The largest promotional products distributor in North America has zero ChatGPT product cards.

Good ideas travel faster when you share them; forward this to someone who’d find it useful.

Over the past five weeks I ran 20 buyer queries through ChatGPT, Perplexity, and Google AI Mode. Same queries, six separate sessions, every result recorded.

One pattern kept showing up that I couldn’t explain away: the biggest, most established companies in the category were declining. Smaller, narrower specialists were gaining. Every session.

I want to walk through what the data actually shows — and what I think it means for anyone running a catalog business right now.

---

The Observation

4imprint is the largest promotional products distributor in North America. Over a billion dollars in revenue. Fifty thousand products. Decades of brand recognition.

ChatGPT query: "laser engraved tumbler BPA free bulk pricing" — Kodiak Wholesale appears as a named recommendation with pricing. Bulk Tumblers appears with a product card. No 4imprint.

Google AI Mode query: “laser engraved tumblers bulk corporate order” — 4imprint appears as a named product recommendation. Kodiak Wholesale also appears. Both platforms, different mechanisms.

That's what prompted me to look more carefully at the broader pattern.

The Pattern

Here’s the company-level trend across my final two sessions — 60 total responses each, 20 queries across 3 platforms:

In my dataset, every specialist I tracked gained visibility across those two sessions. Every broad-catalog distributor declined.

One Hypothesis for Why

I can’t say definitively why this pattern appeared. But here’s the explanation that fits the data best.

When a buyer types “laser engraved tumblers for employee onboarding, 50 units” into ChatGPT, the AI is trying to match that query against product data simultaneously across multiple fields: decoration method, use case, minimum order quantity, production time.

A specialist store built around one category tends to have richer, more complete product-level metadata within that category. Every SKU has those fields populated. Every product description is written for buyers who ask that exact type of question.

A broad-catalog distributor built their product data for a different era — a website where buyers browsed and filtered. For that format, breadth was an advantage.

The 4imprint Situation

I want to be precise here — this isn’t a criticism of 4imprint.

4imprint is not invisible. They dominated Google AI Mode. Perplexity mentions them regularly. Their brand recognition is real and shows up in the data.

The ChatGPT gap is structural — and it illustrates something important about how different AI platforms actually work. Google AI Mode crawls the open web and weights domain authority heavily. A billion-dollar incumbent with decades of SEO history does well in that environment. ChatGPT's product card system operates differently. It requires a structured product data feed — and that feed now powers both organic discovery results and the new paid ads surface OpenAI launched in June 2026. Shopify merchants get feed integration automatically. Many established distributors rely on catalog infrastructure that predates these requirements entirely. The platform has moved away from in-chat checkout and is now explicitly focused on discovery — which makes product data quality more consequential, not less.

Whether 4imprint has submitted that feed or not, I can’t say. What I can say is that the pattern of Shopify-native stores appearing where large distributors don’t is consistent across every session I ran — and it’s consistent with a feed participation explanation.

The Bigger Question

The interesting question here isn’t really about which companies are winning on ChatGPT today.

If that’s true — and the data suggests it might be — then the companies falling behind in AI commerce aren’t falling behind because of marketing. They’re falling behind because their catalog infrastructure was built for a distribution channel that no longer has a monopoly on buyer attention.

That’s a very different problem. And it requires a very different solution.