From 8.84× to 10.07× ROAS: +56% Revenue for a Furniture Hardware Store
A furniture hardware e-commerce store — selling cabinet pulls, hinges, drawer systems, and kitchen accessories — had campaigns running but stuck. We rebuilt the Google Shopping structure from the ground up and grew monthly revenue by 56% without proportionally increasing spend.
+56%
Revenue Growth
10.07×
Final ROAS
+53%
Conversions
+18%
Conv. Rate
The Situation
Good Baseline. Wasted Potential.
When this furniture hardware e-commerce store came to us, the Google Ads account wasn't broken — it was just stuck. An 8.84× ROAS sounds respectable, but for a product category with high repeat-purchase intent and clear search demand, it was well below what the account should have been capable of.
The underlying issue wasn't the budget. It was structure and signal quality. The algorithm was being asked to optimize across wildly different product categories at once, with a feed that wasn't giving Google enough information to match ads to the right queries.
Before · Nov–Dec 2025
8.84×
ROAS
~3.4%
Conversion Rate
~215
Conversions / Month
2.16%
CTR
After · Jan–Feb 2026
10.07×
ROAS
4.04%
Conversion Rate
329
Conversions / Month
2.30%
CTR
Root Causes
Six Problems Holding the Account Back
The audit identified six structural issues — each one suppressing performance independently, and compounding each other:
All product categories in a single campaign — the algorithm couldn't allocate budget intelligently between high- and low-margin product groups, and bid targets were averaged across incompatible niches.
Merchant Center feed with weak item titles — missing brand, material, and size attributes meant Google couldn't match ads to high-intent, specific search queries.
No A/B testing of ad creatives — a single asset version had been running without rotation, limiting CTR improvement opportunities and creative learning.
Bidding strategy was Maximize Clicks — optimizing for volume, not revenue. The algorithm had no conversion-value signal to learn from.
No Performance Max audience signals — without behavioral data fed to PMax, the algorithm took longer to identify high-converting user segments.
Remarketing mixed into general campaigns — warm audiences (users who viewed products or initiated checkout) weren't being served tailored, high-intent messages at scale.
Does your account have any of these issues?
A free 30-minute audit finds every gap — at no cost.
The Approach
Six Optimizations. Run in Parallel.
Rather than a sequential overhaul, we implemented all six improvements simultaneously over the first two weeks — giving the tROAS algorithm enough runway to learn on clean data before the reporting period closed.
1
A/B Testing Ad Creatives
Launched 3–4 headline variants per campaign testing different angles: assortment breadth, pricing, quality, and delivery speed. Reviewed performance every 7 days and kept the top-performing version after 4 weeks. Final CTR reached 2.30% (up from 2.16%).
CTR +6%4-week test cycle3–4 variants/campaign
2
Merchant Center Feed Optimization
Complete feed audit and rewrite: added keywords, brand, material, and size to all item titles; corrected and detailed Google Product Category taxonomy; added color, material, brand, GTIN/MPN attributes; removed duplicate and error listings; connected automated feed rules for ongoing maintenance.
Title rewriteGTIN / MPNAuto feed rulesTaxonomy fix
3
Campaign Restructuring by Category
Split the single general campaign into four focused campaigns: functional furniture hardware, face hardware (pulls, legs), kitchen accessories (dish racks), and remarketing. Each received an individual budget and tROAS target calibrated to that category's margin and demand profile.
Switched all campaigns from Maximize Clicks to Maximize Conversion Value with a target ROAS. After a 2–3 week learning phase, the algorithm began prioritizing products with higher conversion probability. Targets were monitored and adjusted every 5–7 days for the first month.
Per-campaign tROASWeekly monitoring2–3 week learning period
5
Performance Max Audience Signals
Added audience signals to every PMax campaign to accelerate learning: purchasers from GA4 (180-day transaction window), category and product page visitors, and similar audiences built on the purchaser seed list. This reduced the time to exit learning mode and improved placement targeting from day one.
GA4 purchasersCategory visitorsSimilar audiences
6
Isolated Remarketing Campaign
Separated remarketing into its own dedicated PMax campaign with a higher tROAS target than the prospecting campaigns. Warm audiences saw ads featuring the specific products they had viewed, with messaging focused on availability and price — converting browsing intent into purchase intent.
The 60-day results period (January 17 – February 17, 2026) compared against the pre-optimization baseline (November 17 – December 17, 2025) showed improvement across all nine tracked metrics — with revenue growing faster than spend.
