This retail media case study for a CPG brand pulls back the curtain on how a mid-size personal care company restructured its Amazon Ads and Walmart Connect campaigns over six months — cutting ACOS by 38%, eliminating five-figure monthly wasted spend, and growing attributed revenue 2.4x without increasing total media budget. The lessons apply directly to any challenger CPG brand competing against private label and category giants on the digital shelf.
The Brand, the Problem, and What Was at Stake in This Retail Media Case Study for a CPG Brand
The brand in question is a mid-size personal care company selling across three primary categories: hair care, skin care, and body care. Annual retail media spend at the start of the engagement was roughly $1.4 million split across Amazon Sponsored Products, Sponsored Brands, Sponsored Display, and Walmart Connect Sponsored Products. The brand had grown quickly from a DTC-first startup into a multi-retailer business over three years, and its advertising structure reflected that chaotic growth — campaigns were layered on top of campaigns, targeting logic was duplicated, and no coherent incrementality measurement framework existed.
The core problem was not underinvestment. The brand was spending aggressively. The problem was that media dollars were working against each other. Auto campaigns were cannibalizing exact match keywords. Branded keyword campaigns were eating budget that should have been protecting margin on already-won customers. Sponsored Display retargeting was firing on purchasers within 7 days of a completed order — paying to reach people who had already converted.
At stake was category viability. The brand's category managers at two major retail partners had flagged declining velocity on key SKUs. When retail media spend inflates apparent sales rank without generating true new buyers, velocity metrics deteriorate once spend is pulled back. The brand faced a scenario where its shelf position — both digital and physical — was at real risk unless it could demonstrate organic demand.
"We were essentially renting our own sales rank. The moment we paused a campaign, velocity dropped. That's a signal you've built a dependency, not a sustainable business."
The brand's internal team had strong creative capabilities but limited paid search expertise specific to retail media networks. A specialist retail media consultancy was brought in for a 90-day diagnostic followed by a full restructure. The total incremental investment in the engagement was approximately $48,000 in fees — against a potential $530,000+ annual wasted spend identified in the diagnostic phase.
Strategy and Approach: What Was Decided (and What Was Deliberately Left Out)
The strategic brief came down to three priorities: reduce wasted spend immediately, restructure campaigns for clean measurement, and prove incrementality before scaling. The team reviewed the brand's existing retail media network strategy and found it had none — what existed was a collection of tactical decisions made in isolation without a connecting logic.
Three strategic decisions defined the rebuild:
1. Full campaign audit before any new spend. No new campaigns were launched during the first 30 days. Every live campaign was assessed against a set of criteria: does this campaign target a unique audience or keyword set? Is there a clear conversion hypothesis? Is there a way to measure whether this spend is generating new-to-brand buyers or recirculating existing demand? Campaigns that failed two or more criteria were paused, not optimized.
2. Separation of branded and non-branded keyword spend. Branded terms were isolated into dedicated campaigns with aggressive negative keyword lists applied to all non-branded campaigns. This sounds obvious, but the brand had no clean separation in place. Branded spend was accounting for an estimated 34% of total keyword budget while delivering conversion rates that masked poor performance on conquest terms.
3. No Sponsored Display without an incrementality guardrail. Display retargeting was suspended entirely until a purchase exclusion window of 30 days could be configured. No pixel budget was allocated to customers who had already converted in the trailing 30 days. This alone removed a category of spend that was generating attributed revenue with near-zero incremental value.
What the strategy explicitly did not include: no new ad formats were added, no new retail media networks were onboarded, and no additional budget was requested. The mandate was to make the existing $1.4M work harder, not to spend more.
Implementation: Steps, Timeline, and Tools
The restructure ran across three phases over 26 weeks. The team used Amazon's native campaign manager, Walmart Connect's self-serve platform, a third-party bid management tool for automation rules, and a custom attribution dashboard built on top of each platform's API exports combined with retail sales data from the brand's category management portal.
| Phase | Timeframe | Primary Actions | Key Outputs |
|---|---|---|---|
| Phase 1: Diagnose and Pause | Weeks 1–4 | Full campaign audit, wasted spend identification, pause non-performing campaigns | 37 campaigns paused, $44K/month in spend redirected |
| Phase 2: Rebuild and Segment | Weeks 5–14 | Campaign restructure by match type and audience, negative keyword build-out, purchase exclusion windows set | Campaign count reduced from 94 to 31, keyword list expanded from 1,200 to 3,800 terms |
| Phase 3: Measure and Scale | Weeks 15–26 | Incrementality testing, bid optimization, new-to-brand reporting integration, controlled budget reallocation | ACOS down 38%, new-to-brand attributed orders up 61%, revenue 2.4x |
A critical implementation detail: the team ran a two-week holdout test in weeks 15 and 16, pausing Sponsored Display entirely across a matched set of ASINs to establish a baseline for true incrementality. The holdout confirmed that approximately 28% of previously attributed display revenue was non-incremental — meaning those sales would have occurred regardless of ad exposure. That data point directly informed how Display budget was reallocated in weeks 17 through 26.
