AI browsers are creating new challenges for digital measurement. Built-in ad blockers, cookie restrictions, and consent blockers can all reduce the visibility marketers and analysts rely on to understand performance. At the same time, AI-driven traffic and emerging protocols like UCP are starting to reshape how online journeys may be tracked.
In this webinar, we’ll look at what these changes mean in practice - from recognizing AI traffic to understanding where tracking breaks and how to make it more resilient.
Do you also help set up proxy servers through Cloudflare (or another CDN) so cookies persist longer? What’s the best setup for a local business website? What would you recommend?
Yes—if your goal is ITP bypass and longer-lived cookies, I recommend using a same-origin setup for your sGTM subdomain.
For your case, same-origin is usually the best approach. Along with bypassing ITP, it also lets you load gtm and gtag.js via your CDN.
Have you heard of—or do you see any changes in—how Gemini is now embedded in Chrome? Or do you think it will just adopt default Chrome behavior?
I haven’t noticed any direct impact on tracking from this yet. However, I do expect it to influence something else: the volume of organic traffic from search engines.
Does Stape Bot Detection Power-Up detect AI bots, or do we need additional detection?
Bot Detection mainly relies on user agent plus a database of known bot/spam IPs. It may catch basic crawlers, but I don’t expect it to reliably detect AI/agent traffic—especially if it looks like normal Chrome/headless Chrome. If you need AI-bot identification, you’ll likely need additional tools. I see Bot Detection primarily as spam filtering, not AI-agent detection.
Should we load the CMP via a custom loader / GTM? What’s the recommended approach?
I recommend a file proxy approach so CMP/banner scripts load from your first-party domain. You can do this via Stape’s File Proxy Power-Up, or via the File Proxy Client template in the server container (GitHub). You configure the origin and your first-party path, and the CMP script is served from there.
Does an AI browser interact with the CMP (cookie banner)?
Sometimes it might (especially in an “agentic” mode if you instruct it to click), but in practice I’ve seen cases where AI browsers block consent banners before they even render—so I’d assume they’ll often block it.
If I only use server-side GTM, will I miss data because I don’t have webhooks?
No—server-side GTM can receive and handle webhooks, so you won’t miss data for that reason. The bigger risk is relying on client-side GTM only, because in AI-browser flows, it may not load at all. For best coverage, I’d use server-side GTM + webhooks.
Meta/Google push their own server-side solutions (gateways). What’s the difference vs your approach?
In my opinion, gateways usually cover only a small portion of what a full server-side setup can do. They’re a good starting point if you currently have nothing beyond client-side tracking, but they won’t replace a tailored server-side GTM implementation long-term.
What’s the main distinction between agent traffic and “standard bots/crawlers” (if any)?
I don’t think there’s a single clear indicator. Detection usually relies on multiple signals, and some approaches require CDN/server-log access. If there were one simple flag, this would be easy—so I’d expect ambiguity.
Is there a way to prevent AI browsers from blocking the CMP?
Yes. I’d use the same strategy as for preventing GTM/gtag blocking: proxy and obfuscate CMP scripts. Serve the banner scripts from your first-party domain and avoid obvious filenames/paths (don’t use “cookiebanner/cookiebot”; use neutral names). You can use Stape’s File Proxy Power-Up or the File Proxy Client template.
Do we need to invest in data engineering/analytics infrastructure to avoid being negatively impacted by these changes?
Maybe. I don’t want to prescribe specific investments, but at minimum I’d say: if you haven’t already, invest in server-side tracking.
What’s your vision for the future of consent in the EU?
I expect consent banners are here to stay. I’m skeptical that “omnibus” changes will remove them entirely. We’ve also done a webinar on consent—I’d recommend checking that out.
Do you agree that all browsers will eventually become AI browsers?
Very likely, yes. That said, if Chrome becomes “AI,” it may not create the exact same challenges as AI-first browsers. I also wouldn’t be surprised if Google introduces changes that help.
How can we reliably segment AI traffic as a channel in GA4?
I’ll share the slides/deck afterward, and I recommend an article by Dana DiTommaso on GA4 channels for deeper context and examples.
Will UCP require businesses to fundamentally change their data infrastructure?
For most businesses, I think probably not—platforms like Shopify and other CMS/inventory systems will likely adopt/support it. If you run a highly custom-built stack, you may need to adapt over time—especially if you want to sell via LLM/agent-driven flows.
How do we set up event tracking for a custom app (mobile app)?
I’d use a mobile SDK—for example, the Firebase SDK. Firebase can send events into a server-side GTM container, and from there you can route events to destinations (Meta, etc.).
If UCP enables AI agents to fetch/negotiate via API, how can we log these agentic queries—and can we treat them as top-of-funnel signals?
Conceptually, I can treat agent fetches as top-of-funnel signals. But tracking is complicated: if agents fetch assets directly (like images via direct URL), there’s no HTML load, so no pixels/tags. You might track via middleware/proxy/CDN logs, but I don’t see a clear general solution here—with Stape or without Stape—under typical conditions.
[Stape webinar] Consent explained: practical implementation for web and server-side tracking
Learn how consent works in practice and how to implement it correctly for analytics and advertising. Clear explanations, real demos, and practical guidance.
February 2026 release: Logs and Monitoring features update
Check the new Stape update: simplified log types, automated incoming request tracking, and a general statistics dashboard for easier performance monitoring.
Comments