January 6, 2026
Sonyachny/Depositphotos.com
Did Holiday 2025 Mark a Shift Into the ‘Agentic Shopping’ Retail Era?
According to extensive data provided by Salesforce, agentic AI was the most significant retail story behind this year’s holiday sales season, driving approximately one-fifth (20%) of all retail sales while spurring $262 billion in revenue via the twin streams of customized recommendations and enhanced customer service.
“The 2025 holiday season marked a definitive shift to a new era of ‘agentic’ shopping,” said Caila Schwartz, director of consumer insights for Salesforce.
“While shoppers remained resilient in the face of higher prices, the real story was how retailers leaned on AI and agents to navigate the holiday rush. Agents didn’t just drive $262 billion in sales through high-intent discovery; they became the operational heroes of the season — handling a 142% surge in tasks like returns and shipping updates. This wasn’t just a bigger holiday season than last year; it was a more efficient, intelligent one,” Schwartz added.
Agentic Shopping Dominated the Data, But Other Retail Insights Were Provided
Other findings from Salesforce included:
- During the holiday season spanning Nov. 1 through Dec. 31, online retail sales topped $1.29 trillion globally and $294 billion stateside. These figures represent a 7% uptick and 4% improvement, respectively, year-over-year. That growth accelerated further during the last two weeks of December, hitting 12% on a global basis and 9% in the United States.
- The share of both global and U.S. traffic from AI search channels — including ChatGPT and Perplexity, among others — doubled compared to the year prior, indicating massive adoption of the tech in the retail sphere. Even more striking: “Shoppers referred to retailer websites from AI-based search channels converted nine times more often than those coming through social media referrals,” Chain Store Age’s Dan Berthiaume noted.
- In total, AI agents were called upon to manage 142% more requests throughout the two-month period examined than they had in the two months prior. Further, retail AI and associated agentic bots in the customer service stream were engaged 126% more often during the holiday sales season than during the September-October time frame.
- Perhaps unsurprisingly, buy online and pick up in-store (BOPIS) orders peaked just days before Christmas, on Dec. 22, where 35% of all online orders fell into this category. Overall, almost 20% of all online orders were BOPIS during the holiday retail season, trending upward to nearly 33% in the last five days before Christmas.
Returns also showed growth in terms of volume, with $181 billion in purchases made between November 1 and December 31 having already been returned — that’s 14% of the total purchase count and a 10% increase over last year’s figure. Average sale price went up by about 7% both worldwide and stateside, and seasonal order volumes hiked upward by 3% globally and just 1% in the U.S.
BrainTrust
Raj B. Shroff
Founder & Principal, PINE
Perry Kramer
Managing Partner, Retail Consulting Partners
Neil Saunders
Managing Director, GlobalData
Recent Discussions
Discussion Questions
Do you agree with the notion that holiday 2025 marked the sales season in which retail turned the corner into a new era of agentic AI-driven shopping? Why or why not?
What can be said about the overall sales figures, as well as the BOPIS numbers, in terms of lessons learned?
Poll
Loading …
Turned the corner…from what exactly? It wasn’t zero in 2024, and it won’t be 100% this year (or ever)
And should we really care? If AI searches are replacing google (et al) searches, which in turn replaced various crude methods – like asking people (or using one’s own intuition) – the fundamental proposition of people buying things (that they either want or need) hasn’t changed (nor will it).
Spot on—the ‘fundamental proposition’ hasn’t changed, which is why we should be skeptical of that 20% figure. Just because a marketplace has inducted AI into its search bar doesn’t mean the AI caused the sale; it’s often just a new label on the same old behavior. In 2026, the real detail we need is incrementality—did the AI actually create a sale that wouldn’t have happened otherwise, or is it just taking credit for a ‘calculated’ shopper who was going to buy anyway?
The 2025 holidays marked a milestone for agentic AI infrastructure and adoption. Top retailers invested in agentic for efficient processes like inventory management and labor scheduling. Shoppers sought personalized retail solutions like gift and recipe recommendations. Agentic accelerated results for retail stakeholders — and it’s just getting started.
It’s undeniable that agentic AI is becoming the ‘operational hero’ for backend tasks like inventory, but we have to be careful not to conflate operational efficiency with consumer demand. While a 20% ‘influence’ stat sounds like a milestone, the real challenge for 2026 is distinguishing between AI that simply facilitates a sale and AI that actually creates incremental growth. High adoption is great, but until we can prove these agents are driving ‘new-to-brand’ customers rather than just assisting shoppers who were already there, the 20% remains more of a correlation than a direct win.
