Anthropic launched Claude Desktop Live Artifacts on April 20, 2026, collapsing the cost of building a custom dashboard to a single prompt. Six dashboards every WooCommerce owner will want to build first—today’s revenue by source, an abandoned cart leaderboard, ROAS per campaign in the last hour, top product velocity, returning-versus-new split, and a stockout-risk tracker—each demand a specific server-side event stream in BigQuery. The default WooCommerce-to-BigQuery ETL syncs six entity types from the REST API and zero behavioural events, so five of the six dashboards return empty by default.
Why Six Dashboards, and Why Now
Anthropic shipped the template; the agency community translated it; the gap that remains is in BigQuery.
Anthropic launched Claude Desktop Live Artifacts on April 20, 2026, with three reference examples that set the template for everything that followed: a daily ad-spend tracker, a competitor watchlist, and a morning brief stitching email, calendar, and open pull requests into one widget (Anthropic documentation). Inside the first three weeks, the agency and consultancy community translated those templates into vertical use cases, and the WooCommerce version writes itself.
Four days after launch, Duke Digital Media Community published a 30-minute walkthrough showing how to assemble a dashboard from a single prompt. The build cost collapsed from “commission a Looker engagement” to “type two sentences.” That collapse is what makes the question worth asking now: which dashboards does a WooCommerce store owner want first, and what does each one need from the warehouse underneath?
Every dashboard a Shopify owner takes for granted is now a two-minute Live Artifact prompt—if the events underneath exist in BigQuery in a queryable shape.
The MCP connection is the easy part. Google’s remote BigQuery MCP server reached general availability in May 2026, authenticated via OAuth 2.0. The wiring story is finished. The data story is the one that determines whether any of the six dashboards actually return numbers.
The Six Dashboards Every WooCommerce Owner Will Want
List-shaped, each one short. For each: the question it answers, the event it needs, whether the default ETL has it.
These are the dashboards Shopify owners take for granted and WooCommerce owners traditionally commission. With Live Artifacts shipped, six of them are now prompt-shaped:
1. Today’s Revenue by Traffic Source
Answers the question every owner asks first thing in the morning: where did today’s sales come from? Needs purchase events with utm_source, utm_medium, and utm_campaign on the same row. The default WooCommerce REST API exports order rows with neither timestamps to the minute nor any traffic attribution. Default verdict: empty unless attribution is captured server-side.
2. Abandoned Cart Leaderboard
Lists items added to carts that didn’t convert today, ranked by total cart value at risk. Needs add_to_cart and begin_checkout events with cart_value and item_id, both of which are pre-purchase behavioural events GA4 enhanced ecommerce defines but WooCommerce REST API does not expose. Default verdict: empty.
3. ROAS per Campaign in the Last Hour
Joins purchase revenue against ad spend per campaign at hourly granularity. Needs purchase events stamped with campaign IDs AND ad spend from Google Ads, Meta CAPI, and Microsoft Advertising joined at the session level. The default WooCommerce stack has neither side of this join. Default verdict: empty—and the most expensive empty on the list, because most ad spend decisions get made on this signal.
You may be interested in: Claude Desktop’s BigQuery MCP: 90-Second Setup, Empty WooCommerce Results
4. Top Product Velocity (This Hour vs Same Hour Last Week)
Shows which products are accelerating right now compared to the same hour-of-week last week. Needs view_item and purchase events timestamped to the minute, plus a week of history with matching timestamps. The WooCommerce ETL captures orders at created_at granularity but not view events, so the “view” half of the velocity ratio is missing. Default verdict: order-side data only, no velocity calculation possible.
5. Returning vs New Customer Split
Splits today’s purchases by whether the buyer has shopped before. Needs a stable user_id stamped server-side across sessions, plus join history. The WooCommerce customers table has customer_id for logged-in users but no consistent identifier for guest checkouts that later return, and the orders table does not carry a session_id. Default verdict: partial—works for registered customers, fails for the guest-then-account flow that’s most retail buyers.
6. Stockout-Risk Tracker
Highlights products whose add-to-cart velocity exceeds remaining inventory in the next 24 hours. Needs current inventory levels joined to recent add_to_cart events per SKU. Inventory exists in WooCommerce products but the cart events do not exist anywhere in the default warehouse. Default verdict: half the join is missing.
One of six populated. Five of six empty. The pattern is consistent enough to predict the result before the prompt finishes typing.
The Pattern Behind All Six
Five dashboards depend on at least one event the REST API was never designed to expose.
The WooCommerce REST API exposes six entity types: orders, products, customers, coupons, refunds, shipping (WooCommerce REST API documentation). Every ETL connector pulls from those endpoints, so the BigQuery destination mirrors the same gap. GA4’s enhanced ecommerce specification recommends 12 events for retail, most of them pre-purchase: view_item_list, select_item, view_item, add_to_cart, view_cart, remove_from_cart, begin_checkout, add_shipping_info, add_payment_info, purchase, refund, view_promotion (GA4 documentation). Two map onto WooCommerce REST entities. Ten do not.
| Dashboard | Event Needed | In Default WooCommerce ETL? |
|---|---|---|
| Today’s revenue by source | purchase + utm_* | purchase yes, utm no |
| Abandoned cart leaderboard | add_to_cart, begin_checkout | No |
| Hourly ROAS per campaign | purchase + campaign + ad spend | purchase yes, rest no |
| Top product velocity | view_item + purchase | purchase only |
| Returning vs new split | stable user_id across sessions | Partial (registered only) |
| Stockout-risk tracker | add_to_cart + inventory | Inventory only |
Five of six dashboards need at least one behavioural event from GA4’s enhanced ecommerce spec. None of those events exist in the default WooCommerce-to-BigQuery ETL.
