GA4's ecommerce reports are the most valuable reports in the platform and the most painful to use. The Monetization section gives you a flat table of products sorted by revenue, a purchase journey that's too aggregated to act on, and a set of exploration templates that take twenty minutes to configure and still don't answer the question you actually had. Most ecommerce teams end up exporting CSVs and building pivot tables in Sheets — which works until you need to do it again next Monday.
With GA4 connected to an AI assistant through MCP, the workflow inverts. Instead of building a report to find an answer, you ask the question and the report builds itself. Product performance, category-level analysis, AOV diagnosis, revenue attribution by channel — each one is a prompt, not an exploration. If you haven't connected GA4 yet, the no-code setup guide takes five minutes.
What GA4 ecommerce data MCP actually exposes
Before diving into prompts, it helps to know what's available. GA4 tracks ecommerce through a chain of item-scoped events: view_item, add_to_cart, begin_checkout, add_shipping_info, add_payment_info, purchase. Each event carries item-level parameters — item_name, item_id, item_category, item_brand, price, quantity, and optionally up to five item_category levels for hierarchy.
When you connect GA4 to Claude or ChatGPT through a GA4 MCP server, the AI can pull all of this. Item-scoped metrics — item views, add-to-carts, purchases, item revenue — are available alongside session-scoped metrics like conversion rate and sessions. The combination is what makes the analysis useful: you can ask "which products get viewed but never added to cart" and "which traffic sources drive the most product revenue" in the same chat.
Two things MCP does not expose that you might expect. First, individual order details — GA4 doesn't store order-line-level data in a way the Data API surfaces cleanly, so you won't get "show me order #4521." For that, use your Shopify or backend system. Second, margin data. GA4 tracks revenue but not cost of goods, so any margin or profitability analysis needs a data source outside GA4.
With those limits noted, what follows is the ecommerce analysis that GA4 + AI handles well.
Product performance: the four prompts that matter
Most ecommerce teams look at product performance once a quarter in a slide deck. With AI, the same analysis takes four prompts and five minutes — and you'll actually do it weekly.
Which products are selling and which are stalling
"Rank my top 30 products by item revenue for the last 30 days. For each, show me item views, add-to-cart count, purchases, item revenue, and view-to-purchase rate. Sort by revenue descending."
This is the baseline product performance table. The column that matters most isn't revenue — it's view-to-purchase rate. A product with 10,000 views and a 0.3% purchase rate is leaking money. A product with 500 views and a 4% purchase rate is an underexposed winner. Revenue alone hides both signals.
"Which products had the biggest drop in purchases week over week? Limit to products with at least 20 purchases per week to filter noise."
The volume floor is critical. Without it, the AI returns a product that went from 2 purchases to 1 and calls it a 50% drop. With the floor, you get products where the shift is real and worth investigating.
View-to-cart and cart-to-purchase rates
"For my top 20 products by item views, compare view-to-add-to-cart rate and add-to-cart-to-purchase rate. Which products have a healthy view rate but drop off before cart? Which have a healthy cart rate but drop off before purchase?"
This separates two different problems. A high-view, low-cart product has a product page issue — the page isn't convincing enough, the price isn't right, or the imagery isn't working. A high-cart, low-purchase product has a checkout issue — shipping cost surprise, payment friction, or a trust gap at the final step. The fix is different for each.
"Compare the view-to-purchase rate of my top 10 products this month versus last month. Which products improved and which degraded? For any product that degraded by more than 20%, flag it."
This is the early warning system. A product whose conversion rate degrades 20% in a month is either losing organic visibility, facing a new competitor, or has a page that broke. Catching it in month one is the difference between a quick fix and a quarter of lost revenue.
Category and collection analysis
Product-level analysis is useful for tactical fixes. Category-level analysis is where you make strategic decisions — which categories to invest in, which to promote, which to sunset.
"Group my products by item_category and show me total revenue, total purchases, average order value, and conversion rate for each category over the last 30 days. Sort by revenue. Include the number of unique products in each category."
The "unique products" column is what makes this useful. A category with high revenue and three products is a concentration risk. A category with low revenue and fifty products might need curation, not promotion.
"Compare category performance for the last 30 days versus the previous 30 days. Which categories are growing and which are shrinking? Show the percentage change in revenue and conversion rate."
