Shopify: AI search tripled store traffic and orders, and did not replace Google

Something flipped in how buyers find your product. For years the fear was that AI answers would eat your traffic the way they ate publishers. Shopify just put numbers on the table that say the opposite, at least for stores. Let me catch you up on what changed and what to bring to your next review.

The deep cut

• AI reads intent, not keywords. Shopify's example: an agent finds a car seat that fits three across a sedan, not the top-ranked "car seat" page.
• Retrieval runs on SEO, not vibes. GEO copy sits in a folder nobody finds if the page isn't crawled or linked, per NysaWords.
• Track citations apart from rankings. Rank tools show your SEO base works; only citation tracking shows if AI names your brand.

The channel that grew instead of shrank

On its Q2 earnings call, Shopify said AI-driven traffic and orders tripled year over year, and it credited part of a revenue beat to it. Revenue rose 36% to $3.6 billion, past Wall Street's $3.4 billion.

Here is the part worth pinning up. President Harley Finkelstein called AI "a complement to search, rather than a substitute for it." Traditional search sessions are up 1.3x over two years and still hold about a third of all storefront sessions. This is the reverse of what happened to publishers, where AI summaries cut click-through and ad revenue. For stores, AI added a lane instead of closing one.

Why the match got better

Finkelstein's framing is the useful bit. Keyword search ranks by popularity against a handful of words. An AI agent makes many calls into the catalog, working with richer structured data to match a product to the shopper's actual need. His example: a buyer asks for the best car seat that fits three across a sedan. Old search grabs "car seat." An agent reads the dimensions, the vehicle type, and the count of three, then searches all of it at once.

That precision shows up in behavior. Half of AI-referred sessions land directly on a product page, 2.5 times more than traditional search. And 75% of AI-attributed purchases happened outside Shopify's top 100 categories. The long tail, the smaller merchants, got the lift.

The plumbing under the answer

Before you shift budget toward AI-facing copy, know what the answer actually reads. Every AI engine retrieves documents from an index, then a model writes an answer citing what it trusts. As NysaWords lays out, that retrieval step runs on plain SEO: crawlable pages, clean structure, consistent entity signals.

Skip that and the polish is wasted. An orphaned page with no internal links gets cited by nobody, no matter how sharp the writing. A six-second load time loses the retrieval race before content quality matters. There is no clever AI copy that fixes a slow page or a broken schema.

Where the work should live

The order matters more than the labels. First confirm your top pages are indexed, linked, and fast. Then make your product name, category, and description match across your site, your schema, and your third-party listings. Only then rewrite pages to answer narrow questions with dated, specific data.

Speed helps here too. SEO teams still lose hours moving keyword data between tools before a writer can start. Hoe shi Lee describes using Anthropic's Model Context Protocol so an agent pulls keyword data already grouped by intent, turning research from a gate into a direct input. That is the same intent-first shift Shopify is seeing on the buyer side.

Three questions for your team

• Are our highest-value product pages actually indexed, internally linked, and fast, or are we writing AI copy on a broken foundation?
• Do our product name, category, and description match across our site, schema, and third-party listings, or are we sending confused signals to retrieval?
• Are we tracking brand citations in AI answers on a schedule, separate from keyword rankings, so we know which lever is working?