On auction-priced media, creative quality sets your CPM. The working versus non-working split collapsed, and budgets haven't caught up.
For decades the media plan came pre-divided into two categories. Working spend was the money that bought airtime, impressions, the actual delivery of the message to a human being. Non-working spend was everything else, production, fees, the cost of making the thing in the first place. Procurement's job was to push the working share up and the non-working share down, and the logic was sound. When the price of an impression was fixed in a negotiation months before the ad ran, creative could not change what a rating point cost. The work was a fixed input, the media was the variable, so you optimised the variable.
That split made sense for fifty years, on paid social it has quietly inverted, and most budget conversations have not caught up.
Every impression on a social platform is sold in a real-time auction, and the auction does not clear on your bid alone. The platforms are not selling space, they are selling their users' time, and their entire business depends on people staying in the feed. So the auction weights your bid by predictions of how people will respond to your ad, the chosen objective's outcome plus quality and engagement signals. An ad people watch, finish and engage with sends strong signals into that calculation and wins impressions at a lower effective price, most directly when attention itself is what you are buying. An ad people skip loses auctions it would otherwise have won, so every exposure it still gets comes at a higher effective price.
This is the mechanical fact that collapses the working and non-working split. The quality of the creative sets the price of the media. Two brands with identical budgets, identical audience settings and identical flight dates will pay different CPMs, reach different numbers of people and earn wildly different amounts of attention, and the only variable separating them is the work. Creative is not a cost sitting outside the media budget. On auction-priced platforms it sits inside the media cost, deciding what every dollar buys.
The effect is not marginal, and it compounds. In one account we measured, the brand's existing creative completed at a 0.44% view-through rate and a cost per completed view of $0.50. A portfolio built and optimised as a creative system, same brand, same budget, same platform, reached 2.24% completion by week three at $0.09 per completed view. One account, so it is an illustration rather than a law, and part of the gain is the loop doing what loops do, retiring the weakest work and extending the strongest. But the decomposition is the point. Most of that movement is the completion rate itself, more of every thousand impressions converting into watched ads, and the rest is the auction rewarding the stronger signals with cheaper delivery. Both levers are creative levers. The media buying never changed.
Whiskas showed the same mechanics with the two levers visible separately, a 48% lower CPM and a 237% higher completed view rate against the brand's existing benchmark. The media agency did not negotiate that discount, the creative earned it, impression by impression, because the auction rewards work that holds people in the feed.
These numbers move weekly when the loop is running, because a creative system does not launch and hope. Every week the portfolio faces its pause, remix and replace decisions, the weaker ads come out, the stronger patterns get extended, and the average quality of the live portfolio rises. The auction notices, and the cost of attention falls. That decay from $0.50 to $0.09 was not a lucky ad, it was the system improving the portfolio while the data flowed.
Here is the budget conversation this changes. Suppose you want more reach next quarter, and you have two levers available.
Lever one is the familiar one, increase the media budget. A 20% increase buys at best 20% more impressions at your current quality-adjusted prices, and in practice somewhat less, reach curves are concave, the deeper you push into the auction the more the incremental audience costs. If your creative is being skipped, you are buying 20% more skips at rising prices.
Lever two is the one the old split taught us to minimise. Improve the creative, and you change what every impression you were already buying converts into. Moving cost per completed view from $0.50 to $0.09 means the same media budget delivers more than five times the completed views. No plausible media increase produces that. You could not negotiate it, you could not bid your way to it, and it applies to the full spend, not just the increment.
The strange conclusion, which the mechanics fully support, is that the cheapest reach available to most brands is hiding in the creative line. A brand spending heavily on media and minimally on the work has optimised the small lever and ignored the big one, not through any failure of judgment, but because the operating model it inherited was built when the levers genuinely were that way around.
None of this argues for spending recklessly on production, it argues for retiring a category error. The question "what percentage of budget goes to creative versus media" assumes the two are separate pools, one working and one not. On auction-priced platforms the real question is different. What price is our work earning in the auction, and what would it be worth to improve it?
A useful exercise is to pull your account's cost per completed view and model what a 30% improvement would do to delivered completed views across the full media spend, treating the result as directional, gains do not extrapolate perfectly to every dollar. For most brands the number is still large enough that the entire creative budget looks small beside it. That is the working media lever, and it has been sitting in the line item the old model told everyone to squeeze.
None of this repeals the oldest law in media planning. Share of voice still predicts share growth, and brands that grow still tend to hold voice at or above their share of market. What the auction changed is what produces voice. The same budget now buys a different share of the category's attention depending on the work, which means creative quality multiplies what every media dollar contributes to the growth equation rather than substituting for it. The split between working and non-working media described a world where the message and the delivery could be priced separately. That world is gone on the channel that now takes the biggest share of the plan.
Media creates the opportunity to see your ads. Creative determines whether they work.