Why finding the true value of retail media is a tough nut to crack

Incrementality is one of the buzzwords within the itself-very-buzzy world of retail media. An advertiser may see their sales rise on a particular retailer’s website but wonder if this really represents additional revenue.

This idea is nothing new to marketers (“half of my advertising is wasted…”) but it’s particularly pertinent in retail media, where much of the advertising sits close to the goal-line, on the site or app where the purchase is made (the so-called ‘closed loop’). Yes, a particular shopper may have seen an ad and then added that product to their cart, but who’s to say they might not have done so anyway?

The IAB and Media Rating Council’s much welcomed Retail Media Measurement Guidelines published in September 2023 includes a whole section on this topic and offers the following definition:

“Incrementality measures the true value created by any business strategy, determined by isolating and measuring related results, independent of other potential business factors. In other words, incrementality is the potential causal impact of marketing.”

Recently, I spoke to David Pollet, CEO at the very aptly-named retail media measurement platform, Incremental, who described in more detail the measurement challenges faced by advertisers.

Legacy measurement approaches won’t work for retail media

“My belief is that [retail media] is not really totally media. Some of it is media and a lot of it is merchandising,” says Pollet, adding that legacy measurement approaches won’t work for retail media.

Traditionally, media buyers have used a combination of media mix modelling and lift testing to test correlation and causality, respectively (leaving aside multi-touch attribution, which Pollet describes as “a proven failure”).

“The measurement universe has accepted you need to have both. You need to have causality to do validation of your correlation, but you need correlation because it’s more ubiquitous and there are a lot of pockets in the world where you can’t do straight lift tests. It’s also really expensive to keep taking 10% of your advertising inventory and holding it out (to prove uplift).”

The problem is that retail media is different.

“The ads in retail media sit on a page right next to the checkout, right? Your price is there, your promotions are there, your competitor’s pricing is there, their promotions are there and Amazon (for example) has algorithms that are changing the position of everything on a regular basis,” explains Pollet.

How valid is that measurement approach if it is measuring an ad that is now right next to a new competitive offering?

This flux creates lots of variables that make lift tests pretty much impossible to do. Pollet offers a hypothetical scenario:

“Imagine you could [run a lift test on Amazon]. Then what? The best practice would be to run the test and say (for example) ‘this offers 2.2 return on investment’ and you would then apply that for the next 364 days of the year. But an hour later, a new competitor can enter the page and be running a two-for-one offer. So, how valid is that measurement approach if it is measuring an ad that is now right next to a new competitive offering?”

“You can’t measure onsite retail media the same way you’d measure a television ad,” says Pollet. “You need different data. You need different granularity. You need different speed and to accomplish that you need different methodologies because the old methodologies aren’t accurate at that pace.”

Attribution cannot be siloed to inventory source

The other challenge for advertisers is attribution, as Pollet puts it, being “siloed to inventory source”.

“The recipient of the advertising dollars is probably not the ideal candidate for providing measurement of the performance of the media that they sell you,” he explains, drawing parallels with other digital formats such as CTV, where providing both media activation and measurement is seen to be a conflict of interest, and where some companies that do so have sought to divest ad sales.

Pollet is at pains to say he is not implying “bad intentions”, but is highlighting the fact that in this fast-growing industry, retailers are going to build measurement solutions “around the inventory that they offer you”, even if the reality is that “customers don’t think that way.”

“Customers don’t think, ‘I’m engaging in with an ad right now that was targeted using past purchases of SKUs’,” he says. “That’s not what they think. They think – ‘I saw a really cool video ad for this. I’m gonna type it into Amazon. Oh gosh look, there’s a really convenient placement that shows me how to buy it.’”

“That’s a consumer experience. And so the measurement should track back to that.”

The Incremental CEO argues that advertisers have to rethink how they measure retail media at large but also how on-site retail media is measured now. “We believe,” he says, “like all measurement providers, you need to have exposure to all of the consumer touch points that you can get your hands on to understand their influence on the bottom of the funnel.”

Retail media funding is complex

Econsultancy has previously covered the size of the task for CPGs and FMCGs adapting to retail media (from team structures to capability building) and I asked Pollet to paint a picture from the advertiser perspective

“The complex answer is retail media is complex because of the way it’s funded,” he says.

