How a tag manager works
The principle is simple: a developer inserts the GTM container code (a few lines of JavaScript) into the site once. After that, every tag is managed inside the Tag Manager interface without touching the site code again.
The basic concepts:
- Tag — the script to execute (a pixel, an analytics tag, a chat widget)
- Trigger — the condition that fires the tag (page load, a click, a dataLayer event)
- Variable — a dynamic value (product ID, order value, page URL)
Trigger: Purchase event → dataLayer
Tag: GA4 → conversion event "purchase", send order_value
Variable: {{dlv - order_value}} = 49.90
The data layer: the foundation of accurate tracking
A tag manager is at its most effective in combination with a dataLayer — a structured JavaScript object through which the site passes events and data. Without a dataLayer the tags are blind: they see URLs and clicks, but not the context (which product was added to the cart, which funnel step the user is on).
GTM versus direct embedding
| Criterion | Tag manager | Direct integration |
|---|---|---|
| Speed of launch | Fast — no deployment needed | Slower — a deployment is required |
| Control of load order | Limited by tag order | Full control |
| Dependency on developers | Minimal | High |
| SLA reliability | Lower | Higher |
| Recommended for | An MVP and first tests | Production under heavy load |
Common mistakes with GTM
- A tag with no data layer. Tracking events “by clicks” instead of by dataLayer events produces unreliable data: one click can fire several events, or none at all once the markup changes.
- Duplicated tags. Analytics and pixels added both through GTM and directly in the code mean conversions are counted twice.
- An unchecked container. GTM has a Preview & Debug mode — always test a tag there before publishing to production.