Armed With Purchasing Agents, Digital Wallets Could Turn One-Click Checkout Into One-Query Purchases

Deeply interconnected, search and checkout functions on e-commerce platforms have enabled the discovery of products and services. Fintech-as-a-service has enhanced tight integration by improving the security of transactions and simplifying payments. Together they have set the stage for the next wave of innovation in online shopping. 

Enter Large Language Models (LLMs). The integration of LLMs into digital wallets like Shop Pay, Cash App, and Apple Pay could create “purchasing agents” that transform e-commerce. Functioning as AI agents embedded in digital wallets, purchasing agents should simplify the purchasing journey from one click to one query by combining product discovery and payment processes. Their potential could extend beyond e-commerce to the entire online economy, from travel bookings to ride hailing and restaurant deliveries. 

According to our research, by 2030, AI Agents could drive nearly $9 trillion¹ in global online spending. That would include AI integration at the application level—via an AI assistant helping with purchases on a Shopify store—and AI integration at the operating system level—via a personal AI agent that manages all transactions. In the sensitivity table below, we explore the scope of an AI purchasing agent embedded in a digital wallet, highlighting the revenue opportunity by considering various take rates—the percentage of each transaction the digital wallet could earn for facilitating lead generation through its AI-agent services. At the midpoint of the ranges, if 10% of Global Gross spend by AI Agents occurs in digital wallets and they take a 5% lead generation fee, AI purchasing agents could generate $43 billion for digital wallet providers, adding an incremental 50% to the estimated global revenue base of digital wallet providers like Shop Pay, Cash App, Alipay, and Apple Pay today. 

Source: ARK Investment Management LLC, 2024. This ARK analysis draws on a range of external data sources, as of October 24, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

In this blog, we examine why and how digital wallets are likely to incorporate LLM-based search, revolutionizing lead generation and e-commerce.

Our research suggests that purchasing agents could consolidate the online shopping process—search, aggregation, and checkout—and create a seamless “one-query purchasing” experience. Increasingly, intuitive and sophisticated search tools are enabling consumers to find relevant products more quickly and accurately and businesses to increase their conversion rates. Although these tools have boosted customer satisfaction and fueled the growth of online retail platforms,²  28%³ of US shoppers still report that they are frustrated by irrelevant search results and complicated checkout processes, the latter causing 18% of shoppers to abandon their carts. Requiring shoppers to create accounts leads 26% of them to abandon their carts. 

Digital marketplaces are beginning to prioritize one-click checkouts, as ~80%⁴ of US consumers have stored credit/debit cards and other personal information during previous e-commerce purchases. With LLM-based search and one-query purchasing, AI purchasing agents can understand user intent by leveraging purchase histories and preferences and offering an unprecedented level of personalization and efficiency with little input from consumers.

Disintermediating closed loop data aggregators, AI purchasing agents conduct the entire online shopping process—search, aggregation, and checkout—and facilitate one-query purchasing. In the search step, Large Language Models (LLMs) discern purchasing intent and tailor search results. In the aggregation step, LLMs curate and personalize the results based on user preferences. Finally, in the checkout step, LLMs automate and complete the checkout process. 

Some digital wallets are bringing this process to life today. Integrated with Perplexity.ai, Shop Pay addresses search by encouraging users to explore product recommendations and complete purchases, as illustrated below. 

Source: Hines, 2024.⁵ For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

Klarna’s OpenAI-powered shopping assistant combines search and aggregation with product suggestions based on user preferences, as illustrated below.

Source: Source: ARK Investment Management LLC, 2024. Rendered by Prasanna, based on a conversation with the chatbot in the Klarna App. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

As LLMs evolve and refine personalized product recommendations, price comparisons, and payment choices, consumers are likely to engage more and more with their digital wallets and AI purchasing agents. With one-query purchases, shoppers could simply issue a single command to discover, compare, select, and purchase items in a highly efficient, friction-free experience, as depicted below.

Source: ARK Investment Management LLC, 2024. Image rendered using Adobe Photoshop. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

Today, digital wallets typically kick into gear during the final stage of an e-commerce transaction—checkout—when users view their carts and finalize their purchase by selecting a payment method, as shown below.

Source: Farrelly 2021.⁶ For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

As digital wallets increasingly integrate LLM-enabled purchasing agents, consumers are likely to interact with them much earlier in the e-commerce journey, as depicted below.

