View Transitions Are Growing Up — Is Your Tracking Ready?

Bramus Van Damme has spent the better part of 2026 road-testing the View Transitions API on conference stages — Beyond Tellerrand in April, then All Day Hey! in Leeds in May. Two separate talks, same thesis: this API is no longer a curiosity. It’s ready to ship, and smart frontend teams are shipping it.

For the engineers in the room, that’s exciting. For everyone managing analytics, consent pipelines, and tag architecture, it’s a quiet alarm going off in the background.

What View Transitions Actually Change for Tracking

The View Transitions API lets browsers animate between page states — including across full navigations, not just within single-page apps. The effect is cinematic. The tracking implication is structural.

Traditional tag managers fire pageview tags on DOMContentLoaded or window.load. When a view transition intercepts a navigation, neither event fires in the way your GTM container expects. The DOM updates, the URL changes, the user perceives a new page — but your analytics stack may see nothing at all, or worse, fires twice due to race conditions between the transition lifecycle and your existing triggers.

This isn’t hypothetical. The same issue plagued early SPA implementations on Angular and React before the industry standardised on history-change triggers. View Transitions introduces a new lifecycle: pageswap, pagereveal, and the ViewTransition promise chain. None of these map cleanly onto existing GTM trigger taxonomy without deliberate configuration.

The SPA Tracking Playbook, Revisited

If your team already has robust SPA tracking in place — history-state listeners, custom virtualPageview events pushed to the data layer, sanitised URL deduplication — you’re 60% of the way there. The principles transfer. The implementation details don’t, entirely.

The critical audit point is your routeChange or pageview trigger logic. Specifically: is it listening for popstate and pushState overrides, or is it relying on the URL changing in the address bar? With cross-document view transitions (the genuinely new territory Bramus is evangelising), the browser handles navigation natively, which means pushState may not enter the picture at all.

The practical fix: instrument a custom event at the pagereveal lifecycle stage, push it to the data layer with a clean page path payload, and fire your pageview tags off that event rather than any DOM-readiness hook. Then QA obsessively — transition duration, back-navigation, interrupted transitions (user clicks before animation completes), and bfcache restores all need their own test cases.

---

Consent Mode and the Transition Timing Problem

Here’s the wrinkle that keeps me up at night: consent state.

In a standard navigation, your consent management platform initialises on page load, resolves user consent status, and signals downstream tags accordingly. Google’s Consent Mode v2 depends on this resolution happening before any measurement tags fire. In a cross-document view transition, the incoming document initialises its own scripts — but the timing relationship between CMP initialisation and the transition lifecycle completing is non-trivial.

If your CMP takes 200ms to resolve consent state, and your view transition completes in 180ms, you have a window where a pageview tag could fire before consent is confirmed. In the EU, that’s a compliance exposure. In Southeast Asia, where PDPA enforcement in Thailand and the evolving Personal Data Protection frameworks across the region are tightening, this matters increasingly.

The mitigation isn’t complicated, but it requires intentional sequencing: gate your custom pageRevealReady data layer event on a consent-resolved check, not just on transition completion. Your CMP’s callback — whatever form it takes — needs to be part of the trigger chain, not a parallel process.

What Frontend and Analytics Teams Need to Agree On Before Shipping

The gap between a beautiful View Transitions implementation and a broken analytics implementation is almost always a coordination gap, not a technical one. Engineers ship the feature; analytics finds out in the data three weeks later when session counts look wrong.

A minimum viable handshake between both teams before any View Transitions deployment:

1. Define the data layer contract. Agree on the event name, the payload schema (page path, page title, previous page), and who owns instrumentation — the frontend engineer or the tag manager. Document it. Put it in the PR checklist.

2. Agree on what counts as a pageview. If a user clicks a link, the transition starts, and they click something else before it completes — did a pageview occur? On most sites the answer is no. Make sure your trigger logic reflects that, not just your verbal agreement.

3. QA plan is non-negotiable. Test in Chrome (where support is most mature), test in Safari (where it’s newer), and test on mid-range Android devices common in markets like Indonesia and Vietnam, where transition performance under memory pressure can behave differently. A transition that stalls on a Redmi device may leave your tag firing in an undefined state.

Bramus has done the frontend community a service by pushing this API hard through the conference circuit. The implementation patterns are solidifying. Now the tracking layer needs to catch up — deliberately, not reactively.

As more Southeast Asian e-commerce and content properties adopt View Transitions to close the experience gap with native apps — a gap that matters acutely in markets where users hold both a Shopee app and a brand’s mobile web experience to the same standard — the pressure to get this right will only increase. The question worth sitting with: is your measurement infrastructure designed to evolve with the browser platform, or is it still assuming a world where pages load cold every time?

---

At grzzly, we work at exactly this intersection — helping marketing and engineering teams in Southeast Asia build tracking architectures that hold up as web technology moves fast. Whether you’re planning a View Transitions rollout or untangling a data layer that grew organically over three platform migrations, we’ve seen the failure modes. Let’s talk