ROAS Is Not Enough: The Actual Metrics Behind Retail Media Success

[tldr]

TL;DR: ROAS is not enough—and it never was

• Retail media has evolved into a full-funnel discipline, but many organizations still evaluate success using lower-funnel metrics.
• ROAS remains an important KPI, but it shouldn't be the only one driving investment decisions.
• Brands should complement ROAS with incrementality, new-to-brand acquisition ocst, customer lifetime value (LTV), and marginal profitability.
• AI is making better measurement possible, but organizational alignment—not technology—is often the biggest barrier.

[/tldr]

Retail media has changed dramatically over the past several years, especially recently with increased availability of Amazon Marketing Cloud and visibility into customer shopping paths.

What was once a marketing channel centered around sponsored products has evolved into an ecosystem spanning retail media networks (RMNs), DSPs, connected TV (CTV), AI-powered optimization, and increasingly sophisticated audience measurement.

Yet many organizations continue measuring success using the same KPI they've relied on for years:

Return on ad spend (ROAS).

To be clear, last-touch ROAS is still an important metric. But as retail media expands across the customer journey, marketers need measurement frameworks that capture more than just conversions. 

Let’s cover how ROAS became the default, why it’s never been enough, and which additional metrics brands should focus on today.

[tldr]

Watch the full panel discussion

These insights are based on a panel discussion, Designing Full-Funnel Retail Media in the Age of AI, at Signal to Scale, Xnurta's retail media conference, featuring:

• Mike Sinclair (VP, Head of Agency Development, Criteo)
• Sha Atakhanov (VP of Marketing and Ecommerce, Boiron)
• Tom Kluis (GM, Amazon and Ecommerce Marketplace Practice, Right Side Up)
• Joon Choi (Chief Revenue Officer, Xnurta)

The discussion has been expanded here into broader recommendations for retail media leaders.

[/tldr]

Where ROAS can measure—and what it can’t

ROAS remains one of the most useful indicators for evaluating lower-funnel efficiency. The problem is treating it as the only measure of success, which many orgs still do. 

Why marketers still rely on ROAS

• Easy to calculate
• Easy to communicate to leadership
• Effective for measuring demand capture
• Useful for comparing campaign efficiency

Where ROAS falls short

• Doesn't measure demand creation
• Undervalues upper-funnel investments
• Doesn't account for customer quality
• Ignores lifetime value
• Typically only includes the last touch
• Doesn't consider profitability or contribution margin

"ROAS is important, but it shouldn't be the North Star. Once you look at new-to-brand acquisition, cost of acquisition, and shopper cohorts, you get a much clearer view of what's actually working." — Tom Kluis (GM, Amazon and Ecommerce Marketplace Practice, Right Side Up)

Full-funnel retail media requires full-funnel measurement

The biggest issue with overreliance on ROAS (and last-click attribution in general) is that consumers don't experience marketing in silos.

They discover products through multiple channels before purchasing, often interacting with several touchpoints along the way.

Modern retail media platforms such as Amazon Marketing Cloud increasingly make it possible to understand those interactions: Instead of evaluating awareness campaigns separately from conversion campaigns, marketers can begin connecting the entire customer journey.

Tom describes this shift as thinking about the funnel as a spectrum rather than separate stages, a mindset that enables brands to optimize the entire shopping journey instead of isolated tactics.

Traditional vs. full-funnel measurement

‍

Four metrics that belong alongside ROAS

Rather than replacing ROAS, leading retail media teams are expanding their scorecards, and they’re invariably prioritizing four metrics:

• Ad incrementality
• New-to-brand acquisition cost
• Customer lifetime value
• Marginal profitability

1. Ad incrementality

Caveat: this isn’t as much a single metric as it is a mindset, but it’s such a vital part of modern measurement that we needed to mention it.

At its simplest, focusing on incrementality requires consistently supplementing the question:

"Did these ads generate sales?"

with:

"Would those sales have happened anyway?"

By doing that, brands understand whether media is actually influencing customer behavior. 

"Incrementality may be a buzzword, but it should become one of the main goals of retail media campaigns." — Mike Sinclair (VP, Head of Agency Development, Criteo)

[tldr]

Master incrementality testing

Want to get better at incrementality testing? We’ve got several resources to help beginners and experts alike:

• Read our article to get started with incrementality testing.
• Follow it up with our complete guide to marketing experimentation where we cover every tactic and test type in detail.
• Once you’re comfortable with advanced measurement concepts, watch this on-demand webinar to learn how to build your own AI-powered Media Mix Modeling (MMM).

