The intent gap is the growing mismatch between what a shopper already knows when they reach checkout and the experience the checkout actually gives them. The term comes from Rokt, the New York-based e-commerce technology company, which argues that artificial intelligence is widening that mismatch faster than most retailers realize. As AI tools absorb more of the research and comparison that once happened on retail sites, shoppers arrive at the same cart, payment, and confirmation pages in very different states of readiness, and a single generic checkout treats all of them alike.
Consider two buyers landing on the same product page. One spent twenty minutes comparing options inside ChatGPT and arrives all but decided. The other is a repeat customer who typed the address from memory and wants to be recognized, not re-sold. Rokt’s contention, set out in a mid-2026 analysis it calls the intent gap, is that these two need different things at the precise moment most retailers give everyone the same thing.
A concept built on a shift in where shopping starts
Rokt frames the intent gap as the next stage of a change it has tracked for years: the rising value of what it calls the Transaction Moment, the stretch from product selection to order confirmation when a shopper’s attention, trust, and intent all peak. Reporting on the company’s 2026 e-commerce outlook, The Silicon Review described Rokt as the firm that has done the most to articulate and act on the idea that the purchase itself, long treated as a formality, is where advantage now sits. Rokt’s position is that discovery and comparison keep migrating into AI surfaces while the transaction stays on the retailer’s pages, so the checkout is quietly carrying more of the journey than it was ever built to hold.
For most of the past decade, retailers treated the checkout as plumbing, a place to process a payment and little else. Rokt’s reading turns that on its head. If the hardest parts of the decision now happen elsewhere, often inside an AI tool the retailer never sees, then the checkout becomes the first and sometimes the only environment the brand still fully controls once the shopper is ready to act.
The evidence that arrival changes behavior
The clearest outside support for Rokt’s thesis comes from Forrester. In a June 2026 report titled “Death Of The Retail Website As We Know It,” built on a February 2026 consumer panel across the United States, United Kingdom, and Canada, Forrester found that intent increasingly dictates where a shopping journey begins. Answer engines draw the high-consideration research and comparison, brand and retailer sites capture loyal and high-value buyers, and marketplaces such as Amazon absorb routine repeat purchases.
The behavioral data points the same way. According to Shopify’s first-quarter 2026 figures, AI-referred sessions grew eightfold year over year, and those visitors converted at nearly 50 percent higher rates than organic search while spending 14 percent more per order. Adobe reported AI-driven retail traffic up 393 percent over the same period, converting 42 percent more often than non-AI sources, with 37 percent higher revenue per visit. Traffic from social platforms, by contrast, tends to abandon at higher rates. The numbers move by retailer and category, but the direction holds: how a shopper arrives already shapes whether they buy, and by how much.
For retailers, the practical implication is uncomfortable. The same checkout can look like a strong performer against one source and a leaking funnel against another, and a single blended conversion rate hides the difference.
Why personalization tends to stop at checkout
Adapting a site to its traffic source is ordinary marketing practice. Retailers routinely change headlines, offers, and calls to action on landing pages depending on where a visitor came from. Rokt’s observation is that this discipline usually fades once the shopper moves past the product page. A retailer might run a dozen landing-page variants by source and then hand every one of those visitors the same cart, the same payment screen, and the same confirmation page. The context that shaped the first impression rarely follows the shopper into the moment the money moves. If referral context improves relevance at the top of the funnel, Rokt asks, why should it go quiet at the bottom?
Rokt’s argument: arrival context is a relevance signal
Rokt’s answer is that arrival context, meaning what a shopper has already done before they land, is itself a signal worth acting on. The company laid out a related case in its whitepaper “The New Economics of Checkout,” arguing that as comparison shifts into AI, relevance at the point of purchase becomes the differentiator, including the discipline to show nothing when nothing would help. Craig Galvin, Rokt’s chief revenue officer, framed the underlying logic when the company was named in Gartner’s 2026 market guide for retail and commerce media networks, calling the transaction “one of the most valuable and underleveraged” interactions in commerce, a point when a shopper’s intent is clear, and attention is close.