I think agents AI led the back end (operations, customer support, etc) in ways more advanced than ever before. As for shopping, it would seem it was more traditional AI-driven versus agentic. Users still need to prompt AI for most shopping tasks. Maybe holiday 2026 will have more advanced shopper-facing tools that enable more autonomous help tasks but that could still be a few years away. The most amazing statistic to me is that, “Shoppers referred to retailer websites from AI-based search channels converted nine times more often than those coming through social media referrals.” That is pretty mind-blowing. If you don’t have your GEO figured out, “you are cooked”, as my teens would say.
Ai based / augmented searches clearly accelerated customers’ getting to the checkout process. However, when you see a 14% increase in early returns it feels almost disappointing in that customers may not be as thorough in their research. The real story will evolve over the next two years as the AI engines are further integrated to the checkout process somewhat by passing the retailers web sites UI.
Great point in that customers may not have been/be as thorough, over relying on their first chat response for a recommendation.
Consumers increasingly embraced personalized shopping experiences facilitated by artificial intelligence during Holiday 2025. AI agents played a more prominent role in shoppers’ purchasing decisions, as they sought tailored recommendations and seamless transactions. With this shift, consumers gained greater control over their shopping journeys by enabling technology to increase efficiency and autonomy.
SO much PR hype around this! A website used AI chatbots – another showed complimentary items. “Florals for spring…Groundbreaking”
Humans on the sales floor probably engaged 200% more often during the holiday sales season than during the September-October time frame and produced a heck of a lot more sales. Why does no one value them? Geez
I’m hesitant to declare Holiday 2025 as the singular turning point into an agentic, AI-driven retail era — mainly because we simply haven’t yet seen all the data — both the hard numbers and the deeper qualitative insights — needed to make that judgment. What is clear is that AI-based shopping tools, personalized recommendations, and generative-driven discovery have become much more visible in consumer experiences this season than in prior years, and adoption rates among shoppers exceeded many expectations. But a single holiday cycle — even one as important as 2025 — isn’t yet a definitive inflection point in the long arc of retail evolution. Real structural change shows itself over quarters and years, not just in seasonally-peaked bursts. We need to observe sustained behavior change in shopper engagement, conversion lift attributable to AI interventions, and the operational impacts on retailers before calling this a wholesale shift into a new era.
When we look at overall sales and fulfillment metrics — including strong holiday sales and elevated BOPIS (buy-online, pick-up-in-store) performance — there are valuable lessons about consumer expectations and omnichannel orchestration. BOPIS growth underlines that shoppers still value convenience and immediacy, and that AI-enhanced search, intelligent recommendations, and inventory visibility tools help make those hybrid journeys more seamless. What we can celebrate from this season is that AI has moved beyond pilot projects and point innovations into real customer-facing use cases that shaped how many people discovered products, compared options, and completed purchases. But whether that translates into a paradigm shift that endures through 2026 and beyond requires more longitudinal evidence — including return rates, post-holiday retention of AI-driven behaviors, and how the technology impacts key drivers like margin, loyalty, and customer lifetime value.
In short, Holiday 2025 may be an early milestone in the adoption of agentic AI in retail, but it’s too soon to call it the definitive “corner turned.” The season offered encouraging signals that AI is becoming more embedded in the shopping experience, yet retailers and analysts alike must continue to monitor trends over time — not just in peak periods — to determine whether this year truly marks the start of a new era or simply a compelling early chapter in a longer transformation.
Salesforce mislabels retailer-controlled recommendation engines as “agentic shopping.” These are fundamentally different. What they’re measuring is sophisticated manipulation, AI optimizing for retailer revenue through “customized recommendations,” not consumer-controlled agents optimizing for genuine buyer benefit. The 14% return rate exposes the gap; these systems are optimized for conversion, not fit. Real agentic commerce means consumer-controlled agents evaluating options against actual needs, not platform algorithms persuading humans toward higher-margin purchases. Holiday 2025 may mark retailers’ perfecting AI-driven persuasion just as the window closes before consumer-controlled agents make such manipulation obsolete. The battle is real, but it isn’t about whether AI mediates commerce; it’s about who controls the AI and whose interests it serves.
“What they’re measuring is sophisticated manipulation.” Perfect description of how the AI is programmed.
Yep, the return rate blows the cover on this whole conversation. I was a little less even handed in my comments.
Being blunt, I am suspicious of these numbers. While I am sure that they are accurate in their own way, the devil is in the detail.
If we are saying that 20% of sales were influenced in some way by AI – including people using chatbots to generate ideas, or using them to check shipments – then, while they seem a little high, this is feasible. However, this is a very loose use of AI and AI has simply replaced some of the traditional search processes.
If we are saying that 20% of sales/transactions were made directly via AI then this is completely off. We know this because only 22% of holiday sales are made online – so it would suggest that almost every single one of those was driven by AI – which is clearly nonsense.
