Agentic Commerce: How to win and retain customer loyalty

The retail industry is staring at a new future. After decades of optimising the customer journey – fine-tuning every click, scroll and button and linking digital and physical journeys – retailers now face something different: consumers are starting to delegate their shopping decisions to AI Agents. This shift from human-driven shopping through direct engagement with the retailer, to agent-enabled commerce threatens to upend the economics of retail as dramatically as eCommerce did two decades ago while also breaking the loyalty bond between retailer and consumer.

The Scale of Change That’s Coming

The numbers reveal an accelerating transformation. Traffic to US retail sites from generative AI browsers and chat services increased 4,700% year-over-year in July 2025, whilst more than half of consumers anticipate using AI assistants for shopping by the end of the year, largely for conducting research¹. These aren't casual browsers – customers arriving via AI Agents are 10% more engaged than traditional visitors, spending 32% more time on site, browsing 10% more pages and showing a 27% lower bounce rate.

The financial implications are substantial. McKinsey estimates generative AI alone could unlock $240–$390B in value for retailers, equivalent to a 1.2–1.9% margin boost industry-wide². When it comes to Agentic Commerce, these benefits will come through a mix of adapting to the new reality as well as investing to remain relevant to consumers, directly. However, this value won't be distributed evenly. Winners and losers will emerge based on how effectively retailers adapt to a world where AI Agents, not humans, increasingly intervene in the shopping experience.

The Emerging Ecosystem

New entrants, such as OpenAI and Anthropic, have radically transformed the AI landscape in recent years³. And when ecosystems form around new entrants and their innovations, it means industry shifts are happening. OpenAI launched Operator in January 2025 which is now part of ChatGPT. Operator uses agents to automate tasks such as booking travel and restaurant reservations⁴. OpenAI also announced the Agentic Commerce Protocol, developed with Stripe⁵. This protocol lets users complete purchases inside ChatGPT without leaving the chat.

Shopify is building an agentic shopping system⁶. It lets agents access Shopify’s catalogue and create carts that include products from multiple merchants. Perplexity has moved faster than most. It launched “Buy with Pro,” which enables one-click purchases directly from selected merchants⁷ and partnered with PayPal for payments⁸.

New entrants are creating the movement towards Agentic Commerce and partnering with incumbents to accelerate the shift. The trend towards Agentic Commerce is real and cannot be ignored.

The Big Tech Advantage

While the media has been preoccupied with the ‘automagic’ created by Large Language Models from emergent technology companies such as OpenAI and Anthropic, the large incumbent technology companies enter this transition to Agentic Commerce with significant structural advantages. Google Wallet provides payment infrastructure across billions of transactions, offering Google unprecedented insight into consumer purchasing patterns. This transactional data, combined with search history, location information, and browsing behaviour, creates a comprehensive understanding of consumer intent. Google ID, which can be used to log into ChatGPT, Claude and others, will also give Google a data advantage.

When Google develops AI shopping agents powered by Gemini, this data advantage will become decisive – and this is happening now⁹. Google and Walmart announced a fully integrated shopping experience within Gemini, giving customers access to the full assortment and delivery in under 3 hours¹⁰.

Similarly, Amazon's fulfilment network, customer purchase history, and Prime membership data create formidable competitive moat. Apple's ecosystem, seamlessly connecting purchasing across devices, leveraging Apple Pay's transaction data and web search history across Safari, provides another structural advantage. These technology giants can offer AI shopping experiences that span categories and merchants, aggregating consumer intent in ways individual retailers will not be able to match.

The Trust Hurdle

The transition to Agentic Commerce hinges on consumer willingness to trust AI Agents with purchase decisions. Whilst 72% of US consumers have used AI in some form, only 24% feel comfortable using it to make a purchase today¹¹. Furthermore 66% of consumers would refuse to let AI make purchases on their behalf even if they get a better deal¹². Payment security ranks as consumers' biggest concern, followed by privacy worries and loss of control¹³.

Despite these negative sentiments, the adoption for shopping assistance is accelerating. 53% of consumers now experiment with or regularly use gen AI—up from 38% in 2024¹⁴. Nearly 60% are projected to use AI to help them shop, embracing agents for research, price comparison, and product discovery¹⁵. Here is the critical distinction: consumers want AI Agents as collaborators for discovery and comparison, not as autonomous purchasers, yet. As with all technology-led change in the last 20 years (internet use, smartphone adoption and social media participation), societal attitudes and engagement changes with time, and the seeds of change are sown by the generations who first encounter the technology as native, not new. While ‘trust’ barriers exist today, they will not persist in the years to come, as history teaches.

