January 2026 marks a turning point for e-commerce and analytics. With the launch of the Universal Commerce Protocol (UCP) at the NRF trade show in New York last week, Google is accelerating the emergence of agentic commerce: AI agents capable of discovering, comparing… and buying products without going through a website.

⚠️ Note: Google is not the first. OpenAI and Microsoft have already paved the way.

👉 What really changes today is scale: Google can industrialize these uses via Search, Gemini, Wallet/Pay and Merchant Center.

🧠 The click is no longer the starting point

For 20 years, our models were based on a simple funnel:

Visitor → Session → Pages → Conversion

With agentic commerce, this model is exploding. The new “funnel” becomes :

Intention → AI agent → API call → Transaction

➡️ A transaction can take place entirely within a conversational interface, via API exchanges between the AI agent and the merchant’s systems.

✅ No page view. No session. No JavaScript. ➡️ Customer-side tracking becomes partially blind.

📉 The session loses its analytical value

If purchased in Gemini or Search AI Mode :

  • no Google Analytics or Piano Analytics sessions are created,
  • onsite funnels become incomplete,
  • measured sales may be underestimated.

🔁 New key indicators to watch :

  • qualified conversational intentions,
  • successful API calls,
  • quality and completeness of product flows,
  • ability to respond correctly to AI agent requests.

👉 Performance moves from the site to the data exposed to agents.

🧱 Server-side becomes the norm

In a “checkout in conversation” world, the source of truth becomes the back office.

To measure sales correctly :

  • server-side tracking becomes indispensable,
  • Google Analytics (Measurement Protocol) and Piano Analytics (API server-side) must be fed directly from OMS.

🔐 Consent Mode, privacy and RGPD

With :

  • partially off-site routes,
  • machine-to-machine API exchanges,
  • the gradual disappearance of third-party cookies,

👉 consent becomes a central technical signal.

Key implications:

  • server-side events are based on explicit, traceable consent,
  • analytics is evolving towards a more privacy-by-design model, aligned with RGPD.

🧬Identity Linking: attribution becomes deterministic

WithIdentity Linking:

  • the identity (e.g. Google account) can be securely linked to the merchant account,
  • with consent, at the time of purchase,
  • and cross-surface continuity (Search, Gemini, Wallet).

📌 Result:

  • more deterministic attribution,
  • ability to link an AI conversation to a real transaction,
  • ROAS calculated on the basis of the prompt, not the click.

🎯 The prompt becomes a touchpoint

OpenAI and Microsoft have shown that buying “in conversation” works. Google can now go to scale.

In this context :

  • the prompt becomes a marketing contact point,
  • conversation becomes a new funnel,
  • the website is no longer the only option.

🛠️ What analytics teams need to do now

At Optimal Ways, we can already see the roadmap to 2026 taking shape:

✅ Hybrid tracking: client-side + server-side

✅ Increased maturity of product flows(Merchant Center = strategic asset)

✅ Multi-source reconciliation(Google Analytics, Piano Analytics, OMS, platforms, data warehouse, …)

✅ Redefining KPIs: less navigation, more intent and transaction

🧩 In conclusion

OpenAI and Microsoft have shown the way. Google is making it massive.

For Digital Analytics :

  • the website is no longer the only place for conversion,
  • the measurement relies more on the server-side, the actual transaction and theconsented identity.

We will continue our ongoing monitoring of this topic. Stay tuned for future editions of our Optimal Insights newsletter!