The other point I would make here is that there is way too much hyping of AI to try and generate PR noise. What we need are realistic and nuanced views of how AI is having an impact (which it is) rather than exaggerated claims.
Wait a minute. Deep breath. I have buckets of skepticism about this whole conversation. First, we are getting all this from Salesforce? Is it data or Salesforce marketing? They say it was a bigger, and more “efficient and intelligent” holiday season. Really? Am I reading this all correctly that growth in returns outpaced growth in sales? How is that possibly more efficient and intelligent? Sounds expensive, reeeaaalllyyy expensive. So now lazy and sloppy customers can get more stuff faster and then return more of it…faster. Doesn’t sound like retailers would view that as cheerful news. This sounds like the ‘free shipping/free returns’ mentality of two decades ago…on steroids. Good for customer acquisition, and really expensive in the long run as it becomes table stakes.
There’s a very natural evolutionary move taking place here that both customers and retailers are embracing. So agentic AI will become table stakes faster than ecomm became table stakes. Calling it a “new era” sounds like all rainbows and unicorns. It’s not that simple. How about a series of articles that talk about success stories and horror stories and ROI. Yeah…ROI. What is all this costing? Who’s paying? Weren’t we all going to be living in the metaverse by now?
More Discussions
Will Ryan Cohen Score GameStop’s $35 Billion Pay Package?
GameStop CEO Ryan Cohen has been offered a $35 billion pay package based around performance. Will…What Should Saks Global Try Next?
The trouble looks very uncertain for Saks. What is its next move?Should Facial Recognition for Theft Get Another Look?
Retailers and consumers are hotly debating facial recognition tech: theft deterrent, privacy overstepping, or both?
Turned the corner…from what exactly? It wasn’t zero in 2024, and it won’t be 100% this year (or ever)
And should we really care? If AI searches are replacing google (et al) searches, which in turn replaced various crude methods – like asking people (or using one’s own intuition) – the fundamental proposition of people buying things (that they either want or need) hasn’t changed (nor will it).
Spot on—the ‘fundamental proposition’ hasn’t changed, which is why we should be skeptical of that 20% figure. Just because a marketplace has inducted AI into its search bar doesn’t mean the AI caused the sale; it’s often just a new label on the same old behavior. In 2026, the real detail we need is incrementality—did the AI actually create a sale that wouldn’t have happened otherwise, or is it just taking credit for a ‘calculated’ shopper who was going to buy anyway?
The 2025 holidays marked a milestone for agentic AI infrastructure and adoption. Top retailers invested in agentic for efficient processes like inventory management and labor scheduling. Shoppers sought personalized retail solutions like gift and recipe recommendations. Agentic accelerated results for retail stakeholders — and it’s just getting started.
It’s undeniable that agentic AI is becoming the ‘operational hero’ for backend tasks like inventory, but we have to be careful not to conflate operational efficiency with consumer demand. While a 20% ‘influence’ stat sounds like a milestone, the real challenge for 2026 is distinguishing between AI that simply facilitates a sale and AI that actually creates incremental growth. High adoption is great, but until we can prove these agents are driving ‘new-to-brand’ customers rather than just assisting shoppers who were already there, the 20% remains more of a correlation than a direct win.
I think agents AI led the back end (operations, customer support, etc) in ways more advanced than ever before. As for shopping, it would seem it was more traditional AI-driven versus agentic. Users still need to prompt AI for most shopping tasks. Maybe holiday 2026 will have more advanced shopper-facing tools that enable more autonomous help tasks but that could still be a few years away. The most amazing statistic to me is that, “Shoppers referred to retailer websites from AI-based search channels converted nine times more often than those coming through social media referrals.” That is pretty mind-blowing. If you don’t have your GEO figured out, “you are cooked”, as my teens would say.
Ai based / augmented searches clearly accelerated customers’ getting to the checkout process. However, when you see a 14% increase in early returns it feels almost disappointing in that customers may not be as thorough in their research. The real story will evolve over the next two years as the AI engines are further integrated to the checkout process somewhat by passing the retailers web sites UI.
Great point in that customers may not have been/be as thorough, over relying on their first chat response for a recommendation.
Consumers increasingly embraced personalized shopping experiences facilitated by artificial intelligence during Holiday 2025. AI agents played a more prominent role in shoppers’ purchasing decisions, as they sought tailored recommendations and seamless transactions. With this shift, consumers gained greater control over their shopping journeys by enabling technology to increase efficiency and autonomy.
