Quick Answer: Claude Desktop Live Artifacts launched April 20, 2026 with native connectors for PostHog, Mixpanel, Datadog, Airtable, Apify, Gmail, Shopify and Zapier — but not Google Analytics or Meta Ads. Google’s BigQuery MCP server reached general availability in May 2026 with OAuth 2.0 and no API keys required. Together those facts collapse the architecture decision for any WooCommerce store wanting ad performance in a Live Artifact: the data has to live in BigQuery, because BigQuery is the only place Claude Desktop can natively reach and find it.
The Connector List That Forced the Question
Anthropic shipped eight native connectors at launch — none of which is GA4 or Meta Ads, and the absence pushes every WooCommerce store toward the same architectural decision.
Annika Helendi’s May 2026 practitioner review of Claude Cowork’s connector library confirmed what Anthropic’s documentation implies but never quite states: PostHog, Mixpanel, Datadog, Airtable, Apify, Gmail, Shopify and Zapier shipped as native MCP connectors at launch. Google Analytics did not. Meta Ads did not. The Anthropic roadmap has not committed a date to either, and the silence on both is loud — these aren’t obscure platforms that fell off a long list. They’re the two largest ad/analytics destinations for any e-commerce operator on the planet.
Meanwhile, in early May 2026 Google’s remote BigQuery MCP server hit general availability with OAuth 2.0 authentication and no API keys required for the operator. The two facts together — no native GA4 or Meta connector, plus a fully-supported BigQuery MCP — collapse the architectural question for every WooCommerce store that wants ad data in a Live Artifact. The data has exactly one place it can live and be useful.
Claude Desktop Live Artifacts ships with native connectors for 8 platforms — and neither Google Analytics nor Meta Ads is among them.
Four Places Your Data Could Live — Only One Works
Walk through every place a WooCommerce store actually stores ad and conversion data today — only BigQuery survives the connector test.
A WooCommerce operator looking at a Live Artifact dashboard build has four realistic data locations to choose from. Three of them dead-end immediately:
| Data Location | Native Live Artifact Connector? | Workaround Available? | Verdict |
|---|---|---|---|
| Google Analytics 4 | No | Apify scrape of GA4 UI | Compromised — modelled, sampled, rounded |
| Meta Ads dashboard | No | Apify scrape of Meta UI | Compromised — same UI penalties |
| WooCommerce admin | Shopify-only native connector | None for WooCommerce | Dead end |
| BigQuery | Yes (Google MCP, GA May 2026) | Not needed | The only working path |
Each row is worth pausing on. GA4 has no native Live Artifact connector and the Apify workaround scrapes the GA4 UI — which means the Live Artifact reads numbers that have already passed through sampling for high-volume properties, threshold suppression for low-cardinality dimensions, modelled conversions for consent-rejected sessions, and attribution-window remapping. The number on the dashboard is GA4’s interpretation, not the underlying events. Meta Ads is the same story — view-through windows, modelled conversions, attribution settings the platform applies before any number reaches a scraper.
WooCommerce admin is the closest miss. Anthropic shipped a Shopify connector at launch — making the WooCommerce gap structural, not technical. There is no equivalent WooCommerce MCP server in the roadmap, and Shopify’s native connector cannot reach WooCommerce data.
You may be interested in: How WordPress Events Reach BigQuery in Seconds
What a Useful BigQuery Dataset Actually Looks Like
BigQuery being the destination is half the answer — the other half is what’s in the rows, and most WooCommerce BigQuery datasets fail this part quietly.
“The data lives in BigQuery” is the start of the answer, not the end of it. The Live Artifact querying that warehouse will only be as useful as the rows it finds there. Three properties separate a BigQuery dataset Claude can actually use from one that frustrates every query:
- Every ad-traffic event has utm parameters stamped at the row level. Not at session aggregate. Not at user-property level. On every individual event row: utm_source, utm_medium, utm_campaign, ideally utm_content and utm_term. Without row-level stamping, “show me ROAS by campaign in the last hour” is a query that returns nothing useful.
- Conversion events are joined to ad-spend at session or campaign granularity — not just at daily aggregate. A purchase row needs to know which click brought the session in, which campaign that click came from, and what that campaign cost the same day. Daily-aggregate ad spend joined to event-level conversions is the bare minimum; campaign-day or session-level joining is what makes attribution panels work.
- Retention is indefinite. Year-over-year is the comparison every dashboard wants, and it requires more than 12 months of event-level history. GA4’s default retention is 2 months. A useful Live Artifact dataset retains events permanently.
A useful Live Artifact BigQuery dataset has three non-negotiable properties: utm-stamped ad events at row level, conversion-to-spend joins at session granularity, and indefinite retention.
Most WooCommerce stores running GA4’s standard BigQuery export are missing at least two of the three. The GA4 export preserves the raw events GA4 captured — which inherits GA4’s identity fragmentation, default retention, and the gaps where consent rejection or ad blockers blocked the events from firing at all. An estimated 20–30% of conversion events are lost client-side to ad blockers and Safari ITP — and a BigQuery dataset that started from GA4 inherits every one of those gaps.
You may be interested in: GA4 BigQuery Event Limit: What Happens at 1 Million Events
The Apify Workaround and Why It’s a Tax, Not a Solution
The May 2026 marketing community has converged on Apify as the GA4/Meta workaround — it works mechanically, and it pays a data-quality price every single query.
Install the Apify connector. Point a scraper at the GA4 UI or the Meta Ads dashboard. Feed the scraped numbers into a Live Artifact. The pattern is documented, the connector exists, and it works — for some definition of works. The Apify workaround feeds Live Artifacts whatever GA4 or Meta has already modelled, sampled, threshold-suppressed and UI-rounded — every Apify-sourced number has been through the platform’s interpretive layers twice.
