Summarize this article with:
Most Shopping account audits I do surface the same problem. The ROAS number looks reasonable until you strip brand traffic out. Once you isolate non-brand performance, the account is often barely breaking even. Sometimes it isn’t.
Brand queries convert at high rates, inflate the account’s overall ROAS, and mask the fact that non-brand Shopping is either over-spending or under-performing. That’s the starting point for real google shopping optimization. Not bid adjustments. Not title tweaks. Understanding what your Shopping ROAS actually looks like when you remove the easy conversions that would have happened organically anyway.
I’ve managed Shopping campaigns for 500+ ecommerce brands over 12 years. Here’s the sequence that actually moves ROAS, ordered by impact.
What google shopping optimization really is at the account level
Google shopping optimization is the process of aligning your product feed data, campaign architecture, and bidding signals so that Google’s algorithm matches your products to high-intent queries, serves them at the right impression share, and bids efficiently enough to produce profitable conversions.
Three things control that process. First, feed quality: how well your product titles, GTINs, categories, and attributes match the queries your buyers actually use. Google reads the feed, not your product pages. If titles are generic, no amount of bidding optimization fixes the underlying impression quality problem. Second, campaign structure: which campaign type handles which products, how brand traffic is isolated from non-brand, and how the account segments high-margin from low-margin SKUs. Third, Smart Bidding signals: whether conversion tracking is accurate, whether the bidding algorithm has enough conversion volume to optimize, and whether tROAS targets are set realistically given current blended ROAS.
Most google shopping optimization articles focus on tactics within the third layer. But the first two layers determine whether the third layer can function at all.
For ecommerce brands on Shopify or BigCommerce, the feed is the biggest performance driver in the account. Since 74 to 97% of Performance Max spend in ecommerce goes to Shopping placements, the feed is the primary creative asset. Google uses product titles as keyword targeting signals. Weak titles produce irrelevant impressions. Irrelevant impressions produce poor CTR. Poor CTR signals low relevance to Google’s auction system. It either reduces placement frequency or forces higher CPCs to maintain visibility. The whole performance chain flows from the feed up.
Why most Shopping campaigns underperform from the start
The most common setup mistake is straightforward: a brand launches Shopify, installs a Shopping feed plugin, connects Merchant Center, and turns on Performance Max. Everything appears to work. Products are showing up. Conversions are coming in.
But the ROAS is inflated by brand. PMax is spending heavily on queries like “[Brand Name] + product type” because those convert at high rates and the algorithm optimizes toward them. Meanwhile, the non-brand queries (buyers who don’t yet know the brand) are either not triggering at all or triggering at poor impression quality. Both happen because feed attributes are generic.
The second common mistake: no brand exclusion list in Performance Max. Without a brand exclusion list set at the campaign level (under Settings > Brand exclusions in PMax), the campaign bids aggressively on every branded query. In accounts where I’ve added brand exclusions and moved branded traffic to a separate low-bid Search campaign, non-brand ROAS typically drops 20 to 40% from the inflated baseline. That’s not an optimization failure, actually. That’s the true account performance finally becoming visible.
The third mistake is premature tROAS. Setting Target ROAS on a campaign with fewer than 30 monthly conversions is one of the fastest ways to strangle Shopping performance. tROAS restricts spend to preserve its target. When conversion volume is low, the algorithm doesn’t have enough signal to find efficient conversions, so it restricts spend further, which reduces conversions further. Breaking that cycle requires going back to Maximize Conversion Value, generating conversion volume, and transitioning to tROAS only after the data supports it.
The 8-step google shopping optimization sequence
These are ordered by the impact they typically produce on ROAS. Work through them top to bottom. Each one builds on the previous.
1. Conversion tracking validation before anything else. If conversion tracking is misfiring, double-counting, or measuring the wrong event, every Smart Bidding decision downstream is optimizing against bad data. The first check: open the Google Ads conversion actions list and verify that the primary conversion action is a real purchase event with a real conversion value. Not a proxy like “page view” or “add to cart.” Enable enhanced conversions (in Google Ads Settings > Conversions) to improve model accuracy by 15 to 25% at no additional cost. Fixing tracking is unglamorous work that doesn’t move a single dashboard metric directly. But it changes everything downstream.
2. Brand traffic isolation in Performance Max. Add a brand exclusion list to every PMax campaign (Settings > Brand exclusions > Search terms > add your brand name and major misspellings). Create a separate branded Search campaign with a $2 to $5 tCPA target to capture that traffic at a fraction of PMax’s bids. Compare 30-day non-brand Shopping ROAS before and after this change. In most accounts, non-brand ROAS is 30 to 60% lower than the blended number. That gap is your real optimization baseline.
