Measuring Retail Media ROI: Metrics That Actually Matter
The Measurement Problem in Retail Media
Retail media's great promise has always been accountability. Unlike traditional brand media — where the link between a TV spot and a purchase is modeled and inferred — retail media sits close enough to the point of purchase that attribution feels tractable. A shopper sees a sponsored listing, clicks, and buys. The network logs the impression, the click, and the sale. Simple.
Except it isn't. The measurement environment in retail media is considerably messier than that clean story suggests. Networks use different attribution windows. They count conversions differently. They have different definitions of what constitutes a "view" for display. Some networks measure against total brand sales in the retailer's ecosystem; others measure only against clicks. And virtually all of them have an inherent incentive to show you numbers that make your campaigns look as good as possible.
None of this means retail media measurement is hopeless. It means you need to be deliberate about which metrics you track, how you interpret them, and which questions you're actually trying to answer with your data. Here's a framework for doing that well.
Tier 1: The Metrics That Go in the CMO Report
These are the business-level outcomes that justify retail media as a strategic investment — the numbers that matter to leadership and that connect media activity to commercial results.
Sales Lift
Sales lift measures the incremental revenue generated by a campaign above what would have happened without it. It's the most direct measure of retail media's contribution to the business, and it's the hardest to measure accurately.
True sales lift measurement requires a control condition — a group of stores, markets, or audience members who weren't exposed to the campaign — against which you can compare performance. The difference between the exposed group's sales and the control group's sales, holding everything else equal, is your lift.
Most retail media networks offer some version of sales lift reporting. Treat those numbers as directional, not definitive. The cleanest lift measurement comes from third-party studies or holdout tests that you design and control, not from the network's own attribution model.
New-to-Brand Buyer Rate
One of the most valuable things retail media can do is bring genuinely new shoppers into your brand franchise — people who weren't buying your product before the campaign. New-to-brand buyer rate measures the percentage of campaign-driven purchasers who hadn't bought the brand in a defined lookback period (typically 12 months).
This metric matters because it answers a question ROAS doesn't: is your investment growing the franchise, or just efficiently capturing the people who were going to buy anyway? A campaign with a modest ROAS but a high new-to-brand rate might be more strategically valuable than one with a high ROAS that's mostly converting existing buyers.
Share of Search / Share of Voice
Within the retail platform, your brand's share of the sponsored results for key category search terms is a leading indicator of competitive position. A brand that dominates sponsored search for "protein bar" on Instacart will see that advantage compound into organic search ranking over time as click-through rates and purchase rates feed the platform's relevance algorithm.
Tracking share of voice on your key search terms — and monitoring changes as you adjust your investment level — gives you a competitive benchmark that pure sales attribution can miss.
Tier 2: Campaign Optimization Metrics
These are the in-flight metrics that tell you whether your campaigns are performing well mechanically and where to focus optimization efforts. They don't go in the CMO report, but they drive the decisions that eventually show up there.
Return on Ad Spend (ROAS)
ROAS — revenue generated per dollar of ad spend — is the most commonly reported retail media metric and also the most commonly misread one. A high ROAS sounds great. But ROAS without context is nearly meaningless.
A sponsored search campaign targeting existing brand buyers will almost always show a higher ROAS than a display campaign targeting new audiences. The first campaign is fishing in a barrel — reaching people who were already likely to buy. The second campaign is doing harder work, with inherently lower apparent return. If you optimize purely for ROAS, you'll systematically underinvest in the campaigns that are actually growing your business.
Use ROAS as one signal among several, not as the primary optimization lever. Set ROAS targets by campaign objective — higher targets for retention campaigns, lower targets for acquisition and conquest campaigns — rather than applying a single threshold across your entire retail media program.
Click-Through Rate (CTR)
CTR is a useful diagnostic metric but a dangerous optimization target. A high CTR means your creative and placement are generating curiosity. It says nothing about whether those clicks are converting into purchases or whether the people clicking are actually your target audience.
CTR is most useful as a creative performance signal. When a new creative variation dramatically outperforms the control on CTR but underperforms on conversion rate, you've learned something important about the gap between what makes someone curious about your product and what makes them buy it.
Cost Per Acquisition (CPA)
CPA — the total spend divided by the number of purchasers driven — is more meaningful than ROAS for evaluating acquisition-focused campaigns, because it expresses the cost of something concrete (a new customer or a converted cart) rather than a ratio that can be gamed by campaign targeting choices.
Track CPA separately for new buyers versus repeat buyers. The cost to acquire a genuinely new customer is worth paying at a higher rate than the cost to remind an existing buyer to repurchase — and conflating the two produces misleading averages.
Tier 3: Operational Metrics
These are the plumbing metrics — the ones that tell you whether your campaigns are technically healthy and delivering as planned. They matter, but they're easy to confuse with more important signals.
- Impression delivery rate — Are your campaigns delivering the impressions you're paying for, or are you losing budget to under-delivery?
- Frequency — Are you reaching the same users repeatedly with the same creative? Too high a frequency wastes budget and irritates; too low may not generate recall.
- Viewability — Particularly for display placements, what percentage of your impressions were actually viewable by a human being? Industry standards set 50% of pixels visible for one second as the minimum, but push for higher.
- Search impression share — For sponsored search campaigns, what percentage of eligible auctions are you actually appearing in? Low impression share often signals under-bidding on high-value terms.
The Attribution Window Problem
One of the most consequential — and least discussed — measurement decisions in retail media is attribution window selection. An attribution window is the period after an ad exposure during which a subsequent purchase is credited to the campaign.
A 7-day attribution window will show you very different numbers than a 14-day or 30-day window, particularly for categories with longer purchase cycles. Grocery replenishment might have a 7-day cycle; a premium spirits purchase might have a 21-day consideration period.
The problem is that different retail media networks use different default attribution windows, which makes cross-network comparison misleading unless you normalize. Some networks allow you to customize the window; take advantage of that capability and set windows that match your category's actual purchase cycle rather than accepting the default.
Building a Measurement Framework That Holds Up
The most practical approach to retail media measurement is a tiered framework that separates strategic reporting from operational optimization. Here's what that looks like in practice:
- Quarterly strategic review: Sales lift, new-to-brand rate, competitive share of search, and total retail sales trend. These go to leadership and inform budget allocation decisions.
- Monthly campaign review: ROAS by campaign type, CPA by audience segment, creative performance comparison, and network-level spend efficiency. These inform tactical adjustments.
- Weekly operational check: Delivery rates, pacing, bid performance, and any anomalies. These are managed by the execution team and escalated only when something is materially off-track.
The discipline of separating these tiers — not letting operational metrics contaminate strategic conversations, and not making weekly decisions based on quarterly trends — is what keeps retail media measurement honest and useful.
One More Thing: The Vanity Metric Trap
Every retail media network will surface metrics that make their platform look good. Impressions delivered. Total media value. Shopper reach. These numbers are real — they describe something — but they rarely connect directly to business outcomes in a way that informs meaningful decisions.
The brands that get the most value from their retail media measurement are the ones that start with the business question — did this investment grow sales, bring in new buyers, or improve our competitive position? — and then work backward to identify which data sources and metrics actually answer that question. Starting with the data the network provides and working forward to a business conclusion is a much less reliable path.
Agency Five Eighty builds retail media measurement frameworks for CPG brands that want to understand what their investment is actually doing — not just what the network dashboard is showing. From attribution architecture to incrementality testing, we help brands measure what matters.