It’s time to trust in the media metrics that actually matter
Opinion
If we want better media decisions, measurement needs to reflect how people actually make buying decisions, not simply how platforms record them, writes Dan Gee.
As observed by Rory Sutherland, “It’s easier to have a successful career being measurably mediocre than immeasurably brilliant.” That neatly sums up one of marketing’s biggest challenges. We have become exceptionally good at proving activity, but not always at proving how that activity boosts growth.
Media has a wealth of data to work with. We can measure a campaign through impressions, clicks, views, readership and conversions. This shows whether people are seeing and engaging with advertising, and whether campaigns are being delivered as intended.
It’s useful, but incomplete.
Google, Meta and Amazon have made buying media, optimising campaigns and producing performance reports remarkably frictionless. But just because something is easy to buy and easy to measure doesn’t automatically make it the right investment.
What’s often missing is an explanation of how all of this contributes to growth. In essence: true media effectiveness.
It’s not that marketers are making poor decisions. Measurement structures are naturally weighted towards immediate feedback and what can be attributed easily. Hard numbers are attractive, particularly when marketers need to demonstrate results to the rest of the business.
The problem comes when we confuse a metric’s availability with its importance.
Stop relying on the dashboard
Advertising rarely works in the neat, linear way our dashboards imply.
Take Rolex. Nobody decides to spend thousands of pounds on a luxury watch because they clicked on a display advert last Tuesday. That purchase results from years of accumulated influence. Brand campaigns, sponsorships, editorial coverage, and repeated exposure all build familiarity, trust, and desire long before someone is ready to buy.
The final interaction might receive the credit in an attribution report, but it certainly didn’t do all the work.
Single-lens attribution, and even more sophisticated multi-touch models, only capture fragments of the customer journey. They are useful for understanding interactions and optimising activity, but shouldn’t be mistaken for a complete account of what created demand.
If we want better media decisions, measurement needs to reflect how people actually make buying decisions, not simply how platforms record them.
Model, experiment, track
There will never be one universal measurement that proves marketing works. Nor should we expect one.
A more useful approach is to combine different types of evidence, each answering a different question: model, experiment and track. The aim isn’t perfect attribution. It is better investment decisions that lead to stronger business results.
Model: What will help us drive growth?
Start with the commercial problem. Are we trying to grow revenue or market share? Increase penetration? Reach a new audience? Improve profitability?
Then look at broad patterns over time to understand what appears to be driving those outcomes. That might involve econometrics and MMM, but it can also mean looking intelligently across sales, market share, penetration, brand health, media investment, search behaviour and wider market conditions.
Yorkshire Tea provides a useful example. Its sustained brand-building strategy improved preference and recommendation, increased penetration and helped volume share grow from 18% to a market-leading 33%, while maintaining premium pricing. Econometric analysis covering 2019–2023 attributed £25.8m of incremental revenue to the strategy.
The important point isn’t simply that advertising generated a return. It is the chain of evidence: brand strength, penetration, market share and revenue moving together over time.
Experiment: What did this activity actually add?
Where modelling looks for broader patterns, experiments allow marketers to test specific hypotheses.
That might mean leaving comparable geographic areas unexposed, switching a channel on and off in a deliberate pattern, or introducing activity progressively across different markets. The principle is simple: create a credible comparison that helps distinguish incremental impact from activity that might have happened anyway.
A car retailer client of MFM recently took this approach. MFM tested the activity across eight dealerships while leaving two comparable locations untouched. The experiment generated an eight-times return, which translated to strong seven figures of incremental sales.
More importantly, it gave the business greater confidence in the commercial value of the investment. That is what good measurement should do: reduce uncertainty around the next decision.
Track: What is happening now?
Finally, marketers need signals that show whether things are moving in the right direction.
That includes the basics: reach, frequency, cost, attention and media delivery. But it should also include signals closer to the intended outcome: awareness, consideration, branded search, direct traffic, enquiries, new customers, sales and market share.
Attribution belongs here too. It is useful for understanding customer journeys and optimising activity, but it should be treated as one source of evidence rather than the final verdict.
Fast-growing brands such as JAECOO show why this broader view matters. Search interest, awareness, enquiries, registrations and market share can together provide a far more useful picture of momentum than campaign clicks or video views viewed in isolation. None of those metrics proves causation on its own. That is precisely why tracking works best alongside modelling and experimentation.
Measurement should make every campaign smarter
Measurement is about much more than proving whether one campaign worked. Every campaign should leave the business knowing something it didn’t know before.
That requires marketers to become more comfortable with different forms of evidence. Model the broader drivers of growth. Experiment where you need to understand incremental impact. Track the signals that tell you whether the business, brand and media are moving in the right direction.
Advertising is probabilistic, not deterministic. It influences behaviour over time, across multiple touchpoints and in ways that can’t always be neatly attributed to a single interaction.
Brands that optimise only for what platforms can report will ultimately optimise for platform success. Those numbers remain useful, but they are only part of the picture. And not the growth part.
The objective of measurement should be to make better decisions. The best measurement system is the one that leaves you more confident about where to spend the next pound, euro or dollar.
Dan Gee is co-founder and chief strategy officer at media agency MFM