80% of ChatGPT Prompts Don't Trigger the Shopping Feature
Key Takeaways:
-
A new Profound study found that ChatGPT’s shopping feature appeared on just 8.82% of prompt submissions.
-
79.43% of tracked prompts never triggered shopping across the full study window, while only about 6% fell into the reliable 80%+ trigger bucket.
-
Open-ended prompts triggered shopping at 12.1%, compared with just 3.1% for brand-direct prompts.
-
If a prompt triggered shopping on one run, Profound found there was an ~83% chance it would trigger again the next day.
TL;DR: New research from Profound suggests ChatGPT Shopping is a much narrower surface than many brands may assume. Across roughly 2 million unique prompts tracked over several months, shopping showed up less than 10% of the time, and most prompts never triggered it at all.
ChatGPT Shopping has become one of the most talked-about new surfaces in AI-driven product discovery. OpenAI says shopping-enabled responses can show product options with imagery, product details, and links to merchant sites when a prompt suggests shopping intent. OpenAI’s overview of Shopping with ChatGPT Search explains how that product surface works.
But Profound’s new analysis suggests that surface appears far less often than the hype around it might imply. In the company’s data, most prompts never entered the shopping layer at all.
| Finding | Result |
| Overall shopping trigger rate | 8.82% |
| Open-ended prompt trigger rate | 12.1% |
| Brand-direct prompt trigger rate | 3.1% |
| Prompts that never triggered shopping | 79.43% |
| Reliable prompts (80%+ consistency) | ~6% |
| Next-day repeat chance after a trigger | ~83% |
When is ChatGPT Shopping triggered?
The main takeaway from Profound’s study is straightforward: ChatGPT Shopping is real, but it’s far less universal than many brands might expect. Here are the biggest findings.
1. Shopping triggered less than 10% of the time
Across millions of prompt submissions, Profound found that the shopping feature was activated on about 8.82% of them. That makes it a meaningful surface, but not one that brands can count on appearing by default. That kind of unpredictability is not unique to ChatGPT Shopping either. We are seeing similar shifts in Google, where AI Overview citations are pulling from fewer top-ranking pages.
2. Open-ended prompts performed much better than brand-direct ones
Profound found that open-ended prompts triggered shopping at 12.1%, versus 3.1% for brand-direct prompts. Prompts framed around a need or use case were far more likely to activate shopping than prompts built around a specific brand name, which is exactly why improving visibility in ChatGPT search and shopping results is becoming more of a practical challenge for brands.
3. Most prompts never triggered shopping at all
The consistency data was one of the more striking parts of the study. Profound found that 79.43% of prompts never triggered shopping across any run during the study window, while only 0.72% triggered every single time.
4. Prompts that trigger tend to keep triggering
When a prompt triggered shopping on a given run, there was an ~83% chance it would trigger again the next day. Prompts that didn’t trigger were also very unlikely to suddenly start, which is why the pattern looks so all-or-nothing across the dataset. That kind of uneven behavior is not limited to shopping results. It also appears in broader citation patterns, where self-promotional listicles are still getting cited in AI search.
What does this mean for brands trying to show up in ChatGPT Shopping?
Profound didn’t just measure how often shopping appears. The study also pointed to a few practical patterns worth paying attention to as ChatGPT Shopping keeps evolving.
1. Classify before you optimize: Profound’s recommendation is to pull the prompts customers would actually type, then test them repeatedly over time. Since most prompts never enter the shopping layer, brands need to know which ones even have a shot before investing in optimization.
2. Branded queries matter less than many brands think: Brand-direct prompts triggered shopping at just 3.1%, compared with 12.1% for open-ended prompts. The exception is when a branded query still clearly describes a purchasable product need rather than just a brand name.
3. Short-term persistence is real, but it decays: If a prompt triggers today, there’s a strong chance it triggers again tomorrow. But Profound also flags that model updates can reset visibility quickly, which makes ongoing monitoring more important than a one-time audit.
4. Align your ChatGPT strategy with how shopping prompts actually work:
- Focus on discovery-stage language: Profound’s data makes the strongest case for prompts that describe a need before a user has settled on a brand.
- Prioritize shippable product intent: Prompts that describe specific, purchasable product needs were much more likely to enter the shopping layer.
- Track consistency, not just visibility: A one-time trigger isn’t enough. Brands need to know which prompts show stable shopping behavior over time.
Ready to win more visibility in AI search?
ZeroClick Labs helps brands track, understand, and improve how they appear in AI search.
From Google AI Overviews to ChatGPT and other answer engines, we focus on the practical work of visibility: where your brand shows up, where it does not, and what to do next. With 15+ years in digital marketing and search analytics, a dedicated team, and experience working with 200+ clients, ZeroClick Labs brings proven search expertise to a fast-changing AI landscape.
“Our agency had no idea how to approach AI visibility. ZeroClick only does this one thing so they actually know what works. Worth every penny just to not waste time figuring it out ourselves.” – Jay