Skip to content
DocsSearch the docsEnglishBack to siteOpen console

Data streams

A data stream is one tag with its own tag ID. The data stream decides which domain the tag runs on, what is measured automatically, and which server keys are bound to it.

What a data stream actually is#

A data stream is the smallest unit of collection in AdPix: one tag, one tag ID, one domain. A data stream sits under a website, the website under an account, the account under an organization.

Everything about how data is collected is held on the data stream itself, not on the website:

On the data stream What it decides
Tag ID The tag's identity. It ships in the snippet and the collector resolves the organization and website from it.
Domain The hostname the tag runs on. It drives the install check and the receiving indicator.
Collection mode Direct tag or Tag-Gateway. Decides where the tag is served from and where the beacon goes.
Enhanced measurement Which automatic events (scrolls, outbound clicks, downloads, forms, video, site search) are on.
Configure your domains The cross-domain conditions that feed the linker and the self-referral exclusion.
Heatmaps On or off, sample rate, modes, excluded paths and a stop date.
Session replay On or off, sample rate, masking, recording length, excluded paths and a stop date.
Server keys The credentials for server-to-server sends, bound to this data stream — created and revoked under Integrations.

That is why two sites with different settings cannot share one tag: the settings live on the data stream.

Why the ID belongs to the data stream, not the account#

Every beacon that leaves the browser carries exactly one identifier. The collector uses that single string to look up the organization, the website and the data stream, and writes the event where it belongs. An account-level ID would leave nothing to tell one of your sites from another at that boundary.

The shape is fixed: AP- plus 10 uppercase hex characters, for example AP-A1B2C3D4E5. A unique index on that column guarantees no two data streams — in any organization — share one ID.

The ID is public, the server key is not

The tag ID is visible in every page's source and that is fine; it is not a credential. A website's real credential is a server key, created on purpose under Admin → Integrations and shown once. That one never goes in a page.

Where the list lives#

1
Open Admin in the console.
2
Under the Website column, select Data streams.
3
The website's data streams are listed with their name, domain, tag ID and receiving status.
The website's data streams and each one's tag ID.
The website's data streams and each one's tag ID.

The list shows every data stream of the website. They are all web data streams, so the list is not split into tabs; an app backend sends its events through the server-to-server API. Under the list sits the Server keys card, covered in Server-side tracking.

Creating a data stream#

1
Click Add data stream.
2
Enter the website URL and a data stream name. The name is only for you — what matters is the domain.
3
Click Create & continue. A new data stream uses the direct tag; the Direct tag / Tag-Gateway choice appears only where Tag-Gateway collection is available.

The installation instructions open immediately afterwards with the tag snippet carrying this data stream's tag ID. No key is created with a data stream; a module that needs a server key gets one under Integrations.

Under the snippet, the Website ID, the Collector Endpoint (where the tag sends its events) and the Tag ID each have their own Copy button — the three values a CMS plugin, a tag manager or a developer asks for.

Name and domain stay editable afterwards. The tag ID does not.

What is inside a data stream#

Click any row in the list to open Data stream details.

Data stream detail: domain, tag ID and enhanced-measurement settings.
Data stream detail: domain, tag ID and enhanced-measurement settings.

At the top is a status banner: either "Data collection is active in the past 48 hours" or the warning that collection is not active, with a View tag instructions button next to it. Below it the Data stream details card shows four values — name, URL, data stream ID and tag ID — followed by these rows:

Row What it does
Enhanced measurement Turns this data stream's automatic events on and off.
Modify events / Create custom events Takes you to the website's event settings; those are website-scoped, not tracking-code-scoped.
Heatmaps Optional click, move and scroll capture with sampling and a stop date.
Session replay Optional recording of a sample of sessions to watch back. See Session replay.
Server keys A pointer only: server keys are created and revoked under Integrations.
Configure your domains Cross-domain conditions. The row itself summarises how many are set.
Tag-Gateway (first-party) Generate the proxy bundle and run the diagnostic for first-party mode. Shown where Tag-Gateway collection is available, and always on a data stream that already has a gateway.
View tag instructions The snippet, plus the Test installation button.

The two collection modes#

Mode Tag served from Events sent to
Direct tag The AdPix CDN host The AdPix collector, cross-origin and cookieless
Tag-Gateway A subdomain of yours, metrics.<your domain> by default That same subdomain, proxied through to the collector

The direct tag needs no setup and is the right way to start. Tag-Gateway needs a DNS CNAME and a reverse-proxy config on your side, and in exchange both the tag and the beacon become first-party. Generating a gateway bundle never disables the direct tag: the instructions screen offers both snippets side by side and you choose one.

Tag-Gateway is not offered for new data streams at the moment. A data stream that already has a gateway keeps it, along with its Tag-Gateway screen, and keeps collecting through it.

When you need a second data stream#

This is the part of the screen that causes regret later, so be precise about the boundary:

  • A genuinely different domain, or settings that must differ — create a second data stream. The main site and the blog on two domains, with heatmaps on only one of them, is the classic case.
  • Subdomains of the same domain — no second data stream. One data stream on example.com covers its subdomains.
  • Separate reports for two sites — a second data stream does not do this. Create a separate website.
Data streams under one website share one pool

Events do not carry a data stream id. Anything sent with the tag ID of any data stream under a website lands in the same set of reports, separable only by the page's hostname. If two sites need separate numbers, the separation has to happen at the website level, not the data stream level.

One small but consequential detail: the website's primary domain is derived from the first data stream that has a domain, and is kept in sync automatically. The order in which you create data streams matters here.

The receiving indicator and the install test#

Each data stream in the list carries either "Receiving traffic in past 48 hours" or "No data received in past 48 hours". That indicator works off the data stream's domain: events whose page hostname equals the data stream's domain, or is a subdomain of it, are credited to that data stream. A data stream with no domain set yet counts any traffic on the website as receiving.

Test installation, in the tag instructions, is the stricter check and also persists the data stream's verification status; each state is explained in verify data is arriving.

Deleting a data stream#

Deleting a data stream stops the tag and revokes every server key attached to it. You have to type the data stream name to confirm. The events already collected stay in reports, but the tag ID does not come back — so any page still carrying that snippet sends into nothing from that moment on. Remove the snippet from the site first, then delete the data stream.

Frequently asked questions#

Is the tag ID a secret?

No. It is visible in the source of every page on your site and is meant to be — all it does is tell the collector which website an event belongs to. The secret is the data stream's server key, a credential for server-to-server sends that is shown exactly once.

Do I need a separate data stream per subdomain?

No. The data heartbeat and the receiving indicator credit subdomains of the data stream's registered domain to that data stream, so one data stream on example.com also covers shop.example.com. A second data stream is for a genuinely different domain, or for settings that must differ.

If I create two data streams under one website, do the reports separate?

No. Events do not carry a data stream id, and every data stream under a website feeds the same reporting pool. If you want separate reports, create a separate website with one data stream in each.

If I delete a data stream, do I lose the data it collected?

No. Deleting a data stream stops that tag and revokes its server keys, but the events already collected stay in your reports. It cannot be undone, and the tag ID does not come back.

Build with the APIUnderstand where revenue comes from.
Was this page helpful?