The Challenge to Retailers: Four Threats to Today’s Business Model

Agentic Commerce is here today and will cement itself in consumers’ lives in the years to come. This presents four challenges to retailers which have the potential to upend their business models.

1. Disintermediation of Consumer Relationships
When consumers delegate shopping to AI Agents, they interact primarily with the agent interface - ChatGPT, Perplexity, Google Gemini - not with retailer brands. This disrupts the direct customer relationship that retailers have spent decades cultivating. ChatGPT now has more than 800 million weekly users¹⁶, and Google's AI overviews powered by Gemini now reach more than 1.5 billion users per month¹⁷.

When an AI Agent executes a purchase, the consumer's loyalty may attach to the agent that provided the recommendation, not to the retailer that fulfilled the order. AI Agents optimise for objective criteria such as price, delivery speed, product specifications, potentially commoditising retailers into interchangeable suppliers. The loyalty value exchange is diminished, cycles of repeat purchase disrupted and plannability to meet consumer needs is reduced.

2. Margin Compression Through Information Transparency
Agentic Commerce has the potential to introduce near-perfect price transparency, beyond the way in which today’s browser extensions work. AI Agents promise to be able to instantly compare prices across dozens of retailers, factor in shipping costs, apply available coupons, and calculate the true total cost. This capability eliminates information asymmetry that has traditionally protected retailer margins.

The challenge extends beyond pricing. AI Agents will also be able to evaluate product quality through aggregated reviews, assess sustainability credentials, compare specifications, and identify substitute products. This comprehensive analysis will empower agents to negotiate effectively on consumers' behalf, potentially commoditising products and compressing margins across categories.

3. Loss of Cross-Sell and Upsell
Retail relies heavily on merchandising tactics: product placement, bundle offers, complementary item suggestions and strategic promotions. These tactics work because they influence human browsing behaviour—but AI Agents don't browse. They search with purpose. When an agent is tasked with purchasing shoes, it won't be distracted by a prominently displayed jacket or persuaded by a limited-time offer on unrelated items. While Agentic Commerce will apply to eCommerce, the share of sales in physical stores is predicted to decline from 45% in 2024 to 41% in 2026¹⁸. This means that that an increasing share of sales will be conducted digitally and fall within the scope of Agentic Commerce. AI Agents will also play an increasingly important role in helping consumers research and discover items that they want to purchase leading to an unpredictable distortion in sales in physical stores, as well. Both of these dynamics will reduce opportunities for impulse purchases (in store and online) and staff-assisted upselling.

4. Power Shift to the Supplier
As AI Agents begin to bypass retailers’ websites, those retailers will lose some of their significance as the main connection between brands and consumers. This will shift the balance of power towards suppliers. In turn, suppliers are likely to redirect their budgets towards Agentic Commerce platforms to ensure their products are visible, discoverable, and competitively priced. They will also increase investment in their own AI capabilities to optimise algorithmic preferences and develop “agent-first” offers, targeting consumers directly.

As suppliers’ marketing and promotional funding moves upstream, away from retailers and towards Agentic Commerce platforms, retailers may face detrimental consequences. Customer experience, pricing, assortment, loyalty and the retailer’s underlying economics may all be adversely affected. Although this transition will happen gradually rather than abruptly, the shift in power and its implications for retailers will become increasingly apparent over the coming decade.

The Response from Retailers: Four Strategic Imperatives to Win

The rules of engagement for loyalty will change. For retailers to win and retain their relevancy, a re-evaluation of customer engagement strategy and advances in AI are prerequisites to compete against Agentic Commerce. There are four ‘must haves’ for every retailer.

1. Modernise Loyalty Programmes for Agentic Commerce
97% of UK grocery shoppers are members of at least one loyalty programme¹⁹, but traditional closed-loop points-and-perks models won't suffice when AI Agents mediate purchases. The Card Schemes are already preparing for the era of Agentic Commerce – connecting permissioned sharing of transaction data to AI Agents for greater personalisation and relevancy. Visa’s Intelligent Commerce platform enables payments through AI Agents while delivering personalised experiences to customers and applying promotions and points directly to their account²⁰. MasterCard offers something similar²¹. These advancements by the Card Schemes further compound the risk of disintermediation to retailers by channelling loyalty to the customer through the payment card while within the AI Agent’s shopping workflow.