3. Feed title restructure for top-revenue SKUs. Start with the top 20% of SKUs by revenue. Rebuild titles using the structure: Brand + Product Type + Key Attribute + Attribute 2 + Attribute 3. For a garage floor coating: “ArmorGarage Epoxy Garage Floor Kit 2-Car Gray 250 sq ft” instead of “Floor Kit.” The first title matches model-specific, high-intent queries. The second matches almost nothing useful. In accounts where I’ve rebuilt feed titles on top-revenue SKUs, impression share on non-brand queries typically improves 20 to 40% within 30 days. Use a supplemental feed in Merchant Center to push improved titles without touching the primary catalog.
4. GTIN coverage and Merchant Center diagnostics. Pull the Merchant Center Diagnostics tab. How many products are disapproved? What’s the GTIN coverage? Products without GTINs miss premium Shopping placements. Disapproved products have zero impression eligibility. Brands with correct GTINs see roughly 20% more clicks on average. For any branded, manufactured product, GTINs are findable via the GS1 US database or the product’s own packaging. Adding them is 2 to 6 hours of work that unlocks placements immediately. Hustle Marketers’ GTIN guide for Google Shopping covers the lookup process and common errors.
5. Campaign architecture: Standard Shopping for hero products, PMax for the catalog. First, use Standard Shopping for top-margin, top-revenue SKUs. Standard Shopping for top-margin, top-revenue SKUs gives full search term visibility, manual negative keyword control, and product-group-level bid management, everything PMax doesn’t. Then, use Performance Max for the broader catalog: the remaining 80% of SKUs that benefit from scale rather than control. Newer products without performance history and lower-margin items also fit PMax well. A separate branded Search campaign captures brand queries at low CPCs.
6. Smart Bidding calibration. Launch new Shopping campaigns with Maximize Conversion Value and no tROAS target. Wait until 30 to 50 monthly conversions accumulate. Then transition to Target ROAS at 10 to 15% below the observed blended ROAS. Tighten in increments of 10 to 15% every 14 days, not daily. Never change tROAS by more than 15% at a time. Use seasonality adjustments (Google Ads Tools menu > Bid adjustments > Seasonality adjustments) for promotional periods rather than changing tROAS mid-flight, which triggers a learning phase reset.
7. Custom labels for margin-aware bidding. Add custom_label_0 through custom_label_4 to your product feed. Assign labels for margin tier (high/medium/low), performance history (proven/testing/new), and inventory level (clearance/in-stock/pre-order). Create separate campaigns or asset groups by margin tier. Bid higher on high-margin products and lower on clearance items. Without this segmentation, a campaign bidding a uniform tROAS treats a $400-margin item identically to a $12-margin item. For accounts with catalog breadth above 100 SKUs, custom label segmentation typically improves blended ROAS by 8 to 20% within 60 days.
8. Search term analysis and negative keyword expansion. For Standard Shopping campaigns, review the search terms report weekly. Add negatives aggressively: competitor brand names, irrelevant product types, low-intent query patterns. For PMax, the search terms insight report (under Insights) shows the categories driving traffic. You can’t add direct negatives to PMax at the keyword level. But you can add account-level negative keywords (Google Ads Settings > Negative keywords > Add at account level) that block off-target queries across all campaign types. In accounts where I run weekly negative keyword reviews, wasted spend on off-intent queries typically drops 15 to 25% within 60 days.
Performance Max vs Standard Shopping: which to use when
Performance Max replaced Smart Shopping as Google’s default Shopping campaign type. It’s not optional in the same way. But it’s also not the right choice for every product and every account.
Use Standard Shopping for hero products: the top 20% of SKUs by revenue and margin, branded products with GTINs, and products in competitive categories where impression share visibility matters. Standard Shopping gives you search term reports, product-group-level bid control, and full negative keyword management. You can see exactly which queries are triggering which products. That visibility is the primary advantage over PMax.
Use Performance Max for the broader catalog where scale and multi-channel reach matter more than granular control. But don’t run a single PMax campaign across everything: it merges products with vastly different margins, conversion rates, and competition levels into one bidding pool the algorithm can’t segment efficiently.
The bidding sequence that produces real ROAS improvement
The most common bidding mistake on Shopping campaigns is actually trying to improve ROAS by raising the tROAS target. Higher tROAS targets restrict spend, reduce impression share on borderline queries, and concentrate traffic on the safest conversions. That’s not ROAS improvement. That’s ROAS concentration.
Instead, real ROAS improvement comes from improving what happens before the bid: feed quality, brand traffic isolation, and campaign structure. When those are correct, the same tROAS target delivers more revenue because the impressions are higher quality, the queries are more relevant, and the conversion rate on clicks is higher.
If ROAS is too low and the account has healthy conversion volume, the right move is feed work and negative keywords, not tROAS increases. If ROAS is too low and conversion volume is thin, the right move is to drop the tROAS target 10 to 15% temporarily to restore impression share and rebuild conversion data.
What these optimizations produced for real accounts
ArmorGarage, BigCommerce, garage floor coatings. Inherited account had a single PMax campaign covering the full catalog with no brand exclusions and 60% of SKUs missing GTINs. Apparent ROAS looked acceptable. After isolating brand traffic with a separate branded Search campaign and adding brand exclusions to PMax, non-brand ROAS became visible. Feed title rebuild on top SKUs followed, then GTIN addition, then PMax restructure with product-line asset groups. Within 90 days: 1,500%+ ROAS on the rebuilt account. The ArmorGarage case study documents the sequence in detail.