SO much PR hype around this! A website used AI chatbots – another showed complimentary items. “Florals for spring…Groundbreaking”
Humans on the sales floor probably engaged 200% more often during the holiday sales season than during the September-October time frame and produced a heck of a lot more sales. Why does no one value them? Geez
I’m hesitant to declare Holiday 2025 as the singular turning point into an agentic, AI-driven retail era — mainly because we simply haven’t yet seen all the data — both the hard numbers and the deeper qualitative insights — needed to make that judgment. What is clear is that AI-based shopping tools, personalized recommendations, and generative-driven discovery have become much more visible in consumer experiences this season than in prior years, and adoption rates among shoppers exceeded many expectations. But a single holiday cycle — even one as important as 2025 — isn’t yet a definitive inflection point in the long arc of retail evolution. Real structural change shows itself over quarters and years, not just in seasonally-peaked bursts. We need to observe sustained behavior change in shopper engagement, conversion lift attributable to AI interventions, and the operational impacts on retailers before calling this a wholesale shift into a new era.
When we look at overall sales and fulfillment metrics — including strong holiday sales and elevated BOPIS (buy-online, pick-up-in-store) performance — there are valuable lessons about consumer expectations and omnichannel orchestration. BOPIS growth underlines that shoppers still value convenience and immediacy, and that AI-enhanced search, intelligent recommendations, and inventory visibility tools help make those hybrid journeys more seamless. What we can celebrate from this season is that AI has moved beyond pilot projects and point innovations into real customer-facing use cases that shaped how many people discovered products, compared options, and completed purchases. But whether that translates into a paradigm shift that endures through 2026 and beyond requires more longitudinal evidence — including return rates, post-holiday retention of AI-driven behaviors, and how the technology impacts key drivers like margin, loyalty, and customer lifetime value.
In short, Holiday 2025 may be an early milestone in the adoption of agentic AI in retail, but it’s too soon to call it the definitive “corner turned.” The season offered encouraging signals that AI is becoming more embedded in the shopping experience, yet retailers and analysts alike must continue to monitor trends over time — not just in peak periods — to determine whether this year truly marks the start of a new era or simply a compelling early chapter in a longer transformation.
Salesforce mislabels retailer-controlled recommendation engines as “agentic shopping.” These are fundamentally different. What they’re measuring is sophisticated manipulation, AI optimizing for retailer revenue through “customized recommendations,” not consumer-controlled agents optimizing for genuine buyer benefit. The 14% return rate exposes the gap; these systems are optimized for conversion, not fit. Real agentic commerce means consumer-controlled agents evaluating options against actual needs, not platform algorithms persuading humans toward higher-margin purchases. Holiday 2025 may mark retailers’ perfecting AI-driven persuasion just as the window closes before consumer-controlled agents make such manipulation obsolete. The battle is real, but it isn’t about whether AI mediates commerce; it’s about who controls the AI and whose interests it serves.
“What they’re measuring is sophisticated manipulation.” Perfect description of how the AI is programmed.
Yep, the return rate blows the cover on this whole conversation. I was a little less even handed in my comments.
Being blunt, I am suspicious of these numbers. While I am sure that they are accurate in their own way, the devil is in the detail.
If we are saying that 20% of sales were influenced in some way by AI – including people using chatbots to generate ideas, or using them to check shipments – then, while they seem a little high, this is feasible. However, this is a very loose use of AI and AI has simply replaced some of the traditional search processes.
If we are saying that 20% of sales/transactions were made directly via AI then this is completely off. We know this because only 22% of holiday sales are made online – so it would suggest that almost every single one of those was driven by AI – which is clearly nonsense.
The other point I would make here is that there is way too much hyping of AI to try and generate PR noise. What we need are realistic and nuanced views of how AI is having an impact (which it is) rather than exaggerated claims.
Wait a minute. Deep breath. I have buckets of skepticism about this whole conversation. First, we are getting all this from Salesforce? Is it data or Salesforce marketing? They say it was a bigger, and more “efficient and intelligent” holiday season. Really? Am I reading this all correctly that growth in returns outpaced growth in sales? How is that possibly more efficient and intelligent? Sounds expensive, reeeaaalllyyy expensive. So now lazy and sloppy customers can get more stuff faster and then return more of it…faster. Doesn’t sound like retailers would view that as cheerful news. This sounds like the ‘free shipping/free returns’ mentality of two decades ago…on steroids. Good for customer acquisition, and really expensive in the long run as it becomes table stakes.
There’s a very natural evolutionary move taking place here that both customers and retailers are embracing. So agentic AI will become table stakes faster than ecomm became table stakes. Calling it a “new era” sounds like all rainbows and unicorns. It’s not that simple. How about a series of articles that talk about success stories and horror stories and ROI. Yeah…ROI. What is all this costing? Who’s paying? Weren’t we all going to be living in the metaverse by now?