Most marketing teams are still optimising for one version of the web. The customers they want already operate across three. Microsoft's framing is direct. The human web of tabs and searches drives most of the traffic today. The LLM web of AI Answers is where decisions are increasingly shaped. The agentic web of autonomous purchase flows is where a growing share of conversions will happen tomorrow.
The numbers underneath this story are the part most planners are underestimating. Automated traffic is growing eight times faster than human traffic. AI driven sessions nearly tripled in 2025. Agentic browser traffic is up roughly 8,000 percent year on year. This is no longer a thought experiment.
What changes across the three eras
Each era of the web has a different unit of attention and a different way of awarding the click.
**Help me find it.** The human web is search boxes, browsers, and comparison sites. The unit of attention is a session. Advertising wins on relevance to the query and quality of the landing experience.
**Help me choose.** The LLM web is AI Answers, Copilot conversations, and synthesised recommendations. The unit of attention is a conversation. Advertising wins by being included as a credible option in the answer.
**Do it for me.** The agentic web is autonomous agents that find, evaluate, and complete purchases on a user's behalf. The unit of attention is an outcome. Advertising wins by being legible to machine readers and competitive on structured signals like price, availability, fit, and policy.
A brand that builds only for the human web is optimising for a shrinking slice of total demand. A brand that prepares for all three compounds its advantage as each era scales.
What Microsoft just shipped
The April announcements are not standalone features. They form a coordinated stack for the three eras.
**Microsoft Clarity AI Visibility** gives brands insight into which pages are cited in AI answers, how often, and where competitors are appearing instead. A Digitas case study highlighted the practical use. Working with a health client, the largest source of demand was not branded queries but exploratory wellness searches in AI surfaces. Without the visibility data, that demand would have stayed invisible.
**Universal Commerce Protocol and Copilot Checkout** are the infrastructure layer for agentic commerce. UCP ready feeds are now available in Microsoft Merchant Center for US advertisers. The Shopify Catalog integration brings millions of merchants into Copilot with real time product data. Target is enabling loyalty account linking for Circle members. The pattern is clear. Microsoft is building the rails that let products show up correctly when an AI agent is doing the shopping.
**Brand Agents** have expanded from Shopify to WooCommerce. New support for brand and policy materials means advertisers can pass guardrails into the agent itself. Microsoft is reporting average 2x conversion lift compared to unassisted sessions. That is a big number from a still maturing format.
**AI Max for Search** enters open pilot in May. It bundles expanded query matching, asset personalisation, and smart URL routing across Copilot Search and Copilot Answers. The transparency layer matters here. Term exclusions, brand inclusions, and messaging constraints are built in so advertisers can keep guardrails on the AI output.
**Offer Highlights in Copilot** shows differentiators on product detail pages across Copilot, Edge, and Bing. Best Buy is among the early activators. The format pulls structured data straight from Microsoft Merchant Center, which means the work happens at the feed level rather than in ad copy.
**Audience Generation** is in closed pilot in the US and Canada. Natural language prompts translate into targeting settings, recommended audiences, and custom signals. The promise is faster audience build cycles without manual assembly.
What advertisers should be doing now
Three priorities for media plans that need to span the three eras.
First, fix the foundation. Structured product data, accurate feeds, and clean policy documentation are the inputs that determine whether your brand is selectable inside an AI agent's flow. If feed quality has been a back office task, move it forward. Specialist Microsoft Advertising support is most useful here when it connects feed health to actual ad performance rather than treating them separately.
Second, instrument for AI visibility. Clarity's new AI Visibility tools are free to use. If your brand is not measuring how often it appears in AI Answers, a competitor probably already is. The metric to track is share of citation, not just share of impression.
Third, test the early agentic formats before they scale. Brand Agents, Offer Highlights, and AI Max for Search are all in pilot or open beta. The cost of testing now is low. The cost of being late once these formats are at scale is the same compounding disadvantage that hit brands which delayed Performance Max adoption. Activation coupons through Microsoft Advertising make it cheaper to spin up the test accounts needed to evaluate new formats without disrupting existing campaign learning.
The broader shift is uncomfortable but clear. Measurement that stops at the click misses where the choice is being made. Brands that build for agents now will have compounding advantages as volume scales.
This insight is based on content originally published on the Microsoft Advertising Blog, rewritten with added context and perspective by Scepter Digital.
