The three-channel framework for e-commerce
Most profitable e-commerce paid media uses three channels in parallel: Google Standard Shopping for high-intent demand capture, Performance Max for cross-surface reach and scale, and Meta Ads for audience building and lower-funnel retargeting.
Each channel serves a different stage. Shopping captures buyers who are actively comparing products. Performance Max extends coverage to YouTube, Display, and Discover. Meta creates and re-engages demand for buyers who are not yet actively searching.
Google Standard Shopping: your highest-intent channel
Standard Shopping campaigns give you control over bids by product group, allowing you to allocate spend proportionally to margin and conversion rate. Structure your product groups so high-margin, high-converting products are in separate ad groups with higher bids.
Product feed quality determines your impressions and click quality. Ensure product titles include the exact search terms buyers use (brand + product type + key specs), images are clean and on a white background, and GTINs are populated for all eligible products.
- Use product type segmentation to bid differently by category and margin
- Add negative keywords at the campaign level to prevent irrelevant search queries
- Check the search terms report weekly — Shopping pulls in broad match-style traffic
- Prioritize feed optimization for your top 20% of products by revenue contribution
Performance Max: when and how to use it
Performance Max replaces Smart Shopping and extends reach to YouTube, Display, Gmail, Discover, and Maps in addition to Search and Shopping. It requires high-quality creative assets and enough conversion volume to optimize. Below 30 conversions per month, PMax often underperforms Standard Shopping because the algorithm lacks the data to allocate budget intelligently.
The right PMax setup separates campaigns by product category so the algorithm optimizes against similar products with similar margins. A blanket "all products" PMax campaign at a single ROAS target mixes high and low-margin products and produces a blended result that looks acceptable but wastes spend on the wrong SKUs.
- Provide 5+ headlines, 5+ descriptions, images in all required sizes, and at least one video for every PMax campaign
- Set a target ROAS that reflects your category margin, not a blended account-level average
- Run PMax alongside Standard Shopping, not instead of it — Standard Shopping protects your best search placements
- Check PMax insights weekly for audience signals and search theme performance
Meta Ads for e-commerce: awareness and retargeting
Meta fills two functions in an e-commerce funnel that Google cannot. First, it builds product awareness through visual formats — carousel, video, collection ads — before buyers start actively searching. Second, it re-engages cart abandoners and product viewers at a much lower cost per click than Google retargeting.
For cold traffic, dynamic product ads (DPA) targeting broad audiences with Advantage+ Shopping Campaigns outperform static image ads in most categories. Meta's algorithm optimizes delivery toward users with purchase intent signals similar to your pixel event history.
- Advantage+ Shopping Campaigns (ASC) for cold traffic: let Meta optimize audience delivery
- Dynamic Product Ads (DPA) for retargeting: show exact products viewed or abandoned
- Video ads for new product launches where search demand does not exist yet
- Lookalike audiences from purchasers for expansion beyond your current customer base
Measuring cross-channel ROAS correctly
Each platform reports ROAS differently. Google Ads counts a purchase as a conversion even if Meta also touched the same order. GA4 uses last-click attribution by default. The true picture requires a cross-channel view — which means reading all three platforms simultaneously.
With MCP Ads, Claude can pull Google Ads ROAS, Meta ROAS, and GA4 e-commerce revenue in one conversation and calculate the blended ROAS across all paid spend. Ask: "Show me total ad spend across Google Ads and Meta for last month alongside GA4 e-commerce revenue and transactions. Calculate blended ROAS and break it down by platform."
Doing this with Claude connected to the account
The three-channel framework generates three separate reporting problems: product-level Shopping data in one place, PMax asset groups in another, and Meta catalog performance in a third. With Google Ads, Meta Ads, and GA4 connected through MCP Ads, Claude can read all of them and reconcile the numbers against the store.
Prompt 1: "Show Shopping performance by product group for the last 30 days: spend, revenue, ROAS, and impression share. Flag any product group with ROAS under 2 and spend over $200." Claude calls google_ads_shopping_performance and google_ads_shopping_product_groups. What comes back is a product-level view sorted by the problem, not by product ID. As an illustrative example, you might see that a clearance category is absorbing 18 percent of spend at 1.3x ROAS while a full-price category is capped by impression share at 6x — the reallocation writes itself.
Prompt 2: "Compare each PMax asset group against the Standard Shopping campaign that covers the same products, and show total ROAS across Google, Meta, and GA4 revenue for the same period." This uses google_ads_performance_max_asset_groups and google_ads_get_asset_group_performance for the PMax side, meta_ads_campaign_performance for Meta, and ga4_ecommerce and ga4_ecommerce_items for the store-side truth. The reconciliation is the part that matters: platform-reported revenue added together will exceed GA4 revenue, and the response tells you by how much so you can judge the channels on the same base.
Prompt 3: "Raise the target ROAS on the full-price Shopping campaign from 400 to 450 percent and increase its daily budget by 20 percent, and check the Meta catalog for disapproved items." google_ads_update_target_roas and google_ads_update_campaign_budget are writes: Claude shows current and proposed values and waits for approval. meta_ads_catalog_diagnostics is a read and comes back immediately with feed errors that would otherwise stay hidden until sales dip.
For a weekly cadence, the ecommerce revenue optimizer in the skills library runs this whole sequence and stops at each write for your approval.
Where the manual way still wins
Merchant Center is not part of the connection, so feed disapprovals, price mismatch warnings, and shipping settings on the Google side still have to be checked there. Claude can see that a product group has no impressions; it cannot always see that the reason is a disapproved feed item. Margin data is also absent from every ad platform, so a ROAS target that looks healthy might be losing money on low-margin lines — only your product cost sheet knows.
Inventory is the other blind spot. Pushing budget into a product group that is about to sell out is a decision that needs the warehouse, not the ad account. Let Claude find the imbalance; you decide whether the stock and the margin support acting on it.
