5 (More) Questions To Ask Before Working with a Retail Media Network

5 Advanced Questions to Ask Before Committing to an Retail Media Network (Part 2)

Last week we covered the foundational questions every brand should ask before engaging a retail media network. This week, we’re getting into the weeds. These are the questions that reveal how built-out, transparent and data-driven an RMN really is. 

It’s no secret that RMNs are the Wild Wild West. 

These five questions are the ones even sophisticated teams often forget to ask. They expose how clean the data really is, how measurement actually works, and whether the RMN is capable of driving incremental growth beyond your base of loyal buyers. (And if you’ve been following me for any time you know - growth comes from bring in net-new buyers not getting more from loyal buyers). 

If Week 1 teaches you how to spot red flags, week 2 teaches you how to spot the real deal so you can invest wisely. 

1. How clean and deduped is your data?

Why this matters: First-party data is powerful, but only if identity resolution works. Poor deduplication and messy identity graphs lead to overcounting, wasted spend, and flawed attribution. Many RMNs still struggle with reconciling online and offline IDs. The messier this is, the less accurate your ad targeting - meaning less effective media and measurement. 

What to listen for:

• What identifiers feed the data: loyalty cards, credit cards, digital IDs, app logins, etc. (some are more stable and privacy-compliant than others)

• How they merge online and offline shoppers — is there a unified identity graph?

• What percentage of transactions are actually tied to identifiable users (vs. anonymous or unlinked purchases) 

2. What percentage of total sales are trackable via IDs?

Why this matters: If only a portion of a retailer’s sales are tied to identifiable data, then “closed-loop attribution” covers only that fraction. The rest remains opaque or modeled — which limits how accurate measurement and lift calculations can be.

It's no secret that privacy laws have made closed-loop attribution mushy and we don't see that trend changing. Still, since retailers data is often opt-in from consumers it's still cleaner than most. For now. 

What to listen for:

• The % of in-store vs. online transactions linked to IDs

• Whether there’s a sizeable blind-spot for non-card holders or cash-only purchases

• How this gap affects confidence in attribution and incremental lift

3. How frequently are audiences refreshed and optimized?

Why this matters: Retail media moves quickly. If audiences are stale, for example refreshed monthly, you miss spikes in shopper behavior, seasonal shifts, or new buyer flows. For maximum efficiency you want near-real-time updates. More updates will also move buyers more quickly into the right categories (eg. lapsed, frequent purchaser etc)

What to listen for:

• Cadence of refreshes: daily, weekly, monthly?

• Whether they support dynamic segments (e.g. recent purchasers, lapsed buyers, high-frequency buyers) and update in near real-time or with lag

• Their capability to adapt audience definitions during a campaign (e.g. to optimize toward incremental buyers rather than safe bet loyal ones). This is often tricker than it sounds because if your budget may be too large for the scale of audience available. 

4. Do I retain audiences or learnings if I stop spending?

Why this matters: Many RMNs treat audiences as walled and tied to spend. If there’s no continuity post-campaign, you lose valuable data and start from zero next time. That limits long-term brand planning and retargeting flexibility. Because the data is often proprietary and valuable - RMNs are not sharing even with their most valuable clients/brands. 

Keep your expectations in check. We don't expect this to change since they have the upper-hand and no incentive to do so without additional dollars. 

Note: Some RMNs have marketing intelligence available for additional money, and some only allow agencies to access this intelligence vs. brand-direct. The price tag for this data is steep.

What to listen for:

• Whether you retain any segment definitions or aggregated learnings (e.g. insights on buyer behavior, category affinity, new-to-brand rates)

• Whether raw or aggregated audiences/data are exportable or usable outside the RMN’s walled garden

• If not exportable — what’s the RMN’s roadmap (or openness) for interoperability or data portability

5. Can you show me a validated case study with real incremental lift?

Why this matters: A polished pitch deck doesn’t equal real performance. Only RMNs with mature measurement frameworks can show transparent, real-world lift results — from holdout tests, not modeled attribution. Because we don't like anyone grading their own homework (ourselves included) we opt for third-party independent partners wherever/whenever possible. 

What to listen for:

• Presence of pre/post or holdout test methodology (vs. post-hoc correlation) 

• Transparent reporting of lift metrics, ideally with confidence intervals or statistical significance rather than just “lifted X%” headlines

Like this? Subscribe to our weekly newsletter for more great insights on CPG brand-building and media buying. Left Hand Agency helps CPG brands win the shelf by pairing memory-building media with measurement that actually matters.