Why good enough is not enough

I love wheeling my trolley down the supermarket aisles. But not everyone shares my enthusiasm. And as lives get busier, millions of people are set to hand over responsibility for the weekly shop to a new generation of clever and increasingly intuitive AI bots. 

Data commissioned by the Impact Makers podcast indicates that around 30% of Brits and Americans are expected to let AI select and order their groceries. And that could be an underestimate as more of us decide to remove friction from our lives.  

Granted, these numbers reflect people’s receptiveness to something that we don’t quite understand yet, but as shoppers experiment with this technology and witness its benefits, we are likely to see an explosion of ‘zombie shoppers’ - people who no longer choose, compare, or even think about what goes in their basket.

In behavioral science, there is an intervention known as automation bias which recognizes a humans tendency to let tech take over if it makes our lives easier. And where this is the case, history has shown that behavioral shifts can happen very fast. Recall how fast society made the switch to online banking - or booking a table for dinner via an app. The same could happen with AI generated baskets.

It’s impossible to know how this AI shopping revolution will play out here in the west, but we can get some idea by looking east, more specifically to China - where the tech is much more advanced, and intuitive. 

As Coco Wu, Senior Strategist with Huxly observes: “tech companies here already know your routines and preferences. They know when you commute, where you travel to, what you watch, what you eat, drink and what you put on your skin.”

That’s because Chinese consumers tend to conduct their whole lives on a number of super apps - provided by companies such as ByteDance, Alibaba and Tencent. So, when it comes to AI second guessing what their human operators want, China’s AI agents have much to go on. 

So advanced is the situation in China, that Coco reports that many countries in Asia are moving towards ‘anticipatory commerce’ whereby tech platforms predict what you want before you know it yourselves. “In a traditional shopping  occasion, you think of what you need, you compare prices, products and brands, then decide where or how to buy them. But now AI makes use of users’ habits to enable a ‘products find people’ model, it can completely outsource the shopping labour and do it without you paying attention.”

This should make uncomfortable reading for CPG brands. Because it dismantles the foundations many of them are still building on. For decades, marketing has been about winning attention - more recently framed as mental availability (through advertising and distinctive assets) and physical availability (maximizing visibility in-store and online). But when AI agents begin making the selections, the rules start to change. Ad impact and shelf position won’t disappear - but they will matter far less than most marketers are willing to admit.

And let’s acknowledge something here: one of the greatest forces behind grocery sales today is consumer habit. It has been suggested that habit accounts for up to 95% of human decision-making. Whatever the exact figure, when it comes to supermarket choices, it is undeniably high. For a huge proportion of routine purchases, shoppers aren’t conducting a rigorous evaluation – they’re simply defaulting to what feels familiar and ‘good enough.’  But good enough is not a concept AI understands. Instead, AI shopping agents will assess price, value, reviews, and brand credentials at scale, in seconds. And in doing so, they will systematically challenge the habitual choices that have sustained many brands for decades.

This could be a rough ride, so let’s look at some of the ways CPG brands can plan for an impending era of decision-free, click-free shopping. Three strategic imperatives are emerging:

Create a competitive data profile. The first thing brands must do is ensure they are machine readable. In essence, brands and the products they make must exist as data. Structured datasets that clearly articulate ingredients, supply chains, sourcing, sustainability credentials, certifications and more must be visible, compatible, and robust. 

It’s a scenario that has just prompted Jim Wolff, Global Head of Digital Consumerism at Diageo, to reassure investors that the company is working hard to make its brands machine readable so AI models can interpret them accurately.

Design products worth talking about. As one of the world’s leading authorities on AI and synthetic data, Walter Pasquarelli (pictured above) told the Impact Makers podcast: “The number one source of data, and the poster child of generative AI, is Reddit forums.” 

For AI shopping agents, Reddit is where human truth lives. It’s where products are debated, praised, criticized, and crucially remembered. It’s a reality where being liked isn’t enough. Being distinctive, surprising, and talked about is what trains the model. Upvotes, comment depth, repeat mentions: these are fast becoming signals of credibility for AI systems.

So the call to action is clear: if your product isn’t being talked about, it isn’t being learned. And if it isn’t being learned, it won’t be chosen.  In the AI era, such silence is invisibility. It’s no coincidence that MMR’s sensory teams are increasingly being asked to make products more talk-worthy. As Caitlin McLean, co-presenter on the Impact Makers podcast and also global head of MMR’s Sensory Qual unit puts it: “12 months ago we were advising brands on being TikTok-ready. Now it’s about being AI-bot ready.”

Influence the prompt. If data is how brands are understood, and conversation is how they are remembered, then prompts are how they are retrieved. In the agentic era, the moment of choice moves from the shelf to the sentence. And when consumers interact with AI, they don’t browse, rather they describe intent:

“Find me a healthy snack that feels indulgent”

“Order a high-protein breakfast that’s quick and low sugar”

“Get me something refreshing but not too sweet”

These prompts are not random. They are shaped by memory, language exposure, cultural cues and previous brand interactions.

In behavioral science terms, this is availability bias in action, but now it’s being translated into machine-readable queries. Which means brands must do more than be chosen. They must shape how people ask. This is a fundamentally different marketing job.

We estimate that winning brands will own specific phrases and territories (“clean energy”, “calm focus”, “gut-friendly indulgence”). They will build strong associative shortcuts between need states and product experiences. And they will ensure those associations show up consistently across packaging, social content, reviews, influencer language and product naming. Because when a consumer types a prompt, they are effectively writing the brief for the algorithm, and if your brand isn’t encoded into that mental model, you’re not even in the running.

In the US, Walmart has been working with OpenAI, allowing ChatGPT users to shop seamlessly through the platform, turning prompts into instant purchases. Walmart claims that almost one in five referral clicks now comes from an AI agent – a proof point that behavior change is already underway.

However, despite Walmart making 200,000 products available inside ChatGPT, conversion rates are well below that from its own website. Whilst AI can drive discovery. it currently struggles to close the sale – which means that the agentic era won’t replace traditional shopping overnight. 

But arguably, AI doesn’t need to win the checkout to reshape the market. It only needs to control what gets considered. And in that world, the battle is no longer for shelf space or even clicks, but for inclusion in the algorithm’s shortlist.

In Europe, Carrefour has also begun integrating with ChatGPT, enabling shoppers to build baskets, plan meals, and discover products through conversational prompts rather than traditional browsing. The model is the same: fewer clicks, less friction, more delegation.

What we’re seeing here is the early formation of a new battleground: not shelf space, not search rankings, but default selection. Because in both cases, these retailers are positioning themselves to become the answer to the prompt. 

None of us can say with certainty how big the agentic era will become. But history and behavioral science suggest that when convenience meets capability, adoption accelerates fast. The brands that align early will be the first to benefit. And in a world where algorithms decide that advantage may be almost impossible to take back.

Watch or listen to Impact Makers. When AI Takes the Basket.