An increasing number of consumers are beginning their shopping journey by asking artificial intelligence for advice, either through conversations with large language models or by using AI assistants available on retailers’ websites.
According to the fourth edition of Salesforce’s State of Commerce report, the use of agentic search as the first step in the purchasing process increased by 200% year on year.
The report’s findings are based on a survey of 3,450 commerce and e-commerce professionals, 4,690 consumers and an analysis of the behaviour of more than 1.5 billion shoppers worldwide.
Companies are now under pressure from two directions. As many as 86% of commerce and e-commerce leaders believe the development of artificial intelligence is significantly raising customer expectations, while their teams are being asked to meet those expectations with the same or even fewer resources.
As a result, the implementation and development of AI solutions have become organisations’ most important priority for the coming year. At the same time, they are also viewed as their greatest challenge.
“Artificial intelligence is transforming commerce from two directions at once. Customers are discovering products in places over which brands do not have full control, through AI assistants, social media and delivery apps, and they expect the same personalised experience everywhere,” said Caila Schwartz, Head of Agentic Commerce Shopper Insights at Salesforce.
“At the same time, AI is changing how commerce teams work, from managing product ranges and fulfilling orders to scaling operations. The key difference is the pace of change. Customers can begin using AI almost overnight. Companies, however, must address many more issues, including reliable data, integrated systems and teams prepared to use new tools effectively.”
“Organisations that are already building these foundations will be best placed to keep up with the transformation,” she added.
AI Is Changing Product Discovery, Increasingly Outside Brand-Owned Channels
The clearest evidence comes from actual behavioural data covering more than 1.5 billion shoppers using e-commerce platforms around the world.
Traffic referred from AI conversations increased in every quarter analysed, rising by between 150% and 428% year on year. By comparison, overall traffic grew by only several or, at most, slightly more than 10%.
Consumer surveys show a similar trend.
Between August 2025 and May 2026, the proportion of people discovering products through brand-owned channels fell by 7%, while the share using traditional search engines declined by 15%.
At the same time, the number of consumers using emerging channels, including AI assistants, AI solutions embedded in social media and delivery applications, increased by 38%.
Commerce leaders clearly understand the direction of this shift. Some 86% believe that large language models will become a key product-discovery tool within the next year.
Organisations are already taking steps to prepare. Their most common investments include improving the quality of product information, optimising content for conversational queries, making product data available to AI platforms and adapting descriptions to natural language.
A customer may, for example, ask an AI assistant to “find waterproof hiking boots costing less than 150 dollars” rather than entering a short sequence of keywords into a conventional search engine.
The transformation is currently most visible at the product-discovery stage, while fully autonomous shopping remains at an early stage of development.
“Customers no longer begin their shopping journey by browsing a list of search-engine links. Increasingly, they ask artificial intelligence what is worth buying, and by the time they reach a brand’s website, they have already formed an opinion,” Schwartz said.
“If a brand is not present in that first conversation, it becomes invisible before the customer journey has even properly begun.”
Agentic AI Adoption Is Growing as Companies Move from Pilots to Scale
Fewer than one-third of organisations currently say they use agentic artificial intelligence, but the pace of adoption is accelerating rapidly.
Almost half of the companies that have not yet implemented such solutions plan to do so within the next six months.
At the same time, 71% of respondents believe that it will be impossible to scale operations effectively across multiple channels without using AI.
Among organisations already using agentic AI, the experimental phase is largely over. Only 4% say they are still focused mainly on pilot projects.
Most are now concentrating on scaling AI across different areas of their operations, ranging from customer service and IT to product-range management.
Companies implementing agentic AI are also reporting measurable benefits. The most frequently cited include increased customer satisfaction, stronger personalisation, faster development of new solutions, greater operational efficiency and higher employee productivity.
The report also highlights a significant gap. Only 32% of organisations have clearly defined success metrics and key performance indicators for AI-related projects.
This means that many companies are seeing positive results from their implementations without having established precise criteria for assessing their effectiveness.
Without appropriate indicators, it is difficult to determine which investments are delivering the best results and should be prioritised for further expansion.
Increasing Technology Complexity Is Becoming a Major Challenge
One reason is that AI is being implemented on top of technological infrastructure that was already under considerable strain.
A growing number of technology vendors, fragmented customer data and poorly integrated sales and communication channels mean commerce and e-commerce teams are deploying AI on foundations that often do not function efficiently.
The report’s main findings illustrate the scale of the problem:
- 78% of commerce and e-commerce leaders say the number of technology vendors used by their organisation has increased over the past two years.
- Only 27% of organisations say their customer data is fully integrated across sales, customer service, marketing and commerce.
- Among companies with fragmented data, 40% say it results in slow or inefficient resolution of customer problems.
- A further 39% struggle to assess the impact of commerce and e-commerce investments.
- Another 39% point to the high cost of maintaining disconnected systems.
Problems with omnichannel sales operations are almost universal. Only 2% of organisations operating across multiple channels report no significant disruption.
The most frequently cited problems are inconsistent prices and promotions, reported by 41%, and the absence of real-time inventory synchronisation, reported by 40%.
The report shows that data-integration challenges are also visible in physical stores.
Although brick-and-mortar outlets remain the main shopping destination during the holiday season for 77% of consumers, the customer journey itself is becoming increasingly digital.
Some 79% of customers use a smartphone while shopping in a store, while 12% ask an AI assistant for advice on choosing a product during their visit.
Customers view all touchpoints as part of a single brand experience. As many as 88% of representatives of business-to-consumer companies believe shoppers expect the same level of personalisation in physical stores as they receive online.
Organisations with Integrated Data Are Already Seeing Measurable Benefits
Companies that have begun unifying their data most frequently report benefits with a direct impact on business growth.
These include improved cooperation between sales, marketing and e-commerce teams, greater effectiveness of AI and automation solutions, and stronger customer retention and loyalty.
“The organisations gaining an advantage are not those implementing the largest number of AI solutions, but those using the development of artificial intelligence as an opportunity to organise and unify their data and define what success genuinely looks like,” Schwartz said.
“Consumers have already decided to use AI. It is now up to companies to catch up by building the right foundations.”
Survey Methodology
The data presented in the report comes from a double-blind survey conducted between 10 April and 4 June 2026.
The study included 3,450 commerce and e-commerce professionals from 20 countries, representing 13 industries.
The findings were supplemented by Salesforce’s analysis of the shopping behaviour of more than 1.5 billion consumers using commerce platforms in 37 countries between the first quarter of 2024 and the first quarter of 2026.
An additional survey was conducted among 4,689 consumers from eight countries between 12 and 18 May 2026.
Detailed information on the sample structure and demographic data is available in the full report.
Source: CEO.com.pl