Category trends are slow-moving and easy to miss. A category that's been declining 8% month over month for three months is down 22% — but nobody noticed because it was never dramatic enough to trigger an alarm. This prompt catches the drift.
"For my top 5 categories by revenue, break down traffic source. Which categories are most dependent on paid traffic? Which are strongest on organic?"
This shapes budget decisions. A category that's 80% paid-dependent is expensive to grow. A category that's mostly organic has more room for profitable scaling through paid — or more risk if organic drops.
"Which product categories have the highest add-to-cart rate but the lowest purchase completion rate? Something is breaking between intent and purchase in those categories."
This is the diagnostic question most ecommerce teams never ask. A category where people add to cart enthusiastically but don't buy is almost always a shipping, pricing, or payment issue specific to that category's price point or weight class.
Revenue attribution: which channels drive revenue, not just traffic
The standard GA4 channel report shows sessions by source. That's the wrong metric for ecommerce decisions. What matters is which channels drive revenue — and specifically, revenue per session.
"Show me sessions, purchases, revenue, conversion rate, and revenue per session by source/medium for the last 30 days. Sort by revenue per session descending."
Revenue per session is the metric that settles budget arguments. A channel with a 0.8% conversion rate and a $120 AOV produces $0.96 per session. A channel with a 2% conversion rate and a $40 AOV produces $0.80 per session. The first channel looks worse by conversion rate and is actually more valuable per visit.
"Compare my paid channels — google/cpc, meta/cpc, tiktok/cpc — on revenue per session and return on ad spend for the last 30 days. Which channel is most efficient?"
If you pass cost data through UTM parameters or have it available via another connector, this prompt gives you a clean ROAS comparison. If cost data isn't in GA4, revenue per session is still the best proxy for efficiency that GA4 alone can provide.
"For each of my top 5 traffic sources, which product categories do they drive the most revenue in? Are some channels better at selling specific categories?"
This is the channel-category matrix that most media buyers build manually in a spreadsheet once a quarter. With AI, it takes ten seconds. The insight — "Meta ads drive 60% of our apparel revenue but almost none of our electronics revenue" — is the kind of finding that changes creative strategy.
"Has my AI Assistant channel grown in the last 30 days? What's its revenue per session compared to organic search? Which products is it driving?"
The AI Assistant channel in GA4 — sessions from ChatGPT, Claude, Perplexity, and similar referrers — is growing fast for most ecommerce sites in 2026. It's worth tracking separately because the purchase behaviour is different: higher intent, higher AOV, fewer page views before purchase. If it's already 3–5% of your traffic, it deserves its own line in your channel review.
AOV diagnosis: why your average order value shifts
AOV is one of the three levers in ecommerce (traffic, conversion rate, AOV) and the one that gets the least structured analysis. Most teams notice AOV when it shows up in a quarterly review. By then, the shift is three months old.
"Show me my average order value daily for the last 60 days. Highlight any week where AOV shifted by more than 10%. Is the current AOV higher or lower than the 60-day average?"
The daily view is what catches the shape. A gradual AOV decline is a product mix problem — you're selling more of the cheap stuff. A cliff is a discount event, a pricing change, or a broken upsell. A spike followed by a return to baseline is a promotion. Each has a different follow-up.
"Break down AOV by traffic source for the last 30 days. Which sources bring the highest-value carts?"
Organic search almost always has a higher AOV than paid social, because organic visitors arrive with higher intent and less price sensitivity. If that pattern is reversed in your data, something unusual is happening — maybe your paid campaigns are targeting high-value buyers effectively, or maybe your organic traffic mix shifted.
"Compare AOV for new visitors versus returning visitors over the last 30 days. Is the gap growing?"
Returning visitors typically have a 15–30% higher AOV than new visitors. If that gap is narrowing, your retention program might be weakening or your loyalty pricing might be too aggressive. If it's widening, your acquisition channels might be bringing in lower-intent traffic.
"Which products appear most often in high-AOV orders versus low-AOV orders? Are there natural bundles forming?"
This is the question that feeds your merchandising strategy. Products that co-occur in high-value carts are candidates for explicit bundles, cross-sell prompts, or "frequently bought together" placements. GA4 doesn't surface this natively — the AI has to cross-reference item-level data with transaction-level AOV.
The weekly ecommerce review: a 15-minute workflow
The prompts above are building blocks. Here's how to use them as a weekly routine.