The MCP connection runs the SQL flawlessly. Claude Desktop describes the schema, generates the query, executes it. Five of the six results come back as empty tables. The model has no way to distinguish “this dashboard has no signal today” from “this dashboard has no source data at all”—both look the same from the query layer.
You may be interested in: The BigQuery Schema Every WooCommerce Live Artifact Will Quietly Demand
How One Prompt Returns All Six Dashboards
The fix is upstream at event capture, not in the prompt or the query.
Two paths fill the gap. The first is the GA4 BigQuery export, free on the GA4 standard tier, which writes one row per event with the full enhanced ecommerce schema. It works for behavioural data but it batches daily on standard GA4 and drops sampled rows under consent restrictions, so the “in the last hour” half of dashboards 3 and 4 still returns nothing useful before noon.
The second path is server-side capture. Events fire from WooCommerce hooks, get sent to a first-party Node.js server on the store’s own subdomain, and stream into BigQuery in seconds with stable user_id and utm parameters stamped on every row. Transmute Engine™ emits the full GA4 enhanced ecommerce event spectrum server-side—view_item, add_to_cart, begin_checkout, purchase, refund—directly to BigQuery alongside order data, so the same Live Artifact prompt that returned five empty tables now returns six populated dashboards.
The same prompt. The same MCP server. The same OAuth flow. The difference is one layer down at the event-capture seam, and it propagates through every dashboard the store will ever try to build on top of the warehouse.
The question isn’t whether Claude Desktop can build dashboards. The question is whether the warehouse it queries has the events those dashboards need. Five of six common dashboards say no by default.
Key Takeaways
- Live Artifacts launched April 20, 2026: Dashboards built from a single prompt, querying BigQuery via the MCP server that reached GA in May 2026.
- Six dashboards every WooCommerce owner wants first: revenue by source, abandoned cart leaderboard, hourly ROAS, product velocity, new-vs-returning split, stockout-risk.
- Five of six return empty by default: The WooCommerce REST API exposes 6 entity types and zero behavioural events. The ETL warehouse mirrors that gap.
- The pattern is structural: GA4 recommends 12 enhanced ecommerce events, most pre-purchase. Only 2 map onto WooCommerce REST entities.
- Server-side capture fixes all six at once: Stream behavioural events directly to BigQuery and the same prompt now returns six populated dashboards instead of one.
Live Artifacts are executable widgets Anthropic launched on April 20, 2026, that render directly inside Claude Desktop conversations. They can query live data sources (including BigQuery via MCP), display interactive charts and tables, and refresh on demand. The user builds them by describing the dashboard in plain language.
Any dashboard that depends on pre-purchase behaviour, session data, or campaign attribution. That includes abandoned cart leaderboards, hourly ROAS by campaign, product view velocity, returning-versus-new customer splits, and stockout-risk trackers. The default WooCommerce-to-BigQuery ETL only syncs completed orders and catalog data—five of the six common dashboards have nothing to query.
An ETL warehouse mirrors the WooCommerce REST API, which exposes six entity types and zero behavioural events. A server-side capture warehouse streams the full GA4 enhanced ecommerce event spectrum—view_item, add_to_cart, begin_checkout, purchase, refund—directly to BigQuery with stable user_id and utm parameters stamped on every row. The same Live Artifact prompt returns six populated dashboards instead of one.
No, but you do need the events GA4 recommends. You can capture them via the GA4 BigQuery export (free on GA4 standard, daily-batch), or via a server-side capture layer that streams them directly. Claude Desktop’s MCP connection doesn’t care which path put the events in BigQuery—it cares whether the rows are there to query.
Three things in BigQuery: behavioural events (view_item, add_to_cart, begin_checkout) timestamped to the minute, a stable user_id stamped server-side across sessions, and utm parameters joined to every event row. With those in place, one prompt produces all six dashboards. Without them, five of six return empty.
References
- Anthropic. “Claude Desktop Live Artifacts.” support.anthropic.com (April 20, 2026).
- WooCommerce. “REST API Documentation.” woocommerce.github.io/woocommerce-rest-api-docs (2026).
- Google Analytics. “Enhanced ecommerce events (GA4).” developers.google.com/analytics/devguides/collection/ga4/ecommerce (2026).
- Google Cloud. “Use the BigQuery MCP server.” cloud.google.com/bigquery/docs/use-bigquery-mcp (2026).
- Duke Digital Media Community. “30-minute Live Artifact dashboard walkthrough.” digitalmedia.duke.edu (April 24, 2026).
- Geeky Gadgets. “Live Artifact applications review.” geeky-gadgets.com (May 2026).
One prompt. Six dashboards. The MCP setup is 90 seconds—the warehouse decides whether five of them return empty. See how Seresa fills BigQuery with the events Live Artifacts actually need.
