AI is beginning to change the way people shop online.

We are moving towards a world where instead of visiting multiple websites, comparing products and deciding where to buy, a customer could simply tell an AI:

“Find me this product for under £100 and make sure I can have it tomorrow.”

The AI does the searching and comparing for them. In some cases it can already complete the purchase as well, and that is likely to grow.

For retailers, that creates an interesting question.

If an AI is deciding where to buy, is your business ready to be chosen?

My first question to any retailer would be much simpler:

Are your systems connected?

Because before worrying about AI shopping, retailers need to make sure the infrastructure underneath their eCommerce operation works properly.

AI is only as good as the information it can see

Imagine an AI shopping agent finds your product. Your website says there are 12 in stock and delivery is available tomorrow. The AI recommends it to the customer based on that information. The customer places the order.

But your website has not been properly synchronised with your ERP or warehouse and you actually have none available.

Who does the customer blame?

Not the ERP.

Not the integration.

And probably not the AI.

They blame the retailer.

As far as the customer is concerned, you promised something and failed to deliver it. That creates a negative customer experience before the order has even left the building.

This is why I think accurate, connected data is going to become increasingly important as AI shopping develops.

“Our stock is reasonably accurate” might not be good enough

One of the questions I would ask a retailer is:

What happens if you suddenly receive a surge of orders?

You might be perfectly capable of keeping your website and ERP reasonably aligned when you receive 20 orders. What happens when you receive 500? If stock levels are being updated manually or only periodically, they can become inaccurate very quickly.

Now imagine an AI shopping service suddenly starts recommending one of your products heavily. Hundreds of orders arrive within a few hours. If your website, ERP and warehouse aren’t properly connected, your systems can quickly become the bottleneck.

And stock is only part of it.

What happens after the customer buys? Is fulfilment information automatically flowing back to your website? Can the customer see that their order has been dispatched? Can they access tracking information?

If those processes still depend on somebody manually moving information between systems, increased order volume magnifies the problem.

What a connected order journey should look like

For a properly integrated retailer, the journey can be remarkably fast.

1. An order is placed on the website.
2. It automatically enters the ERP.
3. A delivery note is created for the warehouse.
4. The warehouse receives the information required to pick and pack the order.
5. A courier label can be generated.
6. The parcel is dispatched.
7. Tracking information flows back to the website and customer.

The digital part of that entire process can happen within minutes. The part that genuinely takes time is physically picking and packing the product.

Without integration, the same process can take hours, sometimes the best part of a day. And that assumes nobody makes a mistake while manually entering the information.

Speed already wins customers

None of this is actually new because of AI. Customers already value speed.

If two retailers sell exactly the same product at roughly the same price, but one can reliably get it to me tomorrow and the other takes several days, speed becomes a major part of my buying decision.

Amazon has demonstrated this on an enormous scale. Customers have become accustomed to placing an order and expecting it very quickly. AI could simply make that competition more visible.

Today, a customer might visit several websites themselves. They may already know a particular retailer. They might like the brand or prefer its website.

But imagine instead they tell an AI:

“Buy me this product. I need it tomorrow.”

There is potentially much less human relationship involved in that transaction. The AI is doing the comparison. That means the information available about your products, availability, pricing, delivery and reputation could become increasingly important.

I think about it a little like SEO

For years, businesses have invested in SEO because being visible when someone searches can determine whether you even get considered.

I think AI shopping could create a similar optimisation challenge. But instead of thinking only about keywords and website content, retailers may also need to think about the quality of the operational information behind the buying experience.

• Is the stock accurate?
• Is the product information complete?
• Is the pricing correct?
• Can the delivery promise actually be met?
• Is the returns information clear?
• What does the retailer’s reputation look like?
• Are customers consistently receiving what they were promised?

Different AI shopping systems will inevitably use different criteria, so nobody can say exactly how every AI agent will decide which retailer to recommend.

