Can this method be used only for promotions?

No. The flow stores information that arrives with one event on the server side and carries it to a later event from the same visitor; which information is carried depends on the rule. UTM source and medium, referrer or list source can be carried the same way for enrichment. Promotion has standard item fields; the others go as custom item parameters, are read under `items.item_params` in the BigQuery export, and need an item-scoped custom dimension to show up in the GA4 interface. This post covers only the promotion context.

Can the record be carried across devices?

Yes, when the key is a `user_id` tied to a signed-in user or another stable identifier sent as an event parameter; since the document path is a variable, reads and writes stay the same. Clicks are mostly anonymous, so a step that merges the `client_id` document into the `user_id` document at login is needed. The record then becomes the history of an identified person, and retention and deletion obligations grow with it.

What changes when the carrier runs in sGTM instead of the browser?

The promotion record is kept server-side in Firestore rather than on the device, and it is written onto `purchase` items in sGTM. For a banner that leads to a product, no cookie or localStorage record has to travel in the browser between click and purchase, and the enrichment rule lives in one place. The consent boundary does not change: in this example the Data Tag fires only with `analytics_storage` consent, so a declining visitor's click never reaches the server.

How is a banner that leads to a collection page tied to a purchase?

The click sends an itemless `select_promotion`, and the active promotion is kept in the browser with the target list ID. When the visitor touches a product from that list with `select_item` or `add_to_cart`, the promotion fields are written to that product's item object, and since that event passes through sGTM, the carrier works as it does for a banner that leads to a product. If the product is added to the cart by another path, no binding happens.

Is the GA4 Promotions report enough for banner performance?

Not for collection banners. On the sales side, the report shows only the products the promotion was bound to; on the impression and click side, the way it handles itemless events does not match the documented behavior. Impressions, clicks and CTR are calculated in BigQuery from event-level promotion parameters; the query is in the post.

Does this setup work under high traffic?

Volume is not a limit. Each event does one read and one write on a single user's document (`client_id` in this post), with no shared record or counter. The issue lies in the data model rather than volume: if two events from the same user arrive back to back, a read and write without a transaction can overwrite one of the records. This was not observed in the test environment, but the rate depends on concurrency, not volume; in production this risk should be closed with a writer that uses a transaction.

Carrying a Banner Click to the Purchase: Item Enrichment with sGTM, Firestore and BigQuery

In sGTM, information carried by one event can be stored in Firestore and written server-side onto the items of a later event from the same visitor. This post applies item enrichment to the gap left by GA4 not carrying promotion data between events: select_promotion is written to Firestore under the GA client_id key and added to matching items when purchase arrives. Architecture, setup and BigQuery reading are shown on a sample flow for banners that lead to a product and to a collection.

TL;DR

In sGTM, context carried by one event can be stored in Firestore under a key and written onto the items of later events. With client_id as the key, the carry stays within one browser; with user_id or another stable identifier, it can work across devices. Here, item enrichment carries the promotion data GA4 drops from select_promotion to purchase items under the client_id key, verified with the BigQuery export; Promotions report quantity and item revenue matched the export.

Get Access Sign In

Subscribers Only

The full article plus 26 more subscriber-only posts come with a subscription.

analytics

Subscribers Only

RELATED

Linking a Banner Click to a Purchase: Promotion Events, the Carrier Layer and the Consent Boundary

Session Integrity from Landing to Purchase: Counting the Break in BigQuery

Measuring Purchase Lag: Google Ads Reports, the GA4 Gap and the Window Decision

Deriving ID and Slot for E-Commerce Homepage Banners

Measuring Consent Rate: Where Accepts, Rejects and Later Changes Get Counted

sGTM or Data Manager: The Wrong Question and the Right Distinction

Joining Google Ads and GA4 at Row Level: A Shared Key and a Two-Layer Reading

GA4 Attribution Reports: Direct Inflation, Model Selection, and Finding the Real Source

Why Google Ads and GA4 Conversions Never Match: Counting Unit and Time Axis

Connecting a Shopify Order to an Ad Click: Choosing a Join Key for Offline Conversions