What AI is not going to fix for you in retail

It feels like everyone, everywhere is talking about AI. What it will change, how fast it will move, and how it might solve some of the biggest challenges businesses face today. It’s being positioned as the answer to everything from inefficiency to innovation.

There is far less conversation about the other side of that coin: what AI is not going to fix for us. That matters just as much.

AI works best when the foundations are already strong. It depends on clear processes and good quality information, and it still needs people who know how to ask the right questions and apply its outputs in a meaningful way. I always say to clients that the right way to tackle any operational issue is 'people, process, systems - in that order'. AI is no exception.

There is a growing tendency to treat it as a shortcut. However, if it's not carefully thought through with a clear objective, measurable outputs and someone accountable for it, it will not perform. It’s far more useful to think of AI as an additional member of the team who needs to be communicated with, given clear objectives, and managed.

It’s also worth being clear about what AI cannot do. It cannot replace strategic thinking, and it cannot generate genuinely strong solutions without being guided by experience and insight. The technology can support decision making, but it cannot take responsibility for it.

Crucially, it’s important to remember that AI comes with limitations. It has no capacity for genuine empathy, very little awareness of when it might be wrong and it rarely admits uncertainty. Essentially, it doesn't say 'I don't know' often enough!

Why customer trust cannot be automated

Customer service is not just about dealing with customers in the most efficient way and resolving their individual issues. It’s about building a connection between a customer and your brand, and creating a point of difference in what can often be a homogenous market.

The impact of this is clear. 86 Percent of buyers are willing to pay more for a great customer experience, yet 32 percent will leave a brand they love after just one bad experience. (Source: 117 Customer Service Statistics You Need to Know in 2026, Desk365.io)

For luxury and premium brands in particular, expectations are even higher. Customers are not just looking to have a problem resolved, they are looking to be wowed.

The challenge is that the return on good service does not always sit neatly in one place. It shows up on different P&L lines across the business - from customer retention to purchase frequency and organic acquisition through recommendation.

What is needed is real buy-in at C level to investing in service as a driver of growth. Without that, it’s very easy for service to become something that is optimised for efficiency rather than experience, particularly as businesses scale and face increasing volumes of customer queries. This is where I see many brands struggle. As they scale-up, the focus shifts towards efficiency and speed. Without a clear view of the bigger picture, that can come at the expense of the very thing that differentiates them in the first place.

What still requires emotional intelligence, ownership and real time judgement

Before I begin work with any client, I read their customer reviews. The most common issue is not that things go wrong, but how those issues are handled.

Can they be resolved? Often this comes down to limitations in operational processes that need to be addressed. But more importantly, was the issue actually resolved? In negative reviews, the customer often does not feel it was, whereas in positive reviews they do.

The difference is usually the level of empowerment of the person they spoke to, and whether they had the tools and information to help.Being able to create a connection, show genuine empathy, and find the right resolution will always be the foundation of a great service experience. AI can only support that final part.

Crucially, a recent HubSpot survey found that 91 percent of unhappy customers will leave without complaining.

Why customer recovery still starts with picking up the phone

What all customers want is a fast resolution and a genuine apology. 60 Percent of consumers say that quick resolution times are critical to their experience with a company. AI can help make that process faster, smoother and more consistent. But when it’s not working, you need a genuine human connection to bring that customer back. In practice, that is often a real apology accompanied by a resolution the customer agrees with.

Too much time spent by both parties going back and forth over email or text often adds nothing but frustration and can ruin what should be a moment of connection. In a world where customers are increasingly used to automation and AI responses, this is a real opportunity to stand out.

It’s also worth noting that it does not have to be the phone, it is the individualised approach that is key. Using the channel that suits the customer best is part of that, whether that is WhatsApp, phone, email or social media. Different channels work for different customers at different times. Research shows that 89 percent of consumers want to connect with brands across multiple channels and expect a seamless transition between them.

When getting a customer out of the AI loop is the only solution

Customer frustration, a lack of suitable resolution and an elongated communication thread are all signs that something is not working. Being able to recognise these signals and pull that customer out of the loop is vital to retaining them.

59 Percent of customers will abandon a brand after a single poor experience. Particularly for eCommerce brands, where the cost of acquiring new customers is high, retaining existing ones is critical to profitability. It can take up to 12 positive experiences to make up for one unresolved negative experience.

The risks of hiding behind automations instead of resolving issues directly

For one scale-up client I worked with, they were responding quickly to issues and, on the face of it, queries were being resolved. However, when we dug into the data those customers were not coming back and Trustpilot reviews were poor. The AI processes being used were very black and white. They worked well to close queries, but when there was a genuine issue, they fell short.

The business also had a requirement for a high volume of technical questions to be answered, which was difficult to upskill quickly. Here, AI worked well. It could access and process the right data efficiently, answer questions directly and even foresee issues such as delayed or lost orders. Where it did not work was in situations that required a more personalised approach and a genuine apology.

That is where the combination of agent and AI became key. An agent was far better able to rebuild the relationship with a disappointed customer, and also to cross sell or upsell while retaining them. Supported by AI, they had better data to hand which enabled them to offer a more tailored service that felt personal. At the same time, customers with simpler but still technical queries were serviced in near real time, which suited the experience they were looking for.

How to get the best out of AI

AI is a powerful tool, but it’s not a shortcut. The businesses that get the most out of it will be