“Often you have a committed amount of money you have to spend with, say, Walmart because it’s attached to other parts of your relationship. So, one challenge is your system has to figure out how to squeeze more value out of that set investment.”

“But what we provide to advertisers is the ideal media mix and… the amount of incremental sales you can expect. And then a measurement methodology for ensuring that it’s actually [happening]…”

But, once brands have this “academic” plan, Pollet highlights the seemingly mundane realities that may impact how efficiently marketers can execute on it. He cites difficulty in moving budgets around between teams, communication challenges, and multiple agencies as some of the challenges. “Very large enterprises have large constraints,” he adds.

Very large enterprises have large constraints.

Alongside what Pollet refers to as business constraints, there are strategic constraints.

“Buying your own branded keyword [in on-site search] may not drive a lot of incremental value, but if you’re Nike, for example, do you want Reebok there [ahead of you]?” he says.

Another consideration is customers shifting their buying channel.

“If I’ve been wearing Nikes for the last 20 years and I just woke up to the fact that it’s easier to have them come to my door (via ecommerce), then that’s not customer acquisition, that’s channel shift. There’s value to being available on that channel… You will lose me if you’re not, but that channel didn’t necessarily acquire me either and so I think for advertisers, there’s a lot of difficulty there.”

But make no mistake, retail media is big, and executives are calling the shots

Talking to Pollet – who is a veteran of martech and adtech having started out “when we called it internet advertising,” 25 years ago at LendingTree and then Bank of America – I want to get a sense of whether retail media’s growing pains are similar to those of more mature ad formats. Is there a parallel with the rise of paid search, for example? Have execs already tackled similar challenges? Or could retail media be doomed to make similar mistakes?

The difference, says Pollet, is that “[Retail media] is suddenly driving tremendous value for the brands. It is a must-have strategy,” particularly since the acceleration of ecommerce during Covid. Whereas “nobody paid any attention to search” when it first emerged in the noughties, before it became “this behemoth”.

Indeed, Emarketer figures show that UK retail media hit £3 billion in just seven years, whereas social media took 13 years and search 17 years to get to the same point. This year, retail media will account for a fifth of US ad spend in 2024, according to a forecast by Advertiser Perceptions.

Pollet expands on this difference, saying, “The executives making the decisions on retail media are executives. This isn’t a search guru.”

“[When I was in charge of digital ads in 2000], I had a propeller on my head. I sat in the corner and all of my money came because the direct mail people weren’t able to get the mail drop out in time and they said, ‘Spend this so Finance doesn’t know we didn’t spend it’.”

The executives making the decisions on retail media are executives. This isn’t a search guru.

Where Pollet does admit similarities is the danger of being “hooked on vanity metrics” which he says, “have the inverse effect of what you’re pursuing and what I mean by that is an algorithm goes out and tries to find where it can sell the most of your product, right? Well, if you’re a huge brand and you already have a lot of customers, the algorithm is going to start to find places where your customers are going anyway to purchase your product.”

“It is going to optimize away, right, and that existed in digital media. If you reel back 15 years, some of the biggest budgets were in Finance. So think about this, you’re back about 15 years ago on the internet. Who’s searching for a financial product on the Internet? Somebody who’s not receiving direct mails, who’s not getting offered them in the bank centre. It has adverse selection, right?

“And the algorithms… couldn’t even see approval rates. It was actually driving up costs because all they were doing was kicking people in who were never going to get approved…”

Accept the truth and “optimise to the future that you want”

“I see these algorithms doing damage to the advertisers’ performance because it is so much easier to go find people who are already gonna buy,” Pollet continues. “And the problem with that is you have to be a really strong leader to say we’re shifting away from this metric…

“You kind of have to be willing to accept the truth, to then go and optimize to the future that you want. And in that way, I think it’s very similar to [existing] digital advertising.”

Still, Pollet is sanguine when faced with this complexity, adding that Incremental is having a lot of success in driving value. He has a final word on how the industry is maturing:

“The maturity cycle that is occurring within commerce and retail media – [brands understand], we need to get to real metrics here and run a real business. We need to deliver back insights that the finance department would accept.”

“If you can make the finance person at the table smile, then you’ve accomplished something as a marketer, right? That’s decision grade insights.”