Bid management automation was applied only after manual review confirmed the campaign logic was structurally sound. Automating a broken campaign structure accelerates waste — the team was explicit about this sequencing with the client.
Results: Before and After Metrics
Measured against the six-month pre-restructure baseline, results at the 26-week mark showed across-the-board improvement. Precise numbers are presented below. Total retail media budget remained fixed at $1.4M annualized throughout the engagement — no incremental budget was added.
| Metric | Before (Baseline) | After (Week 26) | Change |
|---|---|---|---|
| Blended ACOS (Amazon) | 31.4% | 19.5% | −38% |
| Attributed Revenue (Amazon + Walmart) | $2.1M (6-month period) | $5.04M (6-month period) | +2.4x |
| New-to-Brand Orders (Amazon) | 12,400 orders | 19,964 orders | +61% |
| Estimated Wasted Spend (Monthly) | ~$44,200/month | ~$6,800/month | −85% |
| Active Campaign Count | 94 campaigns | 31 campaigns | −67% |
| Walmart Connect ROAS | 3.1x | 5.7x | +84% |
The 2.4x revenue growth figure deserves context: it reflects attributed revenue in the platforms' own reporting. The brand also ran the holdout methodology described above to adjust for non-incrementality, arriving at an incrementality-adjusted attributed revenue multiple of approximately 1.9x — still a material gain, and one the brand could defend to its retail partners as representing genuine demand generation. Understanding the nuances of retail media network ROI measurement was what allowed the team to present both figures with credibility rather than cherry-picking the more flattering number.
Organic ranking on 14 primary target keywords improved by an average of 6.3 positions on Amazon over the same period — consistent with the hypothesis that healthier, more incremental sales velocity produces compounding organic benefits.
Key Learnings: What Worked, What Failed, and What Surprised Us
What worked: The decision to pause before rebuilding was the single highest-leverage action. It forced the team and the client to confront what the campaigns were actually doing rather than rationalizing existing structure. Negative keyword discipline — maintaining over 2,100 negative keywords across the rebuilt account — prevented the campaign structure from re-corrupting itself over time. Separating branded from non-branded spend gave the team clean data within 60 days, which accelerated every subsequent decision.
What failed: The first attempt at Walmart Connect restructuring lagged by about six weeks because the brand's category manager portal access for Walmart's attribution data was delayed through an administrative issue. That gap meant the team was optimizing Walmart campaigns on platform-reported metrics alone for longer than planned. It also revealed how dependent good retail media decisions are on having clean, timely data pipelines — something many brands underestimate when they plan a restructure.
What surprised the team: The speed at which organic rank responded to cleaner, more incremental sales velocity. The hypothesis was that it would take four to six months for algorithmic ranking signals to reflect the restructure. Meaningful movement was visible at week 11 — faster than expected. Industry practitioners have increasingly noted this connection between ad-generated velocity quality and organic rank, but seeing it play out this quickly on 14 keywords simultaneously was unexpected.
"The assumption going in was that pulling back Display spend would hurt velocity. The opposite happened — because the velocity that remained was real. Algorithms appear to weight signal quality, not just signal volume."
A secondary surprise: pausing 37 campaigns in Phase 1 generated almost no complaints from internal stakeholders once the team presented the wasted spend analysis. The data did the persuasion work. Many restructure engagements stall at the political stage — getting buy-in to pause existing campaigns that someone built and defended. Having a rigorous, documented audit made that conversation straightforward.
How to Replicate This: An Actionable Checklist
The following checklist distills the methodology into steps any CPG brand with active retail media campaigns can apply. Adjust timelines based on team bandwidth and account complexity.
Audit phase (Weeks 1–4):
- Pull 90 days of campaign-level performance data from every active retail media network.
- Tag each campaign: branded, non-branded conquest, category defense, retargeting, or auto.
