Retail attribution is broken – why incrementality is the fix marketers have been waiting for

As e-commerce and bricks-and-mortar retail continue to blur, marketers are facing a growing blind spot: measuring the offline impact of digital campaigns. A new whitepaper from independent media agency Magic argues that the problem lies in the flawed assumptions behind traditional attribution - and proposes a more rigorous alternative: incrementality.

Titled The Retail Attribution Illusion – how Incrementality is the Answer to Omnichannel Commerce, the whitepaper calls time on so-called “vanity metrics” and attribution systems that overstate digital performance while failing to connect marketing efforts to true business outcomes.

“The industry is still chasing impressions, clicks and conversions, but those aren’t the numbers that matter,” Magic CEO Shah Ghaffurian says. “Just because someone clicked an ad and bought something doesn’t mean the ad caused the purchase. They might have bought it anyway. That’s the difference between attributed conversions and incremental sales.”

Magic’s argument is simple but pointed: attribution models are not only incomplete, they’re actively misleading - especially for omnichannel businesses where the path to purchase rarely begins and ends online. In contrast, incrementality looks to measure the actual lift a campaign generates beyond what would have happened naturally, providing a more accurate understanding of marketing’s real-world impact.

Marketing needs its maths moment

Ghaffurian’s frustration is clear. “Marketing is one of the last professions to have the rigour of maths and stats applied to it,” he said. “Stocks had it first, then gambling, then banking. Marketing has never really had those kinds of brains to solve these problems - or at least, not for very long.”

That gap in rigour is exactly what Magic, founded six years ago in Melbourne, set out to close. Since launch, the agency has expanded to Los Angeles and London, grown to a 25-person team and won global accounts on the strength of its performance - not its pitch decks.

The agency’s approach marries media buying with advanced modelling, combining experimentation, econometrics and real-time measurement to optimise channel mix and creative strategies. Clients like Sophos, for instance, initially worked with Magic in APAC before expanding the brief globally after consistently outperforming their holding company rivals.

“It’s not just about models. It’s about using them to inform actual media decisions – and then validating those decisions with real-world experiments,” Ghaffurian said. “That loop of test, learn, validate and optimise - that’s where the magic happens.”

From military intelligence to marketing analytics

Ed McMutrie, head of Signal and author of the whitepaper, brings a unique background to the task. A former military intelligence analyst, he now applies his skills in behaviour analysis and pattern recognition to solve marketing problems. He joined Magic after stints at ANZ, Officeworks and several agencies, drawn by the opportunity to integrate data science directly into the day-to-day practice of media buying.

By using this approach, marketers have a much clearer understanding of how their media efforts impact total sales, both online and offline, allowing them to make more informed decisions on where to allocate budgets for maximum return.

Ed McMutrie

“I completed various roles across the New Zealand defence force including a stint in an analytics function, so numbers and behaviour patterns came naturally. When I left the military, I didn’t know what I wanted. I had to find a job in civilian life pretty quickly. I found analytics and the logic clicked,” McMutrie said.

“In the military, the process we went through when it came to analysing people and patterns and their behaviours is 100% transferable to marketing. Now it’s not about finding the bomb maker – it’s about figuring out who’s going to buy the groceries. Incrementality is all about looking at the incremental value driven by marketing efforts. It accounts for non-linear and complex customer journeys, making it ideal for measuring the impact of digital media on in-store sales."

For Ghaffurian, the key is having data science embedded in the whole media buying process. "For example, if you spent $1,000 on search every day for past three years, you wouldn't be able to tell what impact that had on incrementality or the bottom line. You need variation – like over or under-spending for a period – to draw meaningful insights.

“We use that initial model to get a rough picture, then plot inferences – like, TV looks strong based on historical data, so let’s try a heavy TV burst in one market and validate it. Once we’ve run those incrementality tests, we rerun the model to prove effectiveness.”

This, he says, helps to overcome a common issue that businesses only half-believe what a model tells them. “But when you validate it in market, prove the uplift and then feed that back into the model, buy-in skyrockets,” he says. “Then we take that successful approach and roll it out nationally, while testing the next thing. It’s an active, iterative approach.”

Why incrementality matters more than ever

At the heart of the whitepaper is the claim that legacy attribution models - especially those powered by black-box platform algorithms - are increasingly unfit for purpose. Platforms like Meta and Google optimise toward attributed conversions, often inflating their role in the purchase journey and ignoring long-term, offline or delayed effects.

Ghaffurian gives the example of a client with 80 to 90% of revenue coming from physical stores - yet only online performance was being measured. “If someone sees an ad and walks into a store, you don’t know if the ad caused that. But incrementality helps answer that question,” he says.

Incrementality testing allows marketers to understand what really moves the needle - not just what looks good in a report. It supports smarter media planning, more confident experimentation and more credible conversations with the C-suite.

Shah Ghaffurian

“Ultimately, we’re helping marketers speak the language of the business and I think that's where our unique approach brings a lot of value. If the business has an objective of 10% growth, marketing has an objective of brand awareness or online conversion, digital team might have an objective of video views - you’ve got a disconnect.

“At each of these stages they are all trying to talk the language of business, but the metrics they measure aren’t the same as business objectives. If we can help detangle that and get everyone on the same page - suddenly everything aligns no matter what team they are on.”

A new model for media

Magic’s blend of consultancy-style modelling and practical media buying has found traction with performance-led brands, especially those looking to bridge the gap between digital and physical retail. But it also challenges some industry norms, particularly around media measurement, creative evaluation and agency remuneration.

Creative, for instance, is not an afterthought. “Fifty percent of media performance comes down to creative,” Ghaffurian said. “We talk about it a lot because you can have the best data and strategy in the world, but if the creative doesn’t land, nothing works.”

Looking ahead, Magic wants to see incrementality become the default lens for media performance - especially in omnichannel contexts. That means training marketers to test rigorously, refresh models regularly and challenge assumptions baked into platform algorithms.

McMutrie puts it simply: “If your model is based on bad inputs - like inflated attribution - it doesn’t matter how advanced it is. You’ll make bad decisions faster. Incrementality isn’t just a better metric. It’s a safer one.”

McMutrie says using this approach gives marketers a much clearer understanding of how their media efforts impact total sales, both online and offline – allowing them to make more informed decisions on where to allocate budgets for maximum return.

“We’re already seeing a shift away from legacy attribution models. Businesses, particularly those with bricks-and-mortar stores, need to move beyond simplistic tracking methods and embrace incrementality to truly understand the impact of digital media. Measuring both online and offline conversions is no longer a nice-to-have – it’s fundamental to understanding performance.

“The focus now for marketers should be on testing different media tactics across audience, content, creative and treatment – to identify what drives the highest incremental lift in both online and offline sales, including leading metrics such as foot traffic and store visits.”