A shorter path to purchase: What agentic commerce means for the future of ecommerce

Finding a lightweight jacket for an outdoor wedding in Miami shouldn't require eight filters and a dozen open tabs. But that's how ecommerce has traditionally worked: customers adapt their questions to the structure of a website, narrowing a catalog until they hopefully land on the right product.

Agentic commerce offers a different approach. A customer can simply describe the occasion, climate, style, or other considerations, and an AI agent can interpret the request and guide the journey from there.

In the latest episode of Unlocking Gemini, TTEC Digital's Caleb Johnson talks with Mike Mittelman, Gemini Enterprise for CX Retail Lead at Google, about what this shift means for retailers and the customers they serve.

Ecommerce is becoming a conversation

Retail chatbots aren't new. Early versions handled FAQs and simple requests through keywords and predetermined flows. Ask the wrong question, misspell a word, or venture outside the expected path, and the experience often broke down.

Generative AI removes many of those constraints.

Google Cloud Shopping Agent can interpret natural language, maintain context across a conversation, and search large product catalogs without requiring retailers to anticipate every possible path a customer might take.

Consider a home improvement retailer. A customer may arrive with a project in mind rather than a shopping list. There could be millions of product combinations that fit the job. Mapping each scenario in advance isn't practical, but an agent can work from the customer's description to identify relevant options.

The same principle applies to something as simple as buying an outfit. Rather than knowing which fabric performs well in Miami heat, a customer can provide the context and receive recommendations suited to the occasion.

That moves ecommerce closer to the kind of guidance a knowledgeable associate can provide in a store.

Context creates a more useful shopping experience

The conversation doesn't have to stop with text.

Shopping agents can draw on capabilities across Google's technology ecosystem, including translation and generative media, to make digital commerce more responsive to the person on the other side of the screen.

Language is one example. A retailer serving multiple markets can support conversations across languages without creating an entirely separate agent for each one. AI can also help localize product pages and imagery for different audiences.

Generative media adds another dimension. A customer could upload a photo and virtually try on clothes before making a purchasing decision. Mittelman describes how capabilities such as Nano Banana can generate those images today, with video opening the door to even richer experiences.

These aren't simply novel features. They address a persistent challenge in ecommerce: giving customers enough confidence to make a purchase without seeing, touching, or trying on the product in person.

Conversation gives retailers a new kind of signal

There is another valuable advantage to an agent-led journey: the conversational data itself.

Retailers already collect enormous amounts of behavioral data. Search terms, page views, clicks, carts, and conversions show what customers do. A conversation adds context that can be difficult to infer from those actions alone.

Someone searching for “tan blazer” produces a keyword. Someone explaining that he needs something lightweight for an outdoor wedding in Miami provides the retailer with considerably more information about the purchase.

At scale, those interactions can become a source of business intelligence.

Mittelman describes using CX Insights and BigQuery to analyze conversational data for recurring topics, product sentiment, and emerging trends. Retailers can then consider those findings alongside metrics such as average order value, conversion rate, and revenue per visitor.

That could help identify growing interest in a category, uncover demand the current assortment doesn't meet, or show how customers are responding to a new product launch. It gives merchandising, marketing, and CX teams another way to understand demand as it develops.

The value shows up at checkout

A more intuitive journey matters to customers. For retailers, it also must make economic sense.

Mittelman points to Google's early work with Macy's, where the consumer profile most likely to interact with an AI agent was associated with spending increases up to 400%. He also describes ROI models with customers ranging from approximately 3x to 10x, depending on the use case and implementation.

Those results won't translate identically to every retailer, but they demonstrate why the economics of agentic commerce are getting attention. Improving conversion or average order value even modestly can have an outsized effect at retail scale.

The implementation timeline can also make experimentation practical.

Shopping Agent connects with systems retailers already rely on, including product catalogs and add-to-cart functionality. The workflow also includes defining priority customer journeys and determining how the agent should fit within the existing website or mobile app.

Once those prerequisites are in place, Mittelman says Google and partners such as TTEC Digital can move through an intensive six-week build, followed by a controlled rollout that gradually increases traffic. In many cases, he estimates the overall path to full production at roughly two to three months.

That gives retailers an opportunity to test, measure, and refine agentic commerce without waiting for a wholesale ecommerce transformation.

The shift is ultimately less about replacing the search bar than changing how much work happens behind it. Customers bring the question, goal, or occasion; the retailer brings its catalog, data, and expertise — and the agent connects the two.