Web analytics
Browser based analytics loses events. Ad blockers stop the request, cookie restrictions shorten what can be remembered, tracking prevention breaks session continuity, and a page that closes before the tag finishes never reports at all. None of it surfaces as an error, so the gap is invisible in the reports themselves.
Web analytics collects the same events through a subdomain of your own site and delivers them server side. The request is first party rather than third party, and each event is validated and transformed before it reaches its destination.
Sources and destinations
Section titled “Sources and destinations”Every source below can feed every destination below. There is no Integration Type step for this use case.
Supported destinations
| Category | Supported |
|---|---|
| Warehouse | |
| Analytics |
The same website tag that feeds your conversion destinations feeds this one. You do not instrument the site twice.
Google Analytics 4 and Mixpanel are analytics platforms that read the events directly. BigQuery is the warehouse option: events land as rows in a dataset in your own Google Cloud project, where you query them yourself and join them to whatever else you hold. Nothing stops you connecting more than one.
Setting up the subdomain
Section titled “Setting up the subdomain”Events are collected through a subdomain of your own site rather than a Datahash address. This is what makes the collection first party, so cookies survive browser restrictions and ad blockers treat the request as internal rather than external. It has to be in place before any tag will deliver.
- One subdomain per project. If the project already has one, the tag work is all that is left.
- Set it up in Studio, providing the domain and the subdomain name, for example
s2sortracking. - Studio generates the DNS records. Add them at whichever provider manages your domain, either NS records or a CNAME depending on what you are given. Any DNS provider works.
- The subdomain has to sit under the main domain, and cannot be hosted as a primary domain in its own right.
DNS changes take time to take effect, so a step that fails immediately after you add the records is usually worth retrying rather than debugging. See Subdomain setup for the record types and the per-provider steps.
How the data is collected
Section titled “How the data is collected”Website traffic reaches Signals through a tag. You have two options.
- NeoTag is one tag serving multiple platforms. Install it once and add destinations without touching the site again.
- dhPixel replaces a single platform’s base code. Use it when you are working platform by platform.
Both hash personal identifiers in the browser, so the raw value does not leave the page. Plan your analysis around hashed values rather than expecting readable email addresses in the destination.
For Google Tag Manager, add a Custom HTML tag and paste the base code. A Migrate Now option signs in to GTM and migrates existing tags, which then move out of the Non-Migrated section. See Website tag setup for the full reference on both tags.
What reaches the destination
Section titled “What reaches the destination”One event framework serves every destination. The events most implementations send are page views, product views, search, add to cart, begin checkout, purchase, sign-up, lead generation, and custom business events of your own. The same event can go to more than one platform without being defined twice.
Analytics destinations identify people differently from advertising ones. There is no audience to match and no attribution window to respect. What matters is that each event carries the identifier its destination uses to attach the event to a person, which you choose when you connect that destination. An event arriving without one is delivered and then cannot be attributed to anything.
BigQuery has no identifier to nominate. It takes what the source sends and writes it as it arrives, so the joining is yours to do afterwards. Since the tag hashes in the browser, hash the other side of the join the same way: Apply SHA-256 hashing to PII has the BigQuery expression.
Agree event names before go-live. They are what every report, filter and saved view in the destination is built on, and renaming one later leaves the old name in everything already built on it.
The server side path can run alongside an existing browser side implementation. Where both run against the same property or project they report the same activity, so agree which one the business reads.
See Google Analytics 4, Mixpanel and Google BigQuery for the credentials, configuration and required fields for each destination.