Source: ARK Investment Management LLC, 2024. Image rendered using Adobe Illustrator. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

AI-powered purchasing agents are likely to turn one-query purchasing into the new industry standard, creating a large advertising opportunity. Interestingly, “Amazon's Choice” badges on amazon.com serve as an implicit recommender that boosts conversion rates by 25%.⁷ In our view, one-query purchasing will motivate users to adopt and rely increasingly on purchasing agents for similarly curated suggestions, with perhaps even higher conversion rates. That said, to achieve success, digital wallet sponsors will have to guard against inundating consumers with sponsored content and build trust by delighting consumers, one at a time.

Lending perspective to the size of this opportunity, our research suggests that digital wallets helmed by purchasing agents could increase their share of global e-commerce purchase volume from 50% in 2023 to 72% by 2030, as shown below.

Note: The 2018-2023 figures (solid purple line) are historical figures documented in the Worldpay Global Payments Reports; the 2024-2030 figures (dotted purple line) are ARK’s adoption estimates for E-com payment methods, based on data from a range of external data sources, including WorldPay, as of October 24, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Forecasts are inherently limited and cannot be relied upon.

eMarketer estimates that 57%⁸ of consumers will use digital wallets in 2024. If that uptake remains robust, which LLM-power purchasing agents should ensure, digital wallets are likely to capture value from credit and debit cards, which held 22%⁹ and 12%¹⁰ market share, respectively, in 2023. A shift like that could lead overall to faster, more efficient transactions and ever-increasing adoption of digital payment solutions.

If digital wallets integrate LLM-backed search as we expect, search is likely to commoditize. ARK’s research suggests that AI purchasing agents are poised to disrupt the traditional dominance of walled gardens like Google and Amazon. By automating tasks like product comparisons and purchase decisions, those agents can reduce consumer reliance on search engines and marketplaces for product discovery, effectively disintermediating the incumbents. The shift would position AI purchasing agents as challengers capable of breaking the stronghold that Google and Amazon currently have on digital marketplaces.

What steps must digital wallets take to disintermediate search and “walled gardens” to drive consumer adoption in this evolving landscape? Critical to their early adoption, AI-powered purchasing agents will have to offer personalized deals, promotions, discounts, and rewards programs as they gather valuable user data with detailed insights about preferences and purchase histories. Perhaps showing the way, credit card providers have been collecting data and incentivizing¹¹ consumer behavior for the past 50-60 years.¹² In 2022 alone, credit card rewards totaled $40 billion¹³ in the United States. Credit card loyalty programs have reshaped consumer behavior by steering users toward credit card portals instead of third-party aggregators. According to Skift research, for example, when booking flights and hotels, 36% of Gen Z and 32% of millennials¹⁴ have opted for credit card travel portals instead of third-party platforms to maximize their rewards, as shown below. 

Source: ARK Investment Management LLC, 2024, based on data from Skift Research¹⁵ as of October 24, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

Digital wallets should be able to offer their own integrated rewards systems, enabling users to accumulate and spend points throughout the e-commerce journey. As with credit cards, consumer retention will be key to success, enabling AI purchasing agents to personalize services more effectively. Over time, data-driven insights should enable highly tailored, hyper-personalized recommendations, boosting user engagement and adoption. Netflix’s success with this concept is instructive: by personalizing and curating content based on individual viewing habits and preferences, Netflix’s recommendation engine now drives 80%¹⁶ of the content viewed on its platform. As a result, in the race for market share dominance, short-term adoption will be crucial for long-term growth.

In addition, purchasing agents should preclude shoppers’ dependency on one or two marketplaces. Digital wallet providers that streamline and personalize the e-commerce experience could redefine the consumer experience and value chains, enabling consumers to become “marketplace-agnostic” and rely more on AI to aggregate information and facilitate purchases across many providers. As with many “winner takes most” AI opportunities, the disruption could be profound, and digital wallets are well positioned to capitalize on this opportunity. 

We anticipate a future in which LLM-backed purchasing agents are integrated into digital wallets, offering consumers the opportunity to delegate most of their shopping experiences—from product discovery to checkout—to AI-driven purchasing agents. By offering a seamless process along with attractive rewards, purchasing agents tied to digital wallets could see widespread adoption by 2030.