Metric
Before
After
Change
Revenue (Conv. Value)
Baseline
+56%
↑ +56%
ROAS
8.84×
10.07×
↑ +14%
Conversions
~215
329
↑ +53%
Conversion Rate
~3.4%
4.04%
↑ +18%
Clicks
~5,929
7,780
↑ +31%
Impressions
~275,000
338,258
↑ +22%
CTR
2.16%
2.30%
↑ +6%
Avg. CPC
Baseline
−3%
↓ −3%
Ad Spend
Baseline
+27%
↑ +27%
Key Efficiency Insight
Revenue grew +56% while ad spend increased only +27% — meaning the account became proportionally more efficient at the same time as scaling. This is the hallmark of structural optimization: not just spending more to earn more, but earning more per dollar spent.
Campaign Breakdown
Which Campaign Moved the Needle Most
Splitting by category revealed dramatically different performance profiles — something invisible when all categories were blended in a single campaign:
Functional Furniture Hardware
10.75×
ROAS
119 conversions2,187 clicks
↑ +91.81% conversions
Face Hardware (Pulls, Legs)
8.64×
ROAS
117 conversions3,608 clicks
↑ +4.27% conversions
Kitchen Accessories (Dish Racks)
11.73×
ROAS
49 conversions1,071 clicks
↑ +184.31% conversions
Remarketing (Warm Audience)
12.49×
ROAS — highest of all campaigns
42 conversions914 clicks
↑ +93.68% conversions
Standout Finding
The kitchen accessories campaign (dish racks) saw +184% conversion growth after being isolated from the general campaign. When its budget competed against all other categories in a single campaign, it was consistently underfunded. Separation gave it a dedicated budget and a tROAS target calibrated to its actual demand — and it responded immediately.
Is Your Google Shopping Account in the Same Position?
This case study is relevant if your account looks like any of the following:
Your ROAS is "okay" but you feel like you're leaving money on the table without knowing exactly why
All your product categories compete for budget inside a single campaign or single PMax campaign
Your Merchant Center feed titles are generic — missing brand, size, material, or specific product attributes
You're running remarketing audiences inside your main prospecting campaign instead of isolating them
You switched to Performance Max without feeding it audience signals from GA4
Your bidding strategy is still Maximize Clicks or Manual CPC
30 min · Free · We'll show you the gaps, no strings attached
Frequently Asked
Questions About This Case Study
In this furniture hardware case, overall ROAS improved from 8.84× to 10.07× (+14%) within 60 days of splitting campaigns by product category and switching to Target ROAS bidding. More notably, individual categories outperformed the blended average: the remarketing campaign hit 12.49× and the kitchen accessories campaign hit 11.73×. The 14% account-level improvement understates the value — it masks the fact that each category now has a dedicated, optimized budget rather than competing against dissimilar products.
Feed optimization determines which search queries trigger your Shopping ads. Weak titles with missing attributes (brand, material, size, GTIN) result in low-intent impressions and poor match quality. In this case, rewriting titles to include specific attributes and correcting the Google Product Category taxonomy contributed to a 31% increase in clicks and a 6% CTR improvement — indicating the ads began appearing for more relevant, higher-intent queries. The cost-per-click simultaneously decreased by 3%, consistent with improved quality score and relevance.
Remarketing audiences (people who viewed products, added to cart, or initiated checkout) have a fundamentally different conversion probability than cold prospecting audiences. When mixed together, the algorithm averages the tROAS target across both — resulting in underbidding on warm audiences and potential overbidding on cold ones. Isolating remarketing into its own campaign lets you set an aggressive tROAS target suited to warm audiences and serve product-specific creative. In this case, the isolated remarketing campaign achieved a 12.49× ROAS with +93.68% more conversions versus the prior period.
Performance Max uses machine learning to find conversions across all Google channels (Search, Shopping, Display, YouTube, Gmail, Maps). Without audience signals, it enters a cold-start learning mode where initial performance is poor and spend is inefficient. By feeding it audience signals — GA4 purchasers from the last 180 days, category page visitors, product page visitors, and similar audiences built on buyers — you give the algorithm a starting point for who your best customers are. This shortens the learning phase and improves placement quality from the first week of each new campaign.
The tROAS bidding strategy requires a 2–3 week learning period after any significant structural change (new campaign, new asset group, new bid strategy). In this case, all six optimizations were deployed simultaneously in the first two weeks of January 2026, and the results period (January 17 – February 17) captured the post-learning performance. Feed optimization and A/B creative improvements began showing measurable CTR impact within the first 4 weeks. For most accounts, visible improvement is evident within 30 days — full stabilization typically takes 45–60 days.