Retailers’ loyalty programmes must evolve to become an integral part of the Agentic Commerce ecosystem and counteract the dual threat from AI Agents and Card Schemes. Retailer loyalty programmes must become data-rich, dynamically personalised systems that AI agents can query, evaluate, and incorporate into recommendation algorithms in real time. The case needs to be made for AI Agents to link retailer loyalty programmes into their workflows. How this happens is yet to be seen but the starting point is for retailers to modernise their loyalty programmes to make them Agentic Commerce ready.

More broadly, retailers’ loyalty programmes need to expand beyond transactional benefits to maintain a ‘pull’ from consumers. This could include communities, VIP destinations, competitions or other such mechanics which can be attractive to customers. The right mix of loyalty mechanic and softer benefits will depend on a given retailer’s brand and engagement strategy. As the industry stares into a new future, it is timely to revisit the overall loyalty strategy (and the design of loyalty programmes) in the context of Agentic Commerce.

2. Deliver Quantifiable and Differentiated Shopper Benefits That Influence AI Agents
Where AI Agents are making purchasing decisions on behalf of consumers, retailers can differentiate themselves by the benefits they offer. These benefits must be quantifiable, verifiable and programmatically accessible. Because AI Agents will identify the ‘best’ option for their consumer, algorithmically, retailers can compete on parameters such as faster shipping, longer return policies and price guarantees.

Retailers can go further by creating exclusive benefits for AI Agents to gain a competitive edge. Consider sustainability as a member benefit. Consumers increasingly prioritise environmental credentials, but AI Agents need structured data to evaluate claims. Retailers can quantify their carbon footprints, document ethical sourcing and verify sustainability certifications. Making these attributes machine-readable so agents can incorporate them into recommendations will help edge out the competition.

3. Deploy Advanced Behavioural AI Models to Train AI Agents
The most powerful counter-strategy available to retailers today is to create personalisation capabilities that are so sophisticated that they end up training AI Agents directly. While the AI Agent will work autonomously on behalf of a consumer, the consumer is still in the loop.

Around 70% of consumers expect retailers to anticipate their needs and proactively post product recommendations, offers or information at the precise moment that they become relevant²². While AI Agents will be handling the hard work of finding the cheapest product with the best delivery slot and creating baskets across multiple retailers, the consumer (or the AI Agent’s client) will still have the final say. The shopping experience may be iterative, for example. While these iterations, led by the consumer, will train the AI Agent, retailers’ behavioural models can ‘learn’ and profile the consumer by proxy and in so doing surface tailored offers, content and other benefits which will be valued. Every interaction becomes a source of insight, creating a continuous feedback loop that enhances future recommendations. In turn, that retailer becomes recognised as a preferred supplier thanks to the reinforcement from the consumer themselves. In the world of Agentic Commerce, winning the AI Agent’s ‘share of mind’ becomes as important as winning the customer's share of wallet.

4. Build Comprehensive Customer Profiles Through Data Aggregation
While the response to Agentic Commerce requires new influencing strategies of the AI Agents themselves, this does not negate the need for a direct-to-consumer strategy. And in a world of AI-powered loyalty and personalisation, retailers’ need to consider something existential: AI without data is like having a bath with no water.

Understanding consumers requires the assembly of a complete picture from fragmented data sources. Retailers can create unified customer profiles that aggregate first-party transactional data, third-party enrichment sources, and consumer-permissioned financial information through mechanisms like open banking. And a Customer Data Platform is where this data is aggregated.

While 81% of firms with sales greater than $10B have a CDP in place or are in the process of implementing one, this number drops to 27% for smaller firms²³. However, of those with a CDP, only 64% say that the deployment is delivering significant value and only 14% of organisations have achieved a 360-degree view of their customer²⁴. This reveals the execution gap today. Building unified customer profiles from ‘internal’ data (transactions, customer data) becomes an urgent priority to be ready to compete in the Agentic Commerce era. Enhancing these profiles, with ancillary data sources, then becomes critical for competitive advantage. Why? Because better data drives better personalisation, which attracts more customers, which generates more data. This holds true for both Agentic Commerce and direct-to-consumer engagement.

Many retailers are already enriching customer profiles with third party data sets such as location, weather and ingredient data. However, Open Banking (account aggregation and data sharing) represents a particularly powerful data source and is largely untapped. With permission, and very likely a value exchange mechanic, retailers could access a customer’s financial data to reveal spending patterns across retailers, income signals and purchase power. Retailers would then have a richer view of consumers’ financial lives, understand where and when they make purchases and the retailers they prioritise for certain missions. Aggregating this data into consumer profiles completes a retailer’s understanding of their behaviour, making behavioural predictions more accurate and retailer actions more meaningful.