- Identify keyword duplication across campaigns — any keyword appearing in more than one campaign targeting the same match type is a structural problem.
- Calculate the share of total spend going to branded terms. If it exceeds 25% of non-display keyword budget, it needs to be isolated and scrutinized.
- Identify retargeting audiences and check whether purchase exclusion windows are active. If they are not, pause retargeting until they are configured.
Rebuild phase (Weeks 5–14):
- Rebuild campaigns around a strict hierarchy: one campaign per targeting type per category. No exceptions until the structure is stable.
- Build negative keyword lists before launching any new campaigns. Start with a minimum of 300 negatives per campaign cluster.
- Set branded campaigns to exact match only, with all non-branded campaigns carrying brand terms as negatives.
- Configure new-to-brand reporting on Amazon if not already active — this is the primary incrementality signal available natively in the platform.
- Document every campaign's targeting hypothesis in writing. If you cannot state in one sentence what new buyer this campaign is designed to reach, do not launch it.
Measure and scale phase (Weeks 15–26):
- Run a holdout test on at least one ad format before scaling spend back into it. Two weeks is a minimum viable holdout window for Sponsored Display on most mid-size accounts.
- Reallocate budget from paused campaigns only into campaigns that have demonstrated clean conversion logic and measurable new-to-brand contribution.
- Review organic rank weekly for your top 20 target keywords — treat unexplained rank improvement as a positive signal that sales quality is improving.
- Build a single reporting dashboard that shows platform-attributed revenue alongside incrementality-adjusted revenue. Present both numbers to stakeholders every reporting cycle.
- Set a 90-day review to assess whether the campaign count has crept back up. Campaign sprawl is a recurring problem — not a solved one.
Frequently Asked Questions
How long does a retail media campaign restructure typically take for a CPG brand?
Most mid-size CPG brands should plan for a 20–26 week full restructure cycle if starting from an audited baseline. The first four weeks are almost entirely diagnostic — pausing campaigns and cleaning data — with limited visible performance improvement during that window. Meaningful ACOS and ROAS improvements typically become visible between weeks 8 and 12 as the rebuilt campaign structure stabilizes and platforms re-learn bidding signals on the cleaner account.
What is a realistic ACOS reduction target for a CPG brand restructuring retail media campaigns?
Industry practitioners commonly report ACOS reductions of 20–40% in restructure engagements where the original account had significant keyword duplication, no branded/non-branded separation, and active retargeting without purchase exclusions — all structural problems rather than bid-level inefficiencies. The 38% reduction in this case study sits at the high end of that range, partly because the pre-restructure account had all three of those problems simultaneously. Brands with cleaner existing structures should expect more modest gains in the 15–25% range.
How do you measure incrementality in retail media without a third-party measurement tool?
The most accessible native method is using Amazon's new-to-brand reporting, which tracks orders from customers who have not purchased from the brand in the trailing 12 months. Paired with a manual holdout test — pausing a specific ad format on a matched set of ASINs for two weeks and comparing velocity against a control group — brands can establish a practical incrementality baseline without a third-party tool. It is a lower-fidelity measurement than a statistically rigorous geo-holdout, but it is sufficient to make directional budget decisions and identify clearly non-incremental spend categories.
Should CPG brands run Sponsored Display on Amazon if they cannot configure purchase exclusion windows?
No — not for retargeting audiences. Without a purchase exclusion window of at least 14 days (30 days is preferable), retargeting campaigns will systematically serve ads to customers who have already converted, generating attributed revenue that is almost entirely non-incremental. This inflates reported ROAS, creates false confidence in campaign performance, and wastes budget that could be directed toward conquest targeting or new-to-brand keyword expansion. Sponsored Display used for contextual targeting against relevant product detail pages carries a different risk profile and can be run without purchase exclusions.
How should a CPG brand split retail media budget between Amazon Ads and Walmart Connect?
Budget allocation should follow where the brand's category is demonstrably growing — which for most CPG categories through 2026 means Amazon still commands the larger share, typically 60–75% of retail media investment for brands at the $1–5M annual spend level. Walmart Connect warrants meaningful investment when the brand has confirmed retail distribution in Walmart stores or strong dot-com velocity, as the platform's closed-loop attribution is most reliable when there is genuine retail presence backing it. Many practitioners recommend starting Walmart Connect at 20–25% of total retail media budget and adjusting based on demonstrated ROAS rather than allocating based on corporate interest in the platform.