Google Shopping for D2C Brands: The Setup That Actually Gets You ROAS

Why most Shopping campaigns underperform

The product feed is the "creative" in Shopping. Unlike Meta, where you design ads, in Shopping you submit a product feed and Google decides what to show. A bad feed means Google either can't show your products or shows them against the wrong queries — and because Performance Max has replaced most standalone Shopping campaigns, feed quality now drives search, display, and YouTube placements simultaneously, not just the Shopping tab.

The three most common feed problems: (1) product titles using internal naming conventions instead of search-friendly terms, (2) missing GTINs, which measurably suppresses impression volume on eligible products, and (3) no custom labels for segmentation — meaning you can't bid differently on high-margin versus clearance products even though margin varies wildly across a real catalogue.

Campaign structure: don't hand Performance Max a messy feed

Google now pushes Performance Max as the default for Shopping, and for most catalogues that's the right call — but PMax is only as good as its inputs. Handing it a messy feed and broad targeting produces broad, unpredictable results; feeding it clean, segmented data with clear signals produces something closer to the efficiency Google advertises.

Structure your catalogue into tiers before PMax ever sees it: a Bestsellers asset group (your top 20% of SKUs, highest target ROAS), a Core catalogue group (everything else, standard tROAS), and a Clearance group (slow-moving inventory, aggressive tROAS to move stock rather than protect margin). Feed this structure through custom labels, and give PMax first-party audience signals — your customer list and site visitors — to seed it rather than starting cold.

The search term report is still your best optimisation tool

Shopping campaigns don't have keywords, but they — and PMax campaigns with Shopping inventory — still surface a search term (or search category) insights report. Mining this weekly remains the single highest-ROI activity in Shopping campaign management, PMax or not.

For a furniture brand we work with, mining the search term report revealed ₹40,000/month in spend going to "office chair rental" and "furniture on rent" queries — people not looking to buy at all. Adding these as negatives (via account-level negative keyword lists, since PMax handles exclusions differently from standard Shopping) improved ROAS by 0.6x within two weeks.

The feed fields that actually move impression volume

Shopify's Google & YouTube channel app syncs a feed automatically, and most brands accept whatever it generates. That default sync is a starting point, not a finished feed. The fields worth deliberately controlling:

• Title. The single highest-leverage field. A workable formula for most categories is Brand + Product Type + Key Attribute + Size/Colour — "Carryup Merino Wool Crew Socks, Charcoal, UK 7-11", not "CU-SOCK-CHR-M" or a cute product name that nobody searches. Google matches heavily on title text; an internal SKU name is invisible to the queries you actually want.
• GTIN. Missing GTINs measurably suppress eligibility on comparison placements. If your products genuinely have no GTIN (handmade, custom), set the identifier_exists attribute correctly rather than leaving it blank, which Google reads as an error rather than an intentional absence.
• google_product_category vs. product_type. These are different fields doing different jobs — the first is Google's fixed taxonomy (use the most specific matching node, not a top-level one), the second is your own categorization, which is what you'll segment on with custom labels.
• Image. Google's requirements are stricter than your storefront's. A lifestyle shot with text overlay or a watermark risks disapproval; a clean product-on-white primary image with additional lifestyle images as supplementary is the safe structure.
• Availability and price accuracy. A feed that disagrees with your live product page triggers disapprovals and, worse, erodes the account-level trust that affects impression share across the whole catalogue.

Why PMax reported ROAS overstates reality

Performance Max will happily claim credit for a customer who already knew your brand, searched it by name, and was going to buy regardless — branded search and existing demand sit inside the same campaign as genuine net-new acquisition, and PMax's reporting doesn't separate them by default.

The practical sanity check: compare your PMax-reported revenue against actual Shopify orders for the same window, and watch what happens to total orders (not platform-reported ones) when you pause or materially cut PMax spend for a week. A channel generating genuine incremental demand shows a visible drop in total store revenue; one largely harvesting existing demand shows total revenue barely moving while the reported ROAS number disappears. This is the same attribution gap covered in depth in our guide to real CAC versus platform-reported CAC — worth reading before you scale spend on a number that may be counting demand you already had.

FAQs

Should I still run standard Shopping campaigns or only Performance Max? For most catalogues, Performance Max has effectively replaced standalone Shopping — the priority now is feeding it a clean, segmented product feed rather than avoiding it.

What's the single biggest feed mistake D2C brands make? Using internal naming conventions instead of the search terms real customers use, combined with missing GTINs — both directly suppress how often Google can show your products at all.

Can I still exclude irrelevant search queries in Performance Max? Yes, via account-level negative keyword lists — PMax doesn't support campaign-level negatives the way standard Shopping did, but the exclusion mechanism still exists and mining the search terms/categories insights report weekly is still the highest-ROI habit.

How do I stop Performance Max from wasting spend on low-margin SKUs? Custom labels segmenting your catalogue by margin or velocity, combined with tiered target ROAS by asset group, so the algorithm isn't optimising every SKU against the same target.

How should I structure product titles in the feed? Brand + Product Type + Key Attribute + Size/Colour works for most categories — Google matches heavily on title text, so an internal SKU code or a purely creative product name is invisible to the queries you want.

Is Performance Max's reported ROAS trustworthy? Treat it as an upper bound. It blends branded search and existing demand with genuine new acquisition — pause-test against total Shopify orders to see what's actually incremental.

Our take

Google Shopping — now largely Performance Max — rewards feed discipline more than campaign cleverness. Most of the "channel isn't working" complaints we inherit trace back to a feed that was set up once and never revisited, not an algorithm problem. If your Shopping spend feels unpredictable, talk to us about a feed audit before you touch a single bid.