# Black Friday 2026: What AI price comparisons need to see

> DHL and PwC show how shoppers use AI to find Black Friday deals. Test whether an assistant can check the exact product, shipping costs and delivery date.

- Source: https://www.agentic-commerce.sh/en/news/black-friday-2026-ki-preisvergleich/
- Author: Stefan Hamann
- Updated: 2026-11-22
- Language: en

## Which Black Friday deal will arrive by Friday?

One shop wants €89 for the black headphones. Another wants €94. The first deal looks better until you add €6.90 for delivery. If you need the headphones before leaving on Saturday, there’s another question. Which shop can actually get them to you by Friday? That’s a perfectly ordinary thing to ask a shopping assistant during Black Friday week.

DHL eCommerce and PwC published research on 19 and 20 November about what shoppers already hand over to AI when looking for deals. What interests me is the work that happens before the customer clicks. A shop that can show the right product, the full price and a delivery date it can stand behind gives an assistant something useful to compare. The traveller buying a suitcase in [Chapter 8 of my book](https://www.agentic-commerce.sh/book/en/08-from-the-offer-to-the-service-that-follows/) needs the same thing. The headphones show why it matters even earlier, when the assistant is still deciding which shops to put in front of the customer.

## How are shoppers using AI to compare prices?

In [DHL eCommerce’s Black Friday research](https://group.dhl.com/en/media-relations/press-releases/2026/nearly-nine-in-ten-shoppers-use-or-plan-to-use-ai-to-find-the-best-price.html), 88% of the Black Friday shoppers surveyed said they had used or planned to use AI to compare prices. Some 47% said they’d used AI chat or a virtual assistant to browse or buy. The findings come from its 2026 E-Commerce Trends Report, which draws on surveys of 29,000 online shoppers and 5,800 commerce businesses. Those two percentages refer to Black Friday shoppers in the survey. One combines usage with plans to use AI. The other combines browsing with buying. The figures leave open how many orders an agent actually placed.

[PwC’s survey](https://www.pwc.de/de/pressemitteilungen/2026/black-friday-2026-deutsche-planen-rekordausgaben.html) asks a different question. Among 2,000 respondents in Germany, 21% already use an AI tool or shopping assistant for research, rising to 28% among people aged 25 to 34. About a quarter reject automated purchases, and another 24% say protecting payment and personal data is a central condition for them. The two surveys use different populations and questions, so the percentages can’t be compared directly. Both help explain why product selection matters now, even when the customer goes on to place the order themselves.

## What does an AI shopping assistant need from the retailer?

First, the assistant has to check whether it’s looking at the same headphones. Are both pairs black and do they both have active noise cancellation? Is one refurbished? Then it needs current stock and the full price including shipping for the exact variant. A different colour, condition or delivery charge can change the answer before anyone reaches checkout.

One in three Black Friday shoppers surveyed by DHL said they were most likely to abandon a purchase while choosing shipping options. If you can deliver by Friday, that’s something the assistant needs to see while it’s comparing offers. Our [guide to delivery promises and tracking](https://www.agentic-commerce.sh/en/guides/delivery-promises-and-tracking-for-shopping-agents/) walks through the postcode, available stock and dispatch cutoff that go into that promise. If the shop only provides an estimate, the assistant should explain how much uncertainty is left.

## A Black Friday test I’d run in a real shop

Pick one heavily promoted variant. Ask an assistant whether it can arrive at a specific postcode before Saturday for €100 or less all in. Put the item in the shop’s cart and compare the answers. Start with the SKU, then check the item price, shipping and delivery date. Try again after the promotion expires or the warehouse misses its last pickup. The answer should change with the offer. If the assistant can’t confirm the date, that’s what it should tell you.

If you take the test all the way to checkout, record where each step ends. The [i22 study of German online stores](https://www.agentic-commerce.sh/en/news/i22-studie-agentic-commerce-deutschland/) shows why that matters. It counted reaching the choice of payment method as the checkout goal and ended all runs before placing an order. I’d use that same level of precision here. Did the assistant find a suitable offer? Did the same product and total reach the cart? Could the customer tell whether it would arrive in time before placing the order?

## Source and further reading

[Read the original announcement ↗](https://group.dhl.com/en/media-relations/press-releases/2026/nearly-nine-in-ten-shoppers-use-or-plan-to-use-ai-to-find-the-best-price.html)

[How shopping agents get a delivery promise ↗](https://www.agentic-commerce.sh/en/guides/delivery-promises-and-tracking-for-shopping-agents/)

[The Agentic Commerce handbook ↗](https://www.agentic-commerce.sh/book/en/)
