How a USB-C accessory brand turned its spec sheets into AI-friendly answers — and lifted AI-attributed sessions by 38.4% in thirty days.
Tech shoppers no longer browse — they ask AI to compare. The questions come back in port counts, sustained wattage, host compatibility, monitor support. Generic lifestyle copy doesn't survive that filter. AI assistants only cite a product when its PDP exposes the same facts in machine-readable structure — and most Shopify electronics PDPs hide their specs inside JavaScript tabs, image-only infographics, or unscoped tables that language models can't parse.
Nullport sells USB-C docks, Thunderbolt hubs, and charging accessories — 312 SKUs across Shopify and a headless storefront. Every product page held the facts a tech buyer needed: port count, sustained wattage, host compatibility matrix. The problem was that those facts lived behind a JavaScript-rendered tab system. ChatGPT, Claude, Perplexity, and Gemini couldn't reach them — so when shoppers asked AI for a "100W dock with MagSafe passthrough," Nullport simply wasn't in the answer.
Agentic Page mirrored every Nullport PDP into a machine-readable structure organized around the spec attributes that AI buyers actually compare. Average token count per page dropped 79.7% — the AIs got the answer in fewer tokens, and started surfacing it more often. Within twenty-four hours, the first batch of 286 PDPs had been indexed by ChatGPT and Claude. By day seven, the first live ChatGPT conversation triggered a real-time fetch of a Nullport dock page.
Over the next three weeks, spec-query clusters started lighting up. The "USB-C dock for Mac" family of questions began consistently citing Nullport. By the end of the thirty-day window, AI-attributed sessions ran +38.4% above a matched control group — and the cited-recommendation rate on "100W dock for Mac" hit roughly one in four answers. For the first time, Nullport was reachable through the same surface its buyers were actually using.
The Nullport deployment followed the same arc every Agentic Page project tends to follow — fast indexing, slower behavior change, then a discrete moment when AI assistants start citing the brand in real conversations.
Agentic Page deployed across all 312 Nullport SKUs. Mirror pages restructured around spec attributes: port count, sustained wattage, host compatibility, monitor support.
286 of 312 pages picked up by ChatGPT and Claude indexing systems. Initial knowledge-base ingestion confirms the new structure is machine-readable.
First ChatGPT real-time retrieval detected. A user asks for a 100W USB-C dock with MagSafe passthrough; ChatGPT fetches a Nullport PDP mid-conversation to construct the answer.
Spec-query clusters start surfacing Nullport in cited recommendations. The "USB-C dock for Mac" family becomes the most-cited product cluster, with 41% of ChatGPT cited answers in that subset referencing a Nullport page.
AI-attributed sessions stabilize at +38.4% above the matched control group. Average token cost per PDP down 79.7%. AI Visibility Score reaches 92.1 / 100.