The headline numbers buried in ecommerce analytics platforms in early 2026 tell a story most DTC founders have not read yet. LLM-referred traffic converts at 18%. Higher than paid search. Higher than email. Higher than any acquisition channel most brands are actively spending on. And the traffic is growing 40% quarter on quarter, at zero ad spend.
The shift is not about chatbots on your site. It is about AI agents that shop on behalf of consumers. Someone opens ChatGPT and says "find me the best magnesium supplement for sleep under thirty pounds and add it to my cart." The agent does the research, makes the selection, and routes the customer to a checkout. The brand that wins that transaction did not pay for a click. They just had better product data than the competition.
What Agentic Commerce Actually Is
ChatGPT launched Instant Checkout in September 2025, now reaching 900 million weekly users. Google announced its own agentic commerce protocol in January 2026, with Walmart, Target, Shopify, and over 20 retail partners backing it. McKinsey projects the global agentic commerce opportunity at three to five trillion dollars by 2030. Salesforce estimates AI platforms will account for $20.9 billion in retail spending in 2026 alone.
This is not a future thing. The infrastructure is live. The volume is small today but the growth rate is extraordinary. LLM-referred conversion volume grew 1,200% in the second half of 2025 according to multiple retail analytics providers. Brands that build for it now will hold a structural advantage when the volume scales.
The consumer does not browse. They describe what they want to an agent, and the agent selects for them. Your product either passes that selection process or it does not appear.
The Conversion Data Every DTC Brand Should See
From a study tracking 329 DTC brands over 13 months, LLM referrals convert at approximately 18%, making them the highest-converting traffic source in ecommerce. For context: Google Ads converts at 1.82% and Meta at 0.52%. Email, typically the strongest owned channel, sits in the 2 to 4% range for most brands. LLM referrals outperform all of them, unaided.
On an unaided basis, LLM visitors convert at 2.47%. When those same visitors encounter an on-site AI shopping assistant, conversion jumps to 9.84%, a 4x lift from a single interaction. Perplexity traffic drives 57% higher average order value than other sources despite representing a small share of volume. ChatGPT dominates volume share but dropped from 86.7% to 64.5% market share in one year as Gemini climbed from 5.7% to 21.5%.
What this tells an operator: the LLM ecosystem is fragmenting fast. The brands that win will be visible across all the major agents, not optimised for one.
Why AI Agents Convert So Well
When a consumer arrives from an LLM referral, they are not browsing. They have already had a conversation with the agent. They described what they want, the agent did the comparison shopping, and now they are at your product page with intent fully formed. The pre-qualification work happened upstream, before the click.
It is the same reason a warm referral from a trusted friend converts better than a cold ad impression. The trust and decision work were done before they arrived. Your job at that point is not to persuade. Your job is to not give them a reason to leave.
What AI Agents Actually Look For
AI shopping agents make product selections based on machine-readable signals, not human judgement. The criteria are systematic and consistent across the major platforms.
Product data completeness
Agents cross-reference product titles, descriptions, specifications, ingredients, and use cases against the consumer's query. A vague description loses to a specific one. If your product page says 'natural supplement' and your competitor says '300mg magnesium glycinate, clinically studied for sleep onset and cortisol reduction', the agent picks the competitor. Specificity is not just good copy. It is a ranking signal.
Structured data and schema markup
Agents pull from structured product schemas including Schema.org Product markup, Open Graph tags, and price and availability signals. If your Shopify theme does not output clean structured data, you are invisible to the most principled agents. A broken or missing Schema.org Product type is equivalent to having no listing at all from the agent's perspective.
Reviews and social proof
AI agents weight review volume and sentiment. They can parse star ratings, verified purchase signals, and review content. A product with 400 reviews mentioning specific outcomes converts better in agent recommendations than a product with 12 generic reviews. Schema-marked-up review aggregation makes your ratings machine-readable rather than relying on the agent to scrape your page.
Checkout reliability and site speed
Once an agent selects a product, it routes a high-intent customer to your checkout. If your checkout loads slowly, errors under load, or loses cart state, that transaction is gone. Agents also factor in checkout reliability signals over time, meaning repeated failures can affect recommendation rates.
How to Build for Agentic Commerce
None of this requires a new platform or a new budget. It requires making existing infrastructure machine-readable.
Audit your product data from a machine's perspective
Run your top 10 SKUs through Google's Rich Results Test. Check that every product has a complete Schema.org Product type with name, description, brand, offers (price, availability, currency), and review aggregation markup. Fix the gaps before anything else. This audit takes two hours and the results are often eye-opening.
