44% of Retailers Advocate for Agentic Commerce. Only 24% Are Experimenting
By Malin Gaertig · September 14, 2026
Mastercard research finds a gap between agentic commerce enthusiasm and action among Australian merchants. As discoverability becomes infrastructure, the competitive question is shifting to why AI models select one brand over another.
- —Forty-four per cent of surveyed Australian merchants advocate for agentic commerce, but only twenty-four per cent are actively experimenting with it.
- —Four in ten merchants have no defined agentic commerce strategy.
- —Machine-readable catalogues, payment rails and agent-ready checkout make products discoverable, but they do not guarantee selection.
- —Brands need to measure which competitors AI models recommend, which prompts trigger the difference and why those models prefer them.
- —MARK measures brand presence and preference across ChatGPT, Gemini, Claude and DeepSeek to show who appears instead and why.
Open ChatGPT. Ask it to recommend a product in your category. Then ask why it picked those brands.
The second question matters more than the first.
Mastercard's latest agentic commerce research surveyed 575 Australian merchants. Forty-four per cent describe themselves as advocates for agentic commerce. Twenty-four per cent say they're actively experimenting with agentic commerce concepts. Four in ten have no defined strategy at all.
Why are discoverability and selection different in agentic commerce?
Much of the conversation around agentic commerce still treats discoverability and selection as the same thing. They aren't.
Discoverability is gradually becoming infrastructure. Shopify is standardising product catalogues into formats AI agents can read. Payment networks are building agent-ready transaction rails. Product feeds, checkout compatibility and machine-readable data are increasingly becoming table stakes. The technical ability for an AI agent to find your products is becoming easier.
Selection is different because it's the moment a model decides which retailer, product or brand deserves the recommendation. That's where competition moves.
What does Mastercard's agentic commerce readiness framework measure?
Mastercard's readiness framework focuses on the foundations: product data, payment capability, analytics and fraud controls. All are sensible and practical recommendations.
But it also highlights a question retailers will eventually need to answer for themselves: if an AI model recommends a competitor instead of you, how would you know? Which competitor appeared? Which model preferred them? What reason did the model give? Which buying prompts triggered the difference?
Can an agent-ready retailer still lose the recommendation?
A retailer can have a machine-readable catalogue, an agent-ready checkout and a technically sound implementation, yet still lose the recommendation because the model preferred somebody else.
That's the distinction organisations will increasingly need to understand as agentic commerce matures. Technical readiness gets a brand considered. Measuring model preference reveals whether it is actually being chosen — and why.
How does MARK measure AI model preference?
MARK shows brands how they appear across ChatGPT, Gemini, Claude and DeepSeek, who appears instead of them and why. That turns an invisible recommendation gap into evidence leaders can use across strategy, measurement, content, commerce and product data.