The media industry has become increasingly sophisticated in how it measures effectiveness. Econometric modelling, brand tracking and optimisation tools have improved our understanding of advertising’s contribution and how future investment might be better directed. Quite rightly, clients expect us to demonstrate that their advertising spend is working and to use that evidence to make better decisions.
But are we becoming so good at understanding what worked yesterday that we’re making it harder to discover what might work tomorrow?
Consider two competing media recommendations. The first is a conventional plan using established channels and approaches. The second involves a genuinely inventive use of media, perhaps an unexpected placement or a clever use of context. The first comes equipped with years of historical evidence, while the second won’t — it can’t.
So which is more likely to secure investment and which is more vulnerable when budgets come under pressure (budgets always get cut)? We have a problem when we confuse an absence of historical evidence with evidence that something won’t work.
There’s a certain irony here. Over a century ago, Claude Hopkins published Scientific Advertising, arguing that advertising should be grounded in evidence rather than opinion. But he also advocated testing new approaches before committing larger sums.
A century later, there’s a danger that we’ve become better at using evidence to justify what we already do than at using experimentation to discover what we could do better.
Does invention actually make a difference?
A 2024 meta-analysis published in Journal of Advertising by ZMC van Berlo and colleagues examined 26 studies comparing unconventional physical media with traditional formats. The researchers found positive effects on brand associations, attitudes, purchase intentions and word of mouth.
They also identified surprise as an important mechanism explaining these effects, with stronger results when the medium had a meaningful relationship with the message.
Using media differently can therefore change how people respond to advertising, particularly when the execution reinforces the brand’s proposition rather than simply providing novelty.
The research doesn’t establish that every inventive campaign will outperform a conventional alternative, nor does it demonstrate incremental sales effects. But it does suggest that the medium is more than simply a vehicle for delivering advertising; how it’s used can contribute to effectiveness.
What are our models actually telling us?
Suppose an econometric model suggests linear TV has historically generated a stronger return than outdoor advertising. That’s useful information, but it’s important to understand exactly what we have (and haven’t) measured.
We’ve estimated the effectiveness of campaigns represented in our historical data, reflecting their creative executions, placements, investment levels and wider circumstances. We haven’t established that TV will always outperform outdoor or what a genuinely different execution of either channel might achieve.
An inventive approach may generate effects that simply aren’t represented in the historical dataset.
More sophisticated econometric models can incorporate creative quality and context, provided sufficient information and variation exist. But there’s an important distinction between optimisation and discovery: the former helps us make better decisions about approaches we already understand, while the latter requires us to explore those we don’t.
If we continually direct investment towards previously observed winners, those approaches accumulate more evidence and become easier to justify, while genuinely new ideas struggle to demonstrate their value. Over time, this risks reinforcing our existing assumptions rather than challenging them.
Making invention an intelligent investment
Of course, asking clients to be braver is all very well, but it’s worth considering what we’re asking them to risk.
A CMO recommending an established approach can point to years of historical evidence when challenged by the CEO or CFO. An inventive recommendation may offer greater potential returns, but without that evidence it can be considerably harder to defend.
Historical return on investment estimates carry uncertainty too, even if we sometimes treat them as rather more definitive than they deserve. The danger is that we end up favouring the most defensible investment rather than the most effective one.
This is where measurement can help us make experimentation less risky. Hopkins had the right idea a century ago, arguing that we should “constantly seek for better methods, without interrupting plans already proved out”. His approach involved testing campaigns in a limited number of locations, sometimes just “four or five towns”, before committing to wider investment.
We can apply much the same principle today: develop hypotheses about why an inventive execution might deliver additional value, then test them with manageable investments and agreed measures of success. Where practical, that means comparing an inventive approach with a conventional alternative, rather than simply measuring whether a campaign worked.
The question is whether the invention itself made an additional contribution and whether that contribution justified the investment.
At Goodstuff, we’re building on existing evidence while investing in further research to understand the value of invention. We’re also using postcode-sector modelling to construct geographic test and control groups, helping us test the incremental impact of different media approaches.
Meanwhile, our forthcoming research with The Harris Poll will investigate how inventive media generates attention, emotional engagement, meaning, amplification and memory, and how these responses relate to brand and behavioural outcomes.
The purpose of measurement shouldn’t be to make us increasingly confident in doing the same things. It should be to help us manage uncertainty well enough to discover something better.