Enrich product descriptions with specification-level detail
Rewrite your product descriptions to answer the questions an AI agent would ask on behalf of a consumer. Ingredient names and amounts. Use cases and specific outcomes. Comparison to alternatives. Well-structured copy with clear headings and short paragraphs is more parseable than a wall of marketing text. Write for two audiences at once: the human who lands on the page and the agent that decided to send them there.
Clean your Shopify catalog and product feed
Product title format, variant naming, price consistency, inventory accuracy, and image quality all feed into how confidently an agent can recommend your SKU. Remove duplicate listings, fix broken variants, and ensure your Google Shopping feed and your product pages carry consistent data. Inconsistencies across sources reduce agent confidence in your listings.
Instrument your analytics for LLM traffic
Most brands have no idea how much traffic they receive from LLM sources because standard UTM attribution does not capture it. Referrer strings from ChatGPT, Claude, Perplexity, and Gemini are distinct and parseable. In Shopify Analytics or Google Analytics 4, create a channel group that tags chat.openai.com, claude.ai, perplexity.ai, and gemini.google.com as a single LLM referral channel. You cannot optimise what you are not measuring.
What This Looks Like in Practice
A supplement brand we worked with found, on auditing their referrer logs, that they were already receiving 180 to 220 sessions a month from LLM sources with a conversion rate of 14%. They had no idea this traffic existed. Their product pages had weak schema markup, generic descriptions, and their Google Shopping feed had 40% of SKUs with incomplete variant data.
After a four-week enrichment sprint, LLM traffic doubled. Conversion held above 15%. The total incremental revenue from the channel, at zero additional ad spend, was around four thousand pounds in the first 60 days. Small in absolute terms today. Growing 40% every quarter.
The brands that notice this early and build the infrastructure now will not have to compete for this traffic later. They will already own it.
Inside the system
How we build this for brands
For brands we work with, agentic commerce readiness is built into the data infrastructure from day one. That means enriching product catalog data and schema markup to pass machine-readable quality standards, wiring up referrer tracking to isolate LLM traffic as a distinct channel, and running a VOC engine that distils real customer language into product descriptions with the specificity agents favour. The profit dashboards we build include LLM traffic as a separate acquisition source, tracked weekly so the compounding becomes visible before it becomes material.
On the retention side, lifecycle flows built in Klaviyo compound the advantage: brands with strong repeat purchase rates and review volume signal health to agents and score higher in recommendation algorithms. It is the same flywheel in a new context. Part of this runs live for portfolio brands today. The full system is what we deploy when we take a brand on.
Agentic Commerce Audit
Find Out How Visible Your Brand Is to AI Shopping Agents
I will audit your product data, schema markup, and catalog hygiene against the signals AI agents use to make purchase decisions, and give you a prioritised fix list.
Book Your AuditFrequently asked questions
What is agentic commerce and how does it affect DTC brands?
Agentic commerce refers to AI agents that make purchasing decisions on behalf of consumers. A consumer tells ChatGPT or a similar AI what they want to buy, and the agent selects a product, compares options, and routes the consumer to a checkout. For DTC brands, this creates an acquisition channel that converts at approximately 18%, higher than paid social or paid search, and is growing 40% quarter on quarter.
How do AI shopping agents decide which products to recommend?
AI agents make selections based on machine-readable signals: product data completeness, structured Schema.org Product markup, review volume and sentiment, price and availability accuracy, and checkout reliability. Brands with incomplete product data, missing schema markup, or thin reviews are invisible to the most principled agents regardless of how strong their paid media performance is.
What is LLM traffic and how do I track it in Shopify?
LLM traffic is visits referred from AI interfaces such as ChatGPT, Claude, Perplexity, and Gemini. In Shopify Analytics, filter your referral traffic by source and look for chat.openai.com, claude.ai, perplexity.ai, and gemini.google.com. In Google Analytics 4, create a channel group rule for these referrer domains to track them as a distinct channel over time.
Is agentic commerce replacing paid social for DTC brands?
Not replacing. Complementing. LLM traffic currently represents less than 2% of referral traffic for most DTC brands, so it is not a primary volume channel today. But with a conversion rate of 18% and growth of 40% QoQ, it will compound into a significant source over the next 12 to 24 months. Paid social drives volume. Agentic commerce drives intent.
How long does it take to see results from agentic commerce optimisation?
Structured data changes can be indexed by AI systems within a few weeks. Product description enrichment takes 4 to 8 weeks to show meaningful impact on LLM recommendation rates. The quickest win is instrumenting your analytics to measure what you already have, because many brands discover they are already receiving LLM traffic they were not tracking.
About the author
Caner Veli built Liquiproof to global distribution across 3,000+ retailers, then exited. He now runs Purposeful Profits using a combination of operator strategy and AI-powered systems he has built and uses daily, having 10x'd monthly revenue in his own business in the last 90 days.