The scale of retail media growth is quite hard to get your head around. Some estimates say it will account for a fifth of all US ad spend this year.

But it’s easy to understand why advertisers have jumped in. The retailer provides the media and the sales figures (this is the so-called closed loop), allowing advertisers to link their ad spend to sales performance.

You can usurp your competitors through a sponsored search ad on a retailer’s website. Or you can target CTV ads at people who you suspect might be into your new protein-packed snack bar (by using a retailer’s carefully curated audience of low-sugar or nutrient-dense purchasers, for example).

However, as retail media expands, it is undergoing some growing pains – one of which is a lack of sophistication in measurement, and an accompanying lack of transparency.

As advertisers rationalise their budgets, they want to know how to get maximum impact from their retail media spend. Here, the closed loop can be a double-edged sword (forgive the mixed metaphors). When the media provider attributes sales like this, the concern is to what extent they are marking their own homework.

David Pollet, CEO at retail media measurement business Incremental, highlights this dilemma of predominantly last-touch attribution that is “siloed to inventory source”, with no appreciation of the influence of other media platforms.

Brands that don’t have the measurement savvy are likely getting a blinkered view – they may be using return on ad spend (ROAS) figures provided by a retailer whose picture of the world includes only its own ads.

Retail media spend up 12% in 2023 as digital landscape shifts 

ROAS and the ‘selection effect’

This isn’t a slight on retailers, it’s just a reminder that measurement that looks at the customer in the round is tricky to get right. And it’s by no means a new phenomenon for advertisers. Last-touch attribution and the consequent over-indexing of a particular media channel is an old foe.

Marketing Week columnist and VP of brand strategy at Jellyfish, Tom Roach, wrote about the danger of ROAS (and ROI), and its role in encouraging short-term thinking, back in 2022.

Going further back, readers may be familiar with the ‘selection effect’ – the idea that, in this instance, online advertising is often aimed at users who are already warming to the sale, and who may have gone on to purchase regardless.

Think of brand terms in paid search – when a shopper has typed your brand name into a search engine do they really need to be served a PPC ad in order to find your site when the organic listing is right there? Or a shopper who has already browsed a product, then gets retargeted around the web with display ads – if they then come back and buy, how much is it really to do with those ad impressions? There are plenty of caveats in these scenarios, but the underlying idea is that ad platform algorithms are effectively looking for people who are likely to click or go on to buy.

This isn’t a slight on retailers, it’s just a reminder that measurement that looks at the customer in the round is tricky to get right.

Of course, this works both ways. There are also ad formats where ROAS can be superficially low. Skye Frontier, VP of growth at Incremental, namechecks retail media formats that may have low ROAS but can grow sales incrementally, including upper funnel and off-site ads such as a video ad on YouTube, sponsored brands (a big colourful ad unit that the customer notices but doesn’t push one specific product) and conquesting (targeting competitor brand searches on a retailer site).

Frontier has detailed how misleading ROAS can be, writing about how the metric can in fact decline as a CPG brand optimises towards incremental sales on Amazon. In the case study, Frontier writes that “ad spend increases and topline sales grow at an even faster rate, implying an increase in ad-spend efficiency (every dollar invested is generating more topline growth than it has previously). ROAS paints a very different picture, declining 17% during the same period.”

Will retail media bridge the gap between performance and brand marketing?

Off-site inventory brings new challenges

The industry is getting to grips with this demand for better measurement. You can see it in the content marketing from the ad platforms, the work on standards by the IAB and ISBA, and a wellspring of new measurement consultancies.

Criteo’s latest report was titled ‘The ROAS Trap’ and begins as follows: “You’re probably familiar with retail media’s ability to significantly boost conversions. But beneath the surface, there’s a range of metrics and KPIs that brands can use to understand just how much their spend is paying off throughout the funnel. In short, there’s way more to retail media than ROAS.” The report goes on to discuss metrics such as ‘new to brand’ and ‘share of sales’.

IAB Europe ran a survey last year which found that 70% of buyers cited lack of standards as a barrier to investment, with media and attribution measurement the most important areas to address.

If advertisers aren’t clear on where their ads are appearing on the web, this is where the selection effect can really come into play.

Real growing pains have been felt though from retail media’s expansion off-site. There are bigger tech stacks involved, and the lines between shopper and brand marketing teams blur.

Off-site inventory is a big part of retail media’s continuing growth – an October 2023 forecast by eMarketer predicted off-site will account for 18.5% of US retail media ad spend in 2024. But it throws up questions such as how to gauge its impact on in-store sales, and it inevitably brings some of the problems the programmatic world has been dealing with.

If advertisers aren’t clear on where their ads are appearing on the web, this is where the selection effect can really come into play. A retail media network could theoretically serve ads to a known shopper on a low quality website (ad placements that may look like a cheap form of reach on the balance sheet but don’t do anything for the brand), and then snag credit for a subsequent conversion a week or two weeks later on the retailer website using view-based ROAS metrics.

So, just as effective advertising higher up the funnel can be undervalued by ROAS, so too poor placements can be greatly overvalued.

Indeed, this type of activity was the focus of an Adalytics report at the beginning of this year. The firm conducted an analysis on behalf of a Fortune 500 brand and found they had spent over $10m on made-for-advertising (MFA) websites in the second half of 2023, across programmatic, private marketplace, direct buy, social media, and retail media investments. In one section, the report states that “many major brands appear to have their ads placed on MFA sites specifically via retail media networks”.

The challenge for advertisers –  a more joined-up world

As the industry tries to get a grip on standards, there is continuing fragmentation as more retailers create and add to their media offerings.

The off-site proposition only seems to get more complex, as more partnerships are formed.

Read Amazon’s recent pronouncement on addressability (targeting specific audiences) and ad tech and you’ll see some keywords that brand advertisers are looking for – simplifaction, control and transparency.

The task for teams on the brand side is to understand media not simply on a channel by channel basis. Retail media demands more integration, bringing together performance, brand and ecommerce teams to better understand the customer purchase journey and the incremental impact of every ad format amongst the myriad now on the table. 

Ben Davis is insights editor at Econsultancy, which provides e-learning, live-learning online workshops and skills mapping in digital, marketing and ecommerce.