P-REX Hobby, Shopify, hobby parts for Bin Chen. Good tracking, reasonable architecture. Feed titles were generic product names with no compatibility descriptors. After rebuilding titles to front-load brand, part type, and compatibility model number (“Traxxas Rustler 4WD Compatible Drive Shaft 3.2mm Pin” instead of “Drive Shaft”), impression share on model-specific queries improved measurably within two weeks. ROAS hit 9x over 90 days. The P-REX Hobby case study covers the feed restructure process.
ThePetsClub UAE, Shopify Plus, pet food and supplies. PMax running on the full catalog with no brand exclusions, no Customer Match, one undifferentiated asset group. Brand traffic was masking poor non-brand performance. After brand isolation, Customer Match loading (18,000-person CRM list), asset group restructuring by product category, and category depth update in Merchant Center, non-brand ROAS improved significantly. Combined ROAS across the account reached 14x over 90 days.
What I’d audit first in any Shopping account
First, start with the brand traffic question. Pull the PMax search terms insight report. What percentage of spend is going to queries containing your brand name? If it’s above 20% and there’s no separate branded Search campaign, you’re paying PMax rates for traffic that would have converted at a fraction of the cost.
Then, check Merchant Center disapprovals. How many products have zero impression eligibility right now? Five minutes in the Diagnostics tab answers this. In most inherited accounts, between 5 and 15% of the catalog is disapproved for fixable reasons.
Third, check Smart Bidding thresholds. How many monthly conversions does each Shopping campaign have? Below 30, tROAS is starving the algorithm. Below 15, even Maximize Conversion Value may not have enough signal to optimize effectively.
Fourth, pull 20 product titles from top-revenue SKUs at random. Do they match the actual queries in the search terms report? If there’s a large gap between the words in titles and the words in search terms, feed restructuring will move ROAS faster than any bidding change.
What google shopping optimization costs in time and money
Feed title rebuild for top 200 revenue SKUs: 4 to 8 hours, one-time. This is the single highest-impact task, so it’s worth doing carefully. Most of that time is catalog research and supplemental feed setup.
GTIN research and entry for branded products in a catalog of 200 to 500 SKUs: 2 to 6 hours. Finding GTINs via GS1 database and entering them via supplemental feed.
Brand traffic isolation setup (exclusion list + branded Search campaign): 2 to 3 hours. This typically produces immediate ROAS visibility improvement.
Custom label setup across the catalog: 2 to 4 hours, assuming margin and inventory data exists in the ecommerce platform.
Total first-pass google shopping optimization investment for an active account: $600 to $2,000 in specialist time, one-time. The return materializes in the first 30 to 60 days. At $30,000 monthly ad spend, a 20% ROAS improvement adds $6,000 in monthly revenue. Our ecommerce PPC management services cover how this ongoing work is structured.
Why work with Ishant Sharma on google shopping optimization
Twelve years. 500+ ecommerce brands. $780M+ in trackable client revenue. Google Partner and Meta Business Partner. Upwork Top Rated Plus with a 99% Job Success Score and a 5.0/5.0 rating. Clutch Award Winner 2024.
Every new Shopping engagement starts with the same diagnostic: brand traffic share, Merchant Center disapproval rate, conversion tracking accuracy, feed title quality, Smart Bidding thresholds. That audit typically surfaces 3 to 7 structural issues that explain why the account is underperforming, before I change a single bid or campaign setting. ArmorGarage, P-REX Hobby, and ThePetsClub all produced their documented ROAS results through structural changes applied in that diagnostic order.
Hustle Marketers’ ecommerce PPC agency page covers how those engagements are scoped and what the first 90 days typically look like.
What to take from this
Google shopping optimization is a sequencing problem. The feed determines which queries trigger your products. Brand isolation determines whether your ROAS metrics mean anything. Campaign architecture determines whether Smart Bidding has the right product segmentation to optimize profitably. Smart Bidding calibration determines whether the algorithm can find efficient conversions at your target.
Fix them in that order. Feed and brand isolation first, then architecture, then bidding. Every account I’ve taken from “acceptable ROAS” to 9x, 14x, or 1,500%+ followed the same sequence. Not because it’s a magic formula. Because each layer enables the next one, and skipping layers produces fragile results that fall apart when competitive dynamics or seasonal patterns shift.
About Ishant Sharma
Ishant Sharma is a Google Ads specialist and Founder of Hustle Marketers, a Google Partner and Meta Business Partner agency working with e-commerce and lead-gen brands across the US, UK, UAE, and Australia. 12+ years in performance marketing. Trackable client revenue across his work has crossed $780 million. Upwork Top Rated Plus with a 99% Job Success Score and a 5.0/5.0 rating. Clutch Award Winner 2024. Based in Chandigarh, India.