Monday, 15 minutes. Open a chat with GA4 connected. Run five prompts in sequence:
-
"Compare my total ecommerce revenue, transactions, conversion rate, and AOV for the last 7 days versus the previous 7 days. Flag anything that moved by more than 10%."
-
"Which products had the biggest revenue change week over week? Top 5 up and top 5 down, with at least 10 purchases per week."
-
"Which product categories grew and which shrank this week versus last week? Sort by absolute revenue change."
-
"Compare revenue per session by source/medium this week versus last week. Any channel that shifted by more than 15%?"
-
"Are there any anomalies in my ecommerce data from the last 7 days? Flag anything unusual in product performance, category mix, or checkout completion."
Five prompts, fifteen minutes, and you have a complete ecommerce performance review. The standing Monday version of this is the weekly audit from the GA4 prompts library, extended with the product and category lenses.
The output from prompt 2 feeds your merchandising decisions. The output from prompt 4 feeds your media budget. The output from prompt 5 catches the things you didn't know to look for. If any of the five surfaces a problem, the deeper diagnostic prompts from earlier in this article are the follow-up.
What the GA4 ecommerce reports miss — and what to add
GA4's ecommerce data is strong on the what and the where. It's weak on the why. A few common gaps and how to work around them.
Margin and profitability. GA4 tracks revenue, not cost of goods. If you need profit-per-product analysis, you'll need to bring COGS data from Shopify, your ERP, or a spreadsheet. Some AI setups let you paste a CSV into the chat alongside the GA4 data — Claude handles this well for one-off analysis.
Inventory and stock levels. GA4 doesn't know whether a product is out of stock, low stock, or overstocked. A product with declining sales might be a demand problem or a supply problem. Cross-reference with your inventory system before acting on the GA4 signal alone.
Customer lifetime value. GA4 reports transactions, not customers over time. For repeat purchase rate, customer LTV, and cohort analysis, you need either BigQuery exports or a dedicated analytics layer. The GA4 MCP server overview covers the BigQuery + AI path for teams that need this level of depth.
Returns and refunds. GA4's refund event exists but is underimplemented on most sites. If your refund rate varies significantly by product, the revenue numbers in GA4 overstate the actual value of high-return products. Adjust mentally or pipe refund data from your backend.
FAQ
What GA4 ecommerce reports can AI pull? Any report the GA4 Data API supports — product performance, category breakdowns, funnel analysis, revenue by channel, AOV trends, item-scoped metrics like view-to-cart and cart-to-purchase rates. The AI composes the query from your plain-English prompt and returns the result as a table, chart, or narrative.
How do I see product performance in GA4 with AI? Connect GA4 to Claude or ChatGPT through a GA4 MCP server, then ask: "Rank my top 30 products by revenue for the last 30 days. Show item views, add-to-carts, purchases, and view-to-purchase rate." The AI returns the table in seconds. No exploration setup needed.
Can AI tell me why a product is underperforming? AI can narrow down where the problem sits — low views (visibility issue), low cart rate (product page issue), low purchase rate (checkout or pricing issue) — and suggest hypotheses. The full diagnosis walkthrough is in CRO with AI. The root cause usually requires looking at the page itself, not just the numbers.
What's the difference between GA4's built-in ecommerce reports and using AI? GA4's Monetization reports are pre-built and rigid. You see what Google decided to show. With AI, you describe the question — "which categories grew this week," "which products have the best view-to-purchase rate on mobile" — and the report adapts to the question. The flexibility difference compounds over time: the pre-built report answers one question; the AI chat answers twenty in the same session.
How often should I run ecommerce analytics? Weekly for the standing review (five prompts, fifteen minutes). Daily during sales events, launches, or after site changes. Monthly for category-level strategic review and AOV trend analysis. The weekly audit prompts are the starting template.
Does this work with Shopify, WooCommerce, and other platforms?
Yes. If your ecommerce platform fires standard GA4 ecommerce events (view_item, add_to_cart, purchase, etc.), the analysis works identically. Shopify's native GA4 integration, WooCommerce plugins, and custom implementations all produce the same event structure. The prompts above don't depend on the platform — they depend on the events being present in GA4.
Start with the Monday review. Five prompts, fifteen minutes, real answers. If your GA4 isn't connected yet, start a free trial and run the first product performance prompt on your live data — you'll see in one chat whether there's a product leak you've been missing.