The CPG Retail Media Test Plan: A Practical 90-Day Framework

Most CPG retail media programs fail for a boring reason: they never become a program.

They start as a few “try it and see” campaigns, spread across three retailer portals, with creative that was built for Instagram, and a scorecard that changes every month. The budget gets spent. Nobody learns much. Next quarter, the team “retests” the same stuff.

If you want retail media to become a predictable growth lever, you need a test plan that forces answers.

Below is the 90-day framework we use at Hangar12 to turn retail media from “a line item” into a repeatable system: what to test, how to structure budgets, which KPIs to trust, and how to make decisions without falling in love with vanity ROAS.

First: Decide what you’re trying to prove

Before you touch a retailer console, write down the one sentence your 90-day test must answer. Examples:

• “Can we profitably grow new-to-brand buyers on Walmart without discounting?”
• “Does Roundel off-site prospecting create measurable lift on Target.com?”
• “Which two audience strategies on Kroger drive incremental household penetration?”

If your question is vague (“what works?”), you’ll get a vague result (“some things worked”).

The three buckets of retail media tests and why you need all three

A balanced test plan covers three buckets. Skip one, and you’ll over-invest in the wrong behavior.

1) Demand capture
(search / sponsored products)

This is where you harvest intent. It usually looks best on ROAS, and it’s the easiest place to fool yourself.

Test ideas:

• Brand vs non-brand keyword splits
• Competitor conquesting (with strict guardrails)
• Category keyword coverage and bid strategy
• Product page readiness (images, copy, A+ content) paired with media

2) Demand creation
(display, off-site, video/CTV)

This is where you build mental availability and bring new shoppers into the funnel.

Test ideas:

• Off-site prospecting to category buyers who haven’t purchased your brand
• Video vs display creative ladders
• Sequential messaging (benefit → proof → offer)

3) Conversion acceleration
(offers + retargeting)

This is where you reduce friction when someone is already close to buying.

Test ideas:

• Retargeting windows (3/7/14 days)
• Offer types (coupon vs multi-buy vs “buy now” messaging)
• PDP retargeting vs cart abandon vs category browse retargeting

The 90-day plan: Week-by-week

Days 1–14: Baseline and setup
(don’t skip this)

Your first two weeks aren’t about “winning.” They’re about removing chaos.

Checklist:

• Confirm your hero SKUs (don’t test 18 products at once)
• Clean up retail readiness: images, titles, bullets, reviews, inventory, item mapping
• Align your promotional calendar for the next 60–90 days
• Set a measurement plan (more on that below)
• Build 2–3 creative variants per retailer (not one master file)

Output you should have by Day 14:

• A one-page test matrix (tests, budgets, dates, KPIs, success thresholds)
• Tracking plan and reporting cadence

Days 15–45: Run controlled A/Bs, not “multi-variable soup”

This is where most teams blow it by changing everything at once.

Rules:

• One variable per test: audience OR creative OR placement OR bid strategy
• Pre-commit to budgets and run length
• Don’t pause early unless performance is catastrophic

Suggested budget split (starting point):

• 50% capture
• 30% creation
• 20% acceleration

Challenger brands love capture because it looks “efficient.” The problem is: capture alone caps out fast.

Days 46–90: Scale winners and prove incrementality

By week 7, you should know what you’ll scale. Now you need to know what’s real.

Move from “which campaign had the best ROAS” to:

• Which tactic grew household penetration
• Which tactic grew new-to-brand buyers
• Which tactic improved total category share for your brand

If your retail partner supports it, run at least one incrementality test in this phase.

KPIs: what to track (and what to stop worshiping)

Retail media dashboards make it easy to stare at the wrong numbers. Here’s the scorecard we recommend.

Tier 1 - decision metrics

These should drive budget moves.

• Incremental sales / incremental ROAS (iROAS) when available
• New-to-brand (NTB) % and NTB cost (platform-specific definitions vary)
• Household penetration / buyer growth (especially on loyalty-heavy networks)
• Contribution margin after fees (retail media platforms can hide real costs)

Tier 2 - diagnostic metrics

These help you troubleshoot.

• Click-through rate (creative signal, not a business goal)
• Conversion rate (PDP readiness + offer strength)
• Share of voice on key terms
• Frequency / reach (for upper funnel)

Tier 3 - vanity traps

Useful sometimes, dangerous as a north star.

• Blended ROAS without incrementality context
• Attributed sales when you’re running heavy capture (you’ll over-credit)

Measurement: how to avoid paying for sales you would’ve gotten anyway

If you only use last-click attribution, retail media will look like magic.

The fix is a mix of discipline and experiments.

Options (in order of practicality):

1. Holdout tests (best)
   • A portion of shoppers do not see ads
   • Compare exposed vs unexposed
2. Matched market tests (good for bigger budgets)
   • Test in a few markets, hold others dark
   • Compare sell-through and penetration
3. Pre/post with guardrails (fine when you’re small)
   • Use stable baselines
   • Control for promos, distribution changes, seasonality

You don’t need a PhD. You need a method that makes you honest.

Budgeting: how to split dollars without starving learning

Two mistakes show up constantly:

1. Spreading $20k across six networks
2. Putting 90% into Sponsored Products because it “performs”

If you’re early, pick one primary network (where distribution and velocity are strongest) and one secondary network (where you want to grow). Build a win on one. Then expand.

A clean starter split per network:

• 60% always-on capture (protect your shelf in search)
• 25% prospecting (category buyers, not your brand buyers)
• 15% retargeting + offer amplification

Creative: build for the shelf

Retail media creative has one job: help someone choose you in a comparison moment.

What works:

• Big product shot
• One clear benefit
• One proof point (if you have it)
• Offer callout when applicable
• A clean “where to buy” cue that matches the retailer environment

If you need 10 words to explain the ad, the ad isn’t doing its job.

What to do next

If you want, send me:

• the top 3 SKUs you care about,
• the retailers you’re active in,
• your rough monthly budget,

…and I’ll turn this into a specific test matrix (tests, budgets, dates, KPIs) for the next 90 days.

Or if you’d rather skip the trial-and-error, Hangar12 runs retail media as a managed program: strategy, creative, flighting, and measurement.

Learn more about our retail media services →