What zero-click actually names
The term is borrowed from zero-click search, where a query is answered in the results page and the site that supplied the answer never gets visited. Commerce inherited it during NRF season and it has been travelling ever since, usually attached to a large forecast.
Underneath the framing there is a real, observable compression of the shopping journey. It used to be: search, browse several stores, compare pages, read reviews, decide, checkout. Increasingly it is closer to three steps — ask, receive a shortlist, act. The middle of that funnel, the part merchants spent fifteen years optimizing with photography, layout and on-site merchandising, is where an agent now does the work instead, and it does that work by reading rather than looking.
Take the numbers with the appropriate salt
Almost every figure circulating on this topic comes from a vendor with something to sell, ourselves included, so treat the direction as informative and the magnitude as marketing. Consumer surveys report high rates of AI use somewhere in the shopping journey — product ideas, review summaries, price comparison — while the share who report actually completing a purchase on an AI referral remains small. Gartner's projection of AI agents intermediating trillions in B2B purchasing by 2028 is widely quoted and, like all multi-year projections, is a scenario rather than a measurement.
What we can say from our own telemetry rather than a survey: when an agent does arrive at a machine-readable catalog, it behaves exactly as the zero-click thesis predicts. It never renders a page. It queries structure, filters, shortlists. We published a week of it. Nobody clicked anything.
The reframe that makes it actionable
"We are losing clicks" produces bad strategy, because it points at recovering traffic that is not coming back. The mechanism points somewhere more useful.
- The decision moved, the criteria changed. Persuasion does not travel through an API. Attribute completeness, price accuracy, stock certainty, variant clarity and returns terms do. What convinces an agent is not what convinced a browser.
- Absence is silent. A store that cannot be read is not ranked lower. It is not in the set at all, and no analytics view will report the impressions you did not get. This is the difference between losing a competition and never entering one.
- Unverifiable equals excluded. An agent that cannot check a claim against an authoritative source discounts it. Scraped or copied data is precisely the kind that fails that test — which is why serving your own data from your own domain is a trust position, not just a technical one.
- The brand relationship survives, elsewhere. Discovery may be intermediated while fulfilment, returns, support and repeat purchase remain yours. Zero-click threatens the top of the funnel far more than it threatens the customer relationship — provided you are still in the funnel.
What to actually do about it
Nothing exotic, and none of it requires waiting for a protocol to settle. Publish a discovery document so an agent can learn what your store exposes. Serve structured, bounded, read-only catalog search from your own domain, answering live rather than from a snapshot. Make price and stock honest, including the cases where the honest answer is "depends on the variant" — a system that declares what it does not know is more trustworthy than one that guesses confidently. Then fix the boring data gaps, because completeness is now a ranking factor for a reader that cannot infer from a photograph.
That is the whole of KaliCart Bridge, and you can check where a store stands today with ready.kalicart.com. The federated layer goes one step further, answering the cross-store question — which merchant has what the shopper described — with consent and provenance attached, since that is the question a shortlist is actually made of.
The honest limit
Being readable does not make you chosen. It makes you eligible. Agents will still weigh price, availability, delivery and reputation, and a legible catalog full of weak offers will lose to a legible catalog full of strong ones. What readability buys is the right to compete on the merits — which, when the shopper never visits your page, is the only competition left.
Sources
- 7 AI trends shaping agentic commerce in 2026 — commercetools (consumer-behaviour figures)
- Zero-click commerce and the Gartner B2B projection
- How retail AI agents are reshaping product discovery — Zero Click Project
- Related: From ungooglable to unagentable · The second buyer · WooCommerce for AI shopping agents