[/tldr]

How to measure incrementality (the short version)

In general, there are lots of ways to measure incrementality, and while the type of test you choose depends on your goals and resources, the general framework is usually as follows:

1. Compare an exposed audience with a similar control group.
2. Measure lift in sales, conversions, or new-to-brand customers.
3. Focus on incremental business outcomes, not just attributed revenue.
4. Repeat tests regularly to validate performance across channels.

Quick-and-dirty, ad-specific version for retail media marketers: What’s your organic brand share on a keyword? Take 100% minus that number to get your rough incrementality on that keyword. 

2. New-to-brand acquisition cost (NTBCPA)

Not every customer has the same value: A campaign with modest ROAS may still outperform another campaign if it consistently acquires first-time buyers and those buyers behave differently over the long run

Questions marketers should ask include:

• How many first-time customers did we acquire?
• What was the cost per new customer?
• Which campaigns drive the highest-quality customers?
• Which audiences become repeat purchasers?

3. Customer lifetime value

Lifetime value fundamentally changes acquisition economics.

If customers purchase repeatedly over time, brands can justify much higher acquisition costs than first-order revenue alone would suggest.

So rather than optimizing for a single purchase, marketers should evaluate long-term indicators like:

• Repeat purchase behavior
• Subscription rates
• Customer retention
• Revenue over time

4. Marginal profitability

Understanding just marketing metrics isn’t enough to inform marketing investments. The cost of your first unit and the cost of your nth unit are likely very different. Brands that understand marginal costs and marginal profits can more effectively scale with smart investments in ads.

To that end, some non-marketing metrics to focus on include:

• Marginal production costs
• Inventory availability
• Contribution margin
• Customer acquisition costs
• Operational capacity

"Before brands scale full-funnel media, they need to understand their numbers—marginal cost, acquisition cost, lifetime value, and whether they can actually support the demand they're creating." — Tom Kluis (GM, Amazon and Ecommerce Marketplace Practice, Right Side Up)

Driving demand is only valuable if the business can fulfill it profitably.

The metrics that matter most

‍

Better measurement leads to better decisions

Expanding measurement changes everything for the better, including investment decisions. 

What we’ve found is that orgs often discover that campaigns with weaker short-term ROAS can sometimes have long-term viability, and help brands:

• Acquire more new customers
• Generate higher lifetime value
• Improve lower-funnel conversion
• Increase overall business growth

Of course, this isn’t always the case. Some low ROAS campaigns are simply inefficient and should be scrapped. The key is being able to tell the difference between the two, and that’s where better attribution comes in.

One more thing: Enhancing an already solid set of metrics with AI can help you produce even stronger insights, but only if your data is high-quality and your goals are crystal-clear. Get those right before building any workflows.

The biggest challenge often isn't technology

Most retail media organizations already have access to more data than ever before, but still struggle with operational alignment: they still completely separate disciplines like ecommerce, retail media, shopper marketing, brand marketing, and sales.

That fragmentation makes unified measurement—and ultimately better decision-making—far more difficult. The panel agrees: Bringing those teams together around shared business outcomes is often more impactful than adopting another AI tool.

What’s next in retail media?

A few predictions from the panel:

• Retail media will continue becoming more sophisticated.
• AI can provide a depth of analysis that would take teams days or hours to complete, but teams need to challenge the findings and identify what’s real and what’s not.
• Measurement will become more connected.
• Retail media networks will continue expanding their capabilities.
• The brands that gain the greatest competitive advantage won't simply adopt new technology—they'll adopt better ways of measuring success.

And if there’s one takeaway that bears repeating: ROAS still deserves a seat at the table, but it shouldn’t be sitting at the head of it.

[tldr]

Want to build a measurement strategy that goes beyond ROAS? 

Right Side Up helps brands design full-funnel retail media programs that connect AI, measurement, and activation to profitable growth. 

Whether you're scaling Amazon, Walmart, or other retail media efforts, our team can help you build a strategy that measures what matters. 

Contact us today →

[/tldr]