Shopping Now Begins Inside the Assistant, Not the Site

What Happened

The announcement contains four developments that only make sense together.

Direct integration into ChatGPT

The retailer became the first food retailer to integrate directly into the platform. Users select the brand from the applications panel and complete product discovery inside the conversation.

Natural-language commands

The assistant handles need statements rather than product names. A request for a protein-focused breakfast basket returns a menu suggestion, an ingredient list added straight to the basket, and cooking instructions.

Personalisation from purchase history

Past shopping behaviour is analysed, results are ordered by price and campaign advantage, and items nearing depletion are flagged to prompt basket renewal.

AI runs behind the counter too

The company applies AI in ready-basket algorithms, robotic picking and logistics optimisation. The customer-facing assistant is the visible end of a system already in place.

What the Numbers Mean

The 60 per cent basket-addition rate is the actual news here.

Recommendation outperforms search

On a conventional e-commerce site a small fraction of visitors reach purchase. Here three in five searching users accept the recommendation. The difference lies in how the user arrives: someone who types a request to an assistant is already ready to buy. This should not be compared with general visitor conversion.

The decision path shortens

The traditional route runs through search, comparison, site entry, product discovery and basket. All of those steps collapse into one screen. Every step removed is a customer not lost.

The storefront relocates

The long-running argument was whether the brand site or the marketplace mattered more. A third venue has now opened, and it has no shelf, no ranking and no banner. Only an answer.

Early entry is worth more than it looks

Few brands occupy this channel. Entering while competition is sparse means being established when it fills — an advantage that cannot be purchased later.

Who This Affects, and How

The impact tracks product data quality and where purchases are completed.

Those who gain

Brands with complete product information, current pricing and machine-readable catalogues. The entry condition here is data quality rather than advertising budget. Categories with repeat purchase — groceries, cleaning, consumables — fit naturally.

Those who lose

Brands with scattered product data, marketing-language descriptions and prices that go stale. The assistant cannot read such a catalogue and does not recommend what it cannot read. Absence here is not poor ranking; it is non-existence.

Those largely unaffected

Businesses selling exclusively to corporate buyers or through tenders sit outside this channel today. The same assistant logic is arriving in corporate procurement, where today’s preparation will apply.

The indirect chain

A large retailer enters the assistant channel, consumer habit follows, a smaller brand loses traffic because it is absent there, and tries to compensate with advertising at rising cost. Missing a new channel makes the old one more expensive.

What to Do About It

All four can be done without waiting for an integration, and all four help present-day sales too.

Complete your product data

Name, brand, size, contents, usage and price fields should not be empty. Assistants do not infer missing fields; they skip the product. Starting with the twenty best-selling items makes this a week’s work.

Write descriptions in need language

Phrases such as unrivalled quality match no query. Users type gluten-free breakfast or quick dinner for two. Descriptions should answer those sentences.

Keep price and stock live

When assistants display outdated pricing, the responsibility remains with the brand. Price and availability flowing from a single source without delay is the entry condition for this channel.

Make reordering effortless

Repeating a previous order is among the most-used functions here. If your own site has no one-step reorder flow, you are complicating your easiest sale.

The Digital Side

This development starts a race decided by data discipline rather than media budget.

Assistants read information, not advertising

Most recommendation engines carry no purchasable placement. Visibility comes from clear product information, question-and-answer structure and readable pages — one of the few arenas where a small brand competes with a large budget.

The same preparation pays twice

Data work done for assistants also improves visibility in search. The return on this investment is not confined to one channel.

Do not surrender your own channel

Sales through an assistant are valuable, but the customer relationship is not built there. Contact permission, repeat sales and customer data should still accumulate in your own channel. Holding that balance is a core topic in e-commerce consulting work.

Channels cannot be managed unmeasured

Where orders are not tracked by source, a new channel’s contribution stays invisible. Channel-level measurement moves the investment decision from instinct to evidence.

A Solid Digital Foundation

There is a technical precondition for appearing in this channel, and most businesses have never checked it.

Product pages must stand alone

Where part of the information sits in an image and part behind a collapsed tab, machines cannot read it. Size, contents and usage need to exist as text on the page.

Speed is a threshold here too

A slow-responding server causes problems for automated access as well. Load time is a limit not only for human visitors but for the systems reading you.

Preparation comes before the channel

Which assistant prevails is not yet clear, but all of them want the same thing: clean, current, readable product data. A business that builds that foundation is ready whichever channel emerges. How it is done is set out in our approach to digital consulting.

Frequently Asked Questions