For about fifteen years, the promise of digital advertising was precision. You did not need a better ad, you needed a better list.
That promise is largely over, and it did not end because anyone changed their mind about it. It ended because the two things that made it work — the tracking signal and the manual controls — were taken away within about three years of each other. Part one of the ad creative series.
The ranking that keeps coming back
The most-cited effectiveness work on this is Paul Dyson's econometric analysis of what actually moves profitability. Thinkbox published the current version in 2023, built on roughly 28,000 campaigns. It ranks ten drivers by their multiplier effect on profit.
Brand size leads, but no campaign decides how big its brand already is. Read the list for things an advertiser can change this quarter and creative quality is first at 12x, with targeting last at 1.1x.
That is a different research tradition, a different dataset and a different outcome measure from the NCSolutions work on the hub page — which put creative at 49% of incremental sales and targeting at 11% across roughly 450 campaigns. Two methods, one ranking. When independent approaches disagree about magnitude but agree about order, the order is usually the real finding.
Then the signal went
Apple's App Tracking Transparency arrived with iOS 14.5 in April 2021 and required apps to ask permission before tracking a user across other companies' apps and websites. Most people said no.
Opt-in rates settled around 11 to 15% worldwide in the months after launch, per Flurry Analytics' telemetry. Meta's chief financial officer told investors in February 2022 that the change would cost the company roughly $10 billion in revenue that year — which is a useful number precisely because it is the platform's own estimate of what the lost signal was worth.
What that did to advertisers was subtler than "attribution got harder". It degraded the raw material of audience targeting. Custom audiences built on site behaviour got thinner. Lookalike audiences built on those seeds got less like anything. The precision instrument stopped being precise.
And then the controls went too
The platforms' answer was not to rebuild the targeting instrument. It was to remove the dials and do it themselves. Advantage+ on Meta and Performance Max on Google both take audience selection, placement and bidding inside the system and optimise against an objective.
This is generally reported as a loss of control, and it is. But look at what is left. If every advertiser in an auction is using the same automated buying against the same inventory with the same degraded signal, then the buying is not where anyone can differentiate. The remaining input that is entirely yours is what the ad says and how it looks.
That is what "creative is the new targeting" actually means, and it is more literal than it sounds. On a broad campaign with automation doing the delivery, the creative is the targeting instrument: the system learns who responds by watching who responds, and what it shows them is the only variable you set.
What that looks like in an account
My own test for this is deliberately blunt. Run a broad campaign with strong creative as a split test against the same campaign with conventional targeting — interest-based or a lookalike — and watch which side wins. Across medical clinics, travel businesses and business-to-business software companies, the broad side has been winning on the strength of the creative.
I want to be careful about what that is and is not. It is a method and a pattern I have seen repeatedly. It is not a published measurement, I have no aggregate number attached to it, and you should treat it as an argument for running the test in your own account rather than as evidence in itself. The published research above is the evidence. This is what acting on it looks like.
The practical consequence is a reallocation of effort. If targeting is worth 1.1x and creative is worth 12x, then the hours spent refining audience definitions are close to the least valuable hours in the account, and the hours spent producing and testing more creative are close to the most valuable. Most advertisers still have that backwards, because audience settings are visible, discrete and satisfying to adjust, and creative is work.
What to read next
- Ad hooks and attention — what the attention research establishes, and what it does not.
- Ad creative testing — how much creative you actually need, and how to tell a winner from noise.
FAQ
What is the evidence that creative outranks targeting? Paul Dyson's econometric analysis, published in its current form by Thinkbox in 2023 and built on roughly 28,000 campaigns, ranks ten drivers by their multiplier effect on profit.
What actually ended precision targeting? Two things, about three years apart. Apple's App Tracking Transparency arrived with iOS 14.5 in April 2021 and required apps to ask permission before tracking a user across other companies' apps and websites; worldwide opt-in ran at 11 to 15%. Then the platforms removed the manual controls.
Are Advantage+ and Performance Max simply a loss of control? They are, and that is how it is generally reported. But look at what is left: if every advertiser in an auction is buying the same inventory with the same automation against the same objective, what you hand the system is what remains to compete on.
Has Bytown tested this directly? The test used here is deliberately blunt — run a broad campaign with strong creative as a split test against the same campaign with conventional targeting, and watch which side wins. Across medical clinics, travel businesses and business-to-business software companies, the broad side has been winning on the strength of the creative. That is a method and a repeated pattern, not a published measurement.
Sources on this page
- NCSolutions, "Five Keys to Advertising Effectiveness", 2017 and 2023 editions. Roughly 450 consumer-packaged-goods campaigns; machine-learning attribution across digital and television.
- Thinkbox with accelero (Paul Dyson), "The Drivers of Profitability", 2023. Roughly 28,000 global campaigns.
- Flurry Analytics, iOS 14.5 opt-in tracking, 2021. Flurry software-development-kit telemetry; sample size not stated on the page.
- Meta chief financial officer's guidance on the iOS change, reported via CNBC, February 2022. An earnings-call disclosure, not an independent measurement.
- DataReportal citing Statista Market Insights, "Digital 2025: Global Advertising Trends". Digital share of total advertising spend, 2019 and 2024.
