Retail's future is tied up in agentic commerce, yes? Well, up to a point, says BestBuy, Wayfair, and Electrolux with welcome notes of caution struck

In retail circles, it’s all about agentic commerce, right? Certainly the data keeps coming in that agentic commerce is top priority. This week Logicbroker, which pitches itself as provider of an Agentic Commerce Orchestration Engine for enterprise retailers, released "The State of Agentic Commerce Adoption" report, based on a survey of more than 600 enterprise e-commerce leaders.

This finds that more than 90% of enterprise leaders polled expect AI agents to influence at least 20% of online orders by 2027, and more than a third believe AI could in fact shape more than half of all transactions. More than half of organizations say they plan to roll out AI-shopping agents within the next six months. 

Some 95% of enterprises reckon to have already deployed at least one AI-driven commerce capability, while 47% say they plan to invest $1 million or more in AI-driven commerce initiatives over the next 12 months.

And hopes are high for what this will deliver in terms of ROI. Nearly half expect returns within the first year, rising to three-quarters of respondents predicting the same within two years.

So, the bandwagon is rolling and there are no signs of the wheels coming off just yet.

Kingfisher

Just this week DIY and home improvement group Kingfisher announced plans to roll out agentic AI shopping and commerce capabilities based on Google Cloud’s Vertex AI across its B&Q, Castorama, and Brico Depot France brands in Europe following what the retailer calls “meaningful results” from pilot trials at B&Q.  Benefits expected include:

• Vertex AI Search for Commerce enabling a move from rigid keywords to intuitive, conversational discovery, better aligning with modern shopper behavior to deliver faster, more precise results.

• AI-driven shopping through agentic commerce with product and data catalogues unlocked to build pro-active Al shopping agents to assist customers plan complex home improvement projects, generate smart shopping lists, and execute seamless purchases.

With e-commerce  already accounting for more than 20% of Kingfisher's Group sales, CEO Thierry Garnier is looking to this expanded Google relationship to deliver the next phase:

Through this partnership with Google Cloud, we are enabling our customers to search for and buy home improvement products with AI, delivering a fully personalised and easy shopping experience. These investments put Kingfisher at the forefront of AI-powered shopping, delivering meaningful innovation as part of our expanding digital ecosystem, and helping us to meet rapidly evolving customer needs."

Safe to assume that Garnier is an agentic fan. 

Others are perhaps being a little more hesitant about their own strategies here

BestBuy

At BestBuy, CEO Corie Barry says the firm has “taken our time working out way into agentic commerce”. Why? He explains:

You do need to go in with a plan around what ubiquitously needs to be available in agentic commerce versus what is very unique to your brand, your experience, your data. I think it's going to be an ‘and’ world. You're going to need to be present in agentic commerce. In our case, in working with chat, we have really good data that could help [agents].

But the tech is still nascent, he cautions:

Especially in consumer electronics, you would ask [AI] questions and there was a lot of wrong answers coming back because there are so many models in the world. They can be dated before you know it. There's so much data. So we're trying to go in with a plan around [that fact that] we need to be present, but we also need to say, ‘Look, if you want this installed, if you want us to make for an easy upgrade, then you got to come back to the site’.

A “real solid plan” is needed around what is a retailer’s data, he advises:

You also need to think about your site and your content, because not only are you out in an agentic shopping engine, but their bots are coming and scraping your site for information. So your site needs to now be geared not just toward people, but you may actually have hidden pages of data, not for customers, simply for bots that they can scrape so they're more knowledgeable, which, again, you might feed a little bit of that in the Agentic answer and then you say, come back to the site if you want more expertise on it.

So does all that suggest that BestBuy will continue to proceed at a slow pace down this route? Barry argues:

We'll see. It's early....No-one’s figured it out and everyone's testing a bit.

Wayfair

Over at home decor firm e-commerce specialist Wayfair, founder and CEO Niraj Shah is also quick to set agentic expectations appropriately:

I think this notion that LLMs and AI agents can do everything is not the case, but they can do a lot. The companies that can optimize themselves in that world, the same way in the days of search and social media that what we did with Meta, what we did with Google, we do with Pinterest, advantaged us, I think we can do the same thing.

There have been lessons learned among providers, he suggests, pointing to Google:

Their efforts like Google Shopping, which tried to have transactions finished there, ultimately found that consumers want to go lower funnel. But basically what OpenAI the other day said, ‘Well, that’s probably the bulk of what people want to do’. Google already knows this lesson from years of balancing both sides of it.

And it depends what the operating model and target market of retailers are, he notes:

I think the reality is this - the LLMs, they're very helpful for customers to kind of get the landscape of something they don't understand, but it can only take it so far. If it's something the customer can be very articulate about, like, ‘I use Dove soap bars, I use Mrs. Meyer's hand soap, I use Seventh Generation dishwasher tabs etc and I want to replenish that stuff’, [agents] could execute those orders for me. That is a very straightforward exercise.

If it's something where there's a discovery, inspiration, emotion, aesthetic, whether you're talking about home or you're talking about fashion, or you're talking about beauty, I don't think the agent is going to help you going from the beginning to the end and just saying, ‘OK, you want some more lipstick’,  unless you want to re-order the same exact one.

But it won’t choose a new sofa for you, he adds:

What it will do -- and this is where we can optimize for this - is it will say, ‘Wayfair is a great solution for you,’ but there's a lot of choices you [as a human] get to make around the item. You're going to want to understand other items. We have information about what you purchased. We know what styles you like. We know what you have in your house. We can create inspirational imagery with AI that's very specific to you. We can also present you with a lot of options around delivery, assembly, taking away, packaging.

He concludes:

We play in a category that's very bespoke and unique. Basically, home, fashion and automobiles are the only three product categories that customers have so much curiosity and fanaticism about that they'll spend money to basically buy magazines, basically consume media, for the enjoyment of knowing what else is coming, what the trends are, what's available....There’s an emotive content to it that is very hard to capture through Artificial Intelligence.

Electrolux

So is agentic commerce and agentic shopping really the inevitable future for retailers that the likes of Google and OpenAI would have everyone believe as they sign up on brand after another? Yes, says Marienza Benedetti, D2C Ecom Personalization and Growth Manager, at consumer goods firm Electrolux:

There's a future that is knocking at our door, a future that is already happening, and it's the future of agentic commerce.

But there are questions that organizations like hers need to ask themselves first:

The main question that we are asking ourselves at the moment is, ‘For AI systems that will search, that will compare, and that will transact on behalf of the consumers, who will we be optimizing the experience for in this new landscape? Will it be the human or will it be the agent?’. While we are looking, we are still actually looking for the answers. 

One thing that came to us - and felt like a kind of a paradox - is that the more AI will drive the experience, the more human-centric it must become. Why? Because AI will [have to] somehow connect human needs to product features better than we do. It will help us directly optimize not just for clicks, but for lifetime value and for trust. It will surely be not as aggressive and promotional as we are. So it will be more relevant, more contextual, and more respectful. It will be basically empathy at scale.

Or that’s the theory, at any rate. Watch this space.

My take

The direction of travel is clearly set and there will be more and more agentic enthusiasts, like Kingfisher, racing to sign on to this latest e-commerce gambit. But the candor and cautionary notes struck by BestBuy, Wayfair, and Electrolux seem to me to be valuable counter-weights to a hype cycle that might otherwise be assumed to be out of control.