In the UK, the Open Banking method of ‘pay by bank’ attracts an average of 27M payments per month and 90% of eCommerce merchants intend to launch it soon²⁵. This could increase consumer comfort to share financial data via Open Banking, where retailers’ savings from reduced interchange fees are redistributed as incentives to participating customers.

A New Future is in Touching Distance

The transition to Agentic Commerce is not a distant possibility – it is unfolding now as millions of consumers are using AI Agents to research and discover products. The infrastructure and ecosystems are being built for a future where AI Agents can mediate millions of shopping decisions, through to orchestrating purchases and delivery, and their influence will likely accelerate. For retailers, this moment demands a fundamental question: will you be a commoditised supplier in an AI-orchestrated marketplace, or will you be the preferred partner that AI Agents recommend and consumers trust?

The answer lies not in resisting this transformation, but in reshaping businesses to thrive within it. Retailers that modernise their loyalty programmes, create differentiation, deploy sophisticated behavioural AI, and build comprehensive customer profiles will emerge stronger. Those that delay – hoping the shift will be slower or less significant – risk becoming invisible to the very AI Agents that will soon guide their customers' purchases.

The retailers that win in the age of Agentic Commerce will be those that recognise that loyalty must now be earned twice – once from the customer and once from the AI Agent acting on their behalf. The time to prepare for that reality is not tomorrow. It is today, while there is time to act.

Sources

• ¹https://business.adobe.com/uk/blog/generative-ai-powered-shopping-rises-with-traffic-to-retail-sites
• ²https://www.mckinsey.com/industries/retail/our-insights/llm-to-roi-how-to-scale-gen-ai-in-retail
• ³https://www.reuters.com/business/openai-lays-groundwork-juggernaut-ipo-up-1-trillion-valuation-2025-10-29/
• ⁴https://openai.com/index/introducing-operator/
• ⁵https://openai.com/index/buy-it-in-chatgpt/
• ⁶https://www.shopify.com/uk/commerce-for-agents
• ⁷https://www.perplexity.ai/help-center/en/articles/10352906-what-is-shop-like-a-pro
• ⁸https://newsroom.paypal-corp.com/2025-05-14-Perplexity-Selects-PayPal-to-Power-Agentic-Commerce
• ⁹https://blog.google/products/shopping/agentic-checkout-holiday-ai-shopping/
• ¹⁰https://corporate.walmart.com/news/2026/01/11/walmart-and-google-turn-ai-discovery-into-effortless-shopping-experiences
• ¹¹https://www.bain.com/insights/agentic-ai-commerce-hinges-on-consumer-trust/
• ¹²https://www.omnisend.com/latest-news/two-thirds-of-shoppers-say-no-to-ai-shopping-assistants-trust-issues-could-slow-retails-ai-revolution/
• ¹³https://ir.riskified.com/news-releases/news-release-details/global-study-73-shoppers-using-ai-shopping-journey-merchants
• ¹⁴https://www.deloitte.com/us/en/insights/industry/telecommunications/connectivity-mobile-trends-survey.html
• ¹⁵https://www.kearney.com/industry/consumer-retail/article/as-agentic-commerce-reshapes-retail-here-s-how-leaders-can-win
• ¹⁶https://techcrunch.com/2025/10/06/sam-altman-says-chatgpt-has-hit-800m-weekly-active-users/
• ¹⁷https://business.google.com/us/think/search-and-video/ai-powered-search-marketing/
• ¹⁸https://www.salesforce.com/au/retail/retail-industry-trends/
• ¹⁹https://www.gov.uk/government/publications/review-of-loyalty-pricing-in-the-groceries-sector/executive-summary
• ²⁰https://corporate.visa.com/en/products/intelligent-commerce.html
• ²¹https://www.mastercard.com/global/en/business/artificial-intelligence/mastercard-agent-pay.html
• ²²https://business.adobe.com/uk/resources/digital-trends-report.html
• ²³https://www.cdpinstitute.org/news/customer-data-platform-growth-shifts-to-new-markets-cdp-institute-report/
• ²⁴https://www.mastercardservices.com/en/industries/financial-institutions/insights/rethinking-customer-engagement-era-open-banking
• ²⁵https://truelayer.com/blog/payments/the-pay-by-bank-2025-update-is-in/