had to have any embarrassing or difficult conversations about which part they were really paying for. The cake came out of the oven and you bought the whole thing. If you wanted to cut it into slices after the fact, then that was fine.

What’s happened in the past decade is that media-spend, once concentrated in a single expensive to produce channel in such a fashion that creating an asset for it could subsidise all the important upstream work, has now fragmented into myriad line items, each subject to tyrannical scrutiny and microscopic aperture bean counting. Agencies are being asked to do a thing that is not, on close inspection, possible: separate the eggs from the flour, retroactively, for just one individual slice of the cake at a time, and to tell procurement what each one cost.

You cannot unbake the cake.

Long-time readers of mine might recognise a whiff of what’s gone wrong here. The asset was the proxy. The thinking was the underlying variable. But, as they tend to do, the systems optimised against the proxy; production values, edit quality, festival metals, polish, and they did that for fifty years, until everyone honestly did believe that the proxy was the thing. But then the proxy collapsed in value, and all the systems that bundled the costs of the variables into the price of the proxy woke up and discovered they’d forgotten how to get paid for the bit that cost something.

This is our old pal Goodhart, again, or at least a riff on him. Make a measurement an objective and the underlying object dies. The objectives in question; production polish, asset quality, deliverable count, all of that is now coming up worthless on the market. And the body of work upstream of it, which was the actual thing that mints value, has been royally screwed by decades of being treated as overhead.

Right. So the asset is collapsing in value, the bundle has been torn open, and we cannot unbake the cake. What now?

Most of the prescriptions doing the rounds, charge for thinking, not making; reposition as a consultancy; sell strategy by the day, have been doing those very same rounds for fifteen years and have failed to land for fifteen years. And they keep failing to land for a simple reason: most clients don’t want to buy thinking. They want to buy artefacts that have been thoughtfully created. Artefacts are legible. Thinking is not. Procurement can scrutinise the RFP of a deliverable. It cannot scrutinise the RFP of a decision. The market for invisible labour, by an unbroken centuries-long tradition, is very small.

Experience should tell us “how do we sell the thinking?” is a bad question. The better one is “what is the new product?”. Three answers, in roughly increasing order of how sweaty they will be for the people running agencies:

1. Make fewer things, but be the ones who decide on what things. The value isn’t found in producing the next thirty assets. It is the judgment layer that decides which thirty, aligned against objectives, in what order, against what idea, killing which alternatives, and then producing those assets. This is the ‘Comms Strategy to the rescue’ option, the great upstreaming. And no, this is not a new thought, but it has been refused for a long time because clients probably want a neutral, objective steward of what it is that gets made and won’t take advice on what to make from an agency that is geared up to make specific types of things. And so that implies a significant reduction in headcount. That reduction is happening anyway, but because of AI rather than some shift in strategic positioning. I guess we are seeing this already with the growth in smaller, more scalable agencies.

2. Get paid against the brand, not the deliverable. Architects worked this out a long time ago. Their fee is a percentage of project value, not a per-drawing rate. Some version of equity, royalty or outcome-linked compensation is the nice honest answer to the problem of selling judgment, maybe the agencies brave enough to insist on it will end up looking far more like A24 than like WPP. Or maybe it’s never going to be possible to make this work with the current style of relationship agencies have with their clients. I can’t make this work without seeing the media and the creative in the same shop. That is definitely something clients are tickling around, lots of RFPs recently are explicitly looking for both.

3. Recognise that most agencies don’t survive this and shouldn’t. Urgh. The TV proxy bundle was holding up an industry. Maybe unbaking the cake is the end of days. This future has far fewer agencies, they’re much smaller, and they’re working much closer to brands, with strategic judgment cranked up and the production infrastructure cranked down. This is a doomy prediction. But it is what every other industry that lost its bundle, think record labels, magazines, cable TV, has eventually had to accept. Marketing’s turn was always going to come. And now it is here.

The £500k spot made sense when ten million people watched it six times a week. It does not make sense now, and no amount of cheaper directors or platform-native edits is going to make it make sense again. The asset was never really the big thing. The thing was the work that produced the asset, and the cakey bundling was the only mechanism clients had for paying for that work without admitting that they were paying for it.

We are being asked, by clients and procurement and their spreadsheets, to unbake the cake. We cannot. We never could. The agencies that get this, the ones that figure out how to stop trying to itemise the impossible to itemise and start asking what a new product actually is, will look very little like the agencies of the last sixty years. Chances are they will be smaller. I believe that they will be sharper and have a much higher talent bar too.

The rest will spend the next decade explaining what an egg was worth.