But there is a simple principle here.

The more accurate and reliable the information about your business is, the stronger the information available when an AI or a customer is deciding whether to buy from you.

There is a feedback loop

This is where integration stops being purely a back-office IT project.

Accurate systems create more reliable promises. Reliable promises create better fulfilment. Better fulfilment creates better customer experiences. Better customer experiences create repeat customers, positive reviews and recommendations.

Those things strengthen the reputation of the business. And that reputation can then influence future customers, and potentially the AI systems helping those customers make purchasing decisions.

So the infrastructure sitting behind your website can ultimately affect whether you win the next order.

You don’t need to automate everything tomorrow

If I was advising a retailer whose systems were largely disconnected, I wouldn’t suggest trying to integrate absolutely everything at once. I would start with three areas.

First: orders into the ERP. Get rid of the manual entry and make sure orders can start moving through the business immediately.

Second: stock back to the website. Customers need accurate availability, particularly when order volumes increase.

Third: tracking information back to the website and customer. Close the loop so the customer knows what is happening after they have placed their order.

Other areas can follow. Courier labels can still be created manually initially. Product information and descriptions can potentially remain manual if the business doesn’t introduce new products frequently.

You don’t need perfection on day one. You need to remove the biggest operational bottlenecks first.

We have seen what happens when this is done at scale

One project we worked on involved a business processing around 500 to 1,000 online orders every day. Before the integration, processing those orders could consume a significant part of the day. We automated the order flow so transactions were processed continuously throughout the day instead.

But we didn’t stop with orders. Stock and pricing were integrated. The business could manage special pricing for particular customers and at particular times of the year. Products and variants were integrated, including product images. Information from the warehouse management system flowed through the ERP and into the website.

The result was not simply an automated website. It became a connected eCommerce operation.

And once we had got the first site working correctly, the same approach could be replicated. Within a year, the integration had been rolled out across another six sites.

That is where integration becomes particularly valuable. You are no longer solving one manual process. You are building infrastructure the business can grow on.

We have seen this type of preparedness matter before

There is another reason I think businesses should consider this now rather than waiting until AI shopping becomes mainstream.

We saw something similar during COVID. When physical stores suddenly had to close, businesses were forced to rely heavily on their online operations. Companies that already had connected eCommerce infrastructure were in a very different position from those suddenly trying to build it while their traditional sales channels were disrupted.

Depending on the industry, some businesses were able to continue operating online and, in some cases, handle significantly increased online demand. They hadn’t necessarily built those integrations because they predicted a pandemic. They had built them because automation already made commercial sense. When circumstances changed, they were ready.

I think there is a lesson there for AI shopping.

Don’t integrate because you are afraid of AI

There is a danger of businesses spending money on technology simply because everyone is talking about AI. I don’t think that is the right reason.

Connect your systems because it benefits the business today.

• It reduces manual work.
• It saves time.
• It reduces opportunities for error.
• It improves stock accuracy.
• It speeds up order processing.
• It improves the customer journey.
• It allows the business to handle greater volumes without increasing operational effort at the same rate.

Those benefits exist whether AI shopping becomes mainstream next year or takes considerably longer.

But if AI does become an important route through which customers discover and purchase products, there is another advantage. Your systems won’t have been connected yesterday. They will have been operating together for years. Your processes will have been tested. Your fulfilment performance will have history. Your customers will have had years to generate reviews and feedback. And your business will already have the infrastructure needed to handle increased demand.

So rather than asking:

“What do we need to build for AI shopping?”

I think retailers should ask a different question:

“If AI shopping arrived tomorrow and suddenly sent us 1,000 orders, could our business handle them?”

If the answer is no, there is probably work worth doing today, regardless of when AI shopping arrives.

If you’re wondering how ready your own systems are for a surge in orders, that’s exactly the kind of conversation we have with clients every week. Get in touch and we’ll talk it through.