Discovered Labs captured 547 Google AI Mode sessions and over 1,300 background requests in January 2026 — and found that Google is already attributing conversions via a URL parameter called adview_query_id. Your WooCommerce store can be receiving AI Mode-attributed traffic today, even though “AI Mode Ads” hasn’t appeared in your Google Ads dashboard. The full delivery, tracking, and remarketing stack is operational. The reporting column hasn’t shipped yet.

If you’re not capturing adview_query_id server-side, those conversions are arriving stamped — and your store is throwing the stamp away.

What ‘adview_query_id’ Is and Why Server-Side Capture Matters#

Discovered Labs published the first detailed network analysis of Google AI Mode in January 2026. Across 547 AI Mode flows and more than 1,300 requests, they identified a parameter Google stamps on outbound click-throughs from AI Mode: adview_query_id. It’s the internal ID Google uses to attribute downstream conversions back to a specific AI Mode query and ad-view event.

It is not currently exposed as a dimension in standard Google Ads or GA4 reports. That doesn’t mean it isn’t running. Google AI Mode reached 75 million daily active users and over one billion queries per month by early 2026 (Optimyzee, citing Google) — one of the fastest adoption curves for any new Google surface. Ad auctions inside AI Mode resolve server-side within roughly 60 milliseconds of a query, well before the visible AI response finishes generating (which takes 6+ seconds, per Discovered Labs).

A specific placement codenamed aimba — AI Mode Bottom Ads — has its full delivery infrastructure ready in the AI Mode JavaScript bundles. Visible ads aren’t deployed at scale yet. The plumbing is.

What Discovered Labs Actually Found in 547 Sessions#

The published technical observations are conservative and grounded in observable network traffic. Three findings matter for a WooCommerce store owner:

• The ad stack is running: Discovered Labs reports that Google runs complete ad delivery, tracking, and attribution systems in the background of every AI Mode session — even though ads are not visibly displayed in most responses yet.
• Attribution is already firing: Query-to-conversion attribution is already tracking user journeys via adview_query_id. Click-throughs from AI Mode arrive at the destination URL with the parameter attached.
• Remarketing audiences are populating: Users clicking through from AI Mode citations are being added to remarketing audiences today, before any of this surfaces in advertiser dashboards.

The plumbing is live. The reporting column hasn’t shipped. That’s the gap.

You may be interested in: Your Google Ads ROAS Is Lying (And Your WooCommerce Bank Deposit Proves It)

Why Most WooCommerce Stores Are Throwing the Stamp Away#

The standard WooCommerce tracking stack — a Google Tag Manager container, GA4 tag, Google Ads conversion pixel, Facebook pixel, a couple of WordPress plugins — runs in the browser. That’s the problem. There are three reliable ways the adview_query_id parameter gets dropped before any tag reads it:

• Ad blockers strip the request: 31.5% of global users run ad blockers (Statista, 2024). When a blocker prevents the GA4 or Google Ads script from loading, the parameter sits in the URL bar and never gets attached to a conversion event.
• ITP and consent banners truncate the session: Safari’s Intelligent Tracking Prevention caps first-party cookies set by client-side JavaScript at seven days. A user clicks an AI Mode citation, browses, leaves, comes back two weeks later — the session ID and the parameter are gone before the order fires.
• SPA navigation and lazy tagging lose the URL: Many WordPress themes and page builders manipulate the URL on internal navigation. By the time the checkout page fires its conversion tag, the original landing URL — the one carrying adview_query_id — is no longer in document.referrer or window.location.search.

This is the same pattern that’s eroded GA4 attribution for the last three years. Cross-device attribution in GA4 was the first crack. AI Mode attribution is the newest one. The mechanism is identical.

The Server-Side Capture Pattern That Surfaces the Cohort#

The fix is architectural, not configurational. You need to capture URL parameters at the request layer — before the browser has any chance to lose them — and persist them through the session into the order record.

The pattern has three moving parts:

1. Request middleware on the landing URL. When the browser hits your store’s first page, a server-side handler reads every query parameter — adview_query_id, gclid, gbraid, wbraid, fbclid, ScCid — and writes them to a first-party session record on your own subdomain. This runs before any client-side script loads, so ad blockers and ITP can’t interfere.
2. Session persistence keyed to your domain. The session record lives in a first-party cookie or server-side store keyed to the user’s session. ITP’s 7-day cap applies to client-side JavaScript-set cookies — server-set first-party cookies aren’t affected the same way.
3. Order meta write on checkout. When the WooCommerce woocommerce_checkout_order_processed hook fires, your server-side layer attaches the captured parameter set to the order’s meta record. The order now carries an unambiguous AI Mode origin stamp.

Capture at the request layer. Persist first-party. Stamp the order. Three steps, and the AI Mode cohort exists in your own database from day one.

What This Looks Like in BigQuery#

Once orders carry adview_query_id in meta, the cohort is queryable as soon as the data lands in BigQuery. A minimal query that surfaces AI Mode orders looks like this:

SELECT
  DATE(order_created_at) AS order_date,
  COUNT(DISTINCT order_id) AS ai_mode_orders,
  SUM(order_total) AS ai_mode_revenue,
  AVG(order_total) AS ai_mode_aov
FROM `your_project.woocommerce.orders`
WHERE meta_adview_query_id IS NOT NULL
  AND meta_adview_query_id != ''
GROUP BY order_date
ORDER BY order_date DESC;

This isn’t a future capability waiting on Google. The moment adview_query_id is captured into order_meta and streamed to BigQuery, you have a labelled cohort. You can compare AOV, repeat rate, and product mix against your existing gclid cohort. You can join against your CRM. None of this requires Google to release the dimension.

You may be interested in: First-Party Data Strategy

How Seresa Captures adview_query_id Today#

Here’s how you actually do this on a WooCommerce store. Transmute Engine™ — Seresa’s first-party Node.js tracking server — runs on your own subdomain and captures the entire landing URL, with every query parameter intact, at the request layer. adview_query_id, gclid, gbraid, wbraid, fbclid, ScCid all arrive together and persist into the order record automatically. The BigQuery destination ships those order events as a stream, so the AI Mode cohort is queryable the same day it starts existing.

Transmute Engine is a dedicated server, not a WordPress plugin. The lightweight inPIPE plugin in WordPress is the data courier; the engine on your subdomain does the capture, hashing, and routing. That’s the reason the parameter survives the round trip — the capture happens on your server, not in the browser where the parameter can be lost.

Key Takeaways#

• Google AI Mode is already attributing conversions via the adview_query_id URL parameter — confirmed by Discovered Labs’ January 2026 analysis of 547 sessions and 1,300+ requests.
• The reporting column hasn’t shipped, but the data is live. Stores capturing the parameter now will have months of cohort history before Google exposes it as a standard dimension.
• Client-side tracking can’t reliably capture this parameter — 31.5% ad blocker share, ITP’s cookie limits, and SPA URL manipulation all break it before tags can fire.
• Server-side capture at the request layer is the only reliable pattern — middleware reads the parameter, session state persists it, WooCommerce order meta stamps it.
• The cohort is queryable in BigQuery immediately — no need to wait for the Google Ads or GA4 column to appear.

Frequently Asked Questions#

Audit your WooCommerce store’s URL parameter capture before Google Marketing Live exposes the dimension. Start at seresa.io.