That restraint sits at the center of how Rokt describes its technology. In its account of the company’s approach, TechAnnouncer noted that Rokt’s engine will withhold an offer entirely when it cannot clear a quality threshold, and Rokt’s chief AI officer, Claire Southey, has said the smartest move is often to step aside and let the customer finish the purchase. The willingness to show nothing is unusual in an industry that has long measured checkout real estate by how much of it can be filled. The intent gap extends that thinking. What a shopper already did before arriving, Rokt argues, is as revealing for relevance as who the shopper is.
How readiness changes what counts as relevant
Rokt maps the idea across a few broad arrival profiles, and stresses that these are testable hypotheses rather than fixed rules. A research-heavy shopper coming from an answer engine has usually decided already. Reviews that confirm the choice help, while new recommendations that reopen the decision tend to backfire. At payment, an offer tied to the purchase can fit where a generic cross-sell breaks the moment, and at confirmation, the effort already spent leaves the shopper open to loyalty content and education.
A brand-loyal shopper arriving directly carries a relationship. History-based suggestions feel useful rather than intrusive, a familiar and recognized checkout reinforces the tie, and the confirmation page becomes a natural place for referrals or exclusive access. A convenience-driven shopper coming off a marketplace wants the shortest path at the right price. Reassurance on speed and cost lands; express checkout or cashback can help, and almost anything else risks the sale unless it plainly saves money or time. The through-line is straightforward: what a shopper already accomplished before the transaction is a strong clue to what will help during it, up to and including the choice to add nothing at all.
What Rokt is telling retailers to do
Rokt’s guidance leans diagnostic before it turns to technology. The near-term steps it recommends:
- Break out checkout performance by arrival source instead of tracking one blended conversion rate.
- For each major source, find the transaction stage with the sharpest drop-off, which turns a vague conversion problem into a specific one.
- Test whether the upstream signals that already lift conversion, such as referral source and campaign context, can do the same at the transaction, starting with the highest-volume sources.
- Walk the checkout from each arrival context and ask whether it makes sense for someone who compared five products, someone on a third purchase, and someone measuring the brand against one-click rivals.
- Put relevance infrastructure at the point of purchase, since static checkouts respond poorly to different arrival contexts.
A company positioned to make the case
Rokt has spent more than a decade building for the point of purchase, which is part of why its framing carries weight with analysts and partners. Its AI engine, the Rokt Brain, weighs each transaction in real time against more than 1.95 trillion data points a year to decide the next best action, and per Rokt by the Numbers, the network will power more than 10 billion transactions in 2026, reach 165 million monthly active users, and serve over 33,000 clients, including more than half of the largest global e-commerce companies. Rokt Ads posts a 4.03 percent click-through rate, which the company reports is roughly ten times that of Google Display.
The outside record has kept pace. Rokt’s revenue grew to more than $800 million in 2025, according to the Gartner announcement, and the firm landed on the 2025 Deloitte Technology Fast 500, as Dataconomy noted in its analysis of Rokt’s renewed partnership with Fanatics. That deal, covered across trade press including Sporting Goods Intelligence, placed Rokt’s relevance engine across Fanatics.com and league sites worldwide. The Gartner inclusion arrived as commerce media accelerated, with US retail media spend projected to reach $69.33 billion in 2026, up from $58.79 billion in 2025 on eMarketer figures cited in the guide announcement. Rokt continues to publish its own view of the shift through its blog and its LinkedIn page.
Where this goes next
Rokt calls the intent-gap idea early, and says the value of arrival-context signals still has to be proven retailer by retailer. What is less contested is the setup. AI is concentrating value at the transaction, at the same time intent is scattering the places shoppers begin, and those two forces meet on the same cart and confirmation pages. As Elizabeth Buchanan, Rokt’s chief commercial officer, argued in the company’s 2026 outlook, the brands most likely to win this year are the ones “doing less” at the transaction while making what remains far more relevant. If Rokt has read the shift correctly, the retailers that close the intent gap first will be the ones that stop treating every arrival as the same shopper.