The penalties are specific:
Sampling on high-volume properties: GA4 samples queries that exceed its threshold, and the scraped UI number is the sampled estimate, not the actual count.
Threshold suppression: Dimensions with low cardinality get suppressed for privacy, which means the scraped table has gaps that don’t exist in the underlying events.
Modelled conversions: Where consent was lost or cookies blocked, GA4 and Meta model what they think happened — and the scraper reads the model, not the data.
Attribution-window remapping: The dashboard shows conversions under whichever attribution window the platform has decided to display, not the actual click-to-purchase chain.
BigQuery queries the underlying events directly, before any of those interpretive layers touch the numbers. The Apify workaround is a useful bridge for ad-hoc inspection. It is not a foundation for a dashboard that has to be right.
How to Get Your Data There
The architecture pattern for getting WooCommerce ad and conversion data into BigQuery with the row-level properties Claude needs is well-trodden — server-side capture at the hook, streamed to BigQuery with attribution stamped.
Here’s how you actually do this. Transmute Engine™ is a first-party Node.js server that runs on your subdomain (e.g., data.yourstore.com). The inPIPE WordPress plugin captures WooCommerce hooks — every page view, add-to-cart, checkout, and purchase — and sends batched events via authenticated API to your Transmute Engine server. The server stamps each event with utm parameters resolved at session start, joins conversion events to the originating ad click, and streams the rows to BigQuery via the Streaming Insert API in seconds.
Every row that lands in BigQuery already carries the attribution data a Live Artifact needs to answer “which campaign drove this purchase” — at row level, on every event, with indefinite retention. The “where does the data live” question stops being a question. It lives in BigQuery, with utm-stamped rows, conversion joins at session granularity, and a history that compounds month over month.
Key Takeaways
The five operational truths emerging from the May 2026 Live Artifacts connector reality.
- Live Artifacts has 8 native connectors — GA4 and Meta Ads are not on the list: The architectural decision for WooCommerce stores is forced, not chosen.
- BigQuery MCP reached GA in May 2026: OAuth 2.0, no API keys, fully supported by Google — BigQuery is the only natively-reachable destination for ad/analytics data.
- Apify scraping works mechanically and fails analytically: Scraped UI numbers carry sampling, modelling, suppression, and rounding penalties that BigQuery queries don’t.
- Useful BigQuery datasets need three properties: utm parameters stamped at row level, conversion-to-spend joins at session granularity, indefinite retention.
- Server-side capture is how you get the row-level properties: Client-side tracking loses 20–30% to ad blockers and ITP before the row is ever written.
Frequently Asked Questions
Common questions from WooCommerce operators auditing their data layer for Live Artifacts compatibility.
Yes mechanically, but the dashboard reads whatever GA4 has already modelled, sampled, threshold-suppressed and UI-rounded. The Apify-as-intermediary workaround feeds numbers that have been through GA4’s interpretive layers twice — once on the GA4 backend, once on the scrape. BigQuery queries the underlying events directly, without GA4’s modelling layer sitting in between the data and the dashboard.
There is no committed date on the Anthropic roadmap for either. The connectors that shipped at launch cover product analytics, observability, productivity, and one e-commerce platform (Shopify). Roadmap silence on GA4 and Meta is significant — both are obvious candidates, and their absence at launch is a deliberate scope decision rather than an oversight likely to be reversed in weeks.
Three properties matter most. First, every ad-traffic event has utm_source, utm_medium, and utm_campaign stamped at the row level — not at session aggregate. Second, purchase events are joined to ad-spend at the campaign or session granularity, not just at daily aggregate. Third, retention is indefinite so year-over-year comparisons are possible. Without these, the BigQuery MCP connector queries data that can’t answer a WooCommerce operator’s actual questions.
It’s a starting point, not a finished dataset. GA4 BigQuery export gives raw event rows but inherits GA4’s identity fragmentation across cookieless and consent-rejected sessions, retains data only as long as your GA4 retention setting allows (default 2 months), and doesn’t include Meta Ads spend or any non-Google channel. Useful tables typically combine GA4 export with first-party server-side events and a separate ad-spend ingestion job.
Meta has no first-party BigQuery export, but several ingestion paths work: Supermetrics or Funnel.io as managed ETL, the Meta Ads Insights API written to BigQuery via a scheduled Cloud Function, or server-side conversion capture via the Meta CAPI alongside an ad-spend pull. The destination is the same regardless: a BigQuery table joined to your event-level data on date and campaign.
References
Primary sources for the Live Artifacts connector inventory, BigQuery MCP availability, and data-quality claims.
- Anthropic (April 2026). Claude Desktop Live Artifacts launch and connector documentation. Source
- Annika Helendi (May 2026). Claude Cowork Live Artifacts practitioner review and connector library audit. Substack. Source
- Google Cloud (May 2026). BigQuery MCP server general availability announcement with OAuth 2.0 authentication. Source
- Google Analytics (2025). GA4 BigQuery Export documentation — schema, retention, and event-level row structure. Source
- Statista (2024). Global ad blocker usage — 31.5% of internet users worldwide. Source
- WebKit / Apple (2025). Intelligent Tracking Prevention — 7-day first-party cookie limit documentation. Source
- Geeky Gadgets (May 2026). Claude Desktop Live Artifacts integration coverage and use case review. Source
Your ad data has one place to live in the Live Artifacts era. Start the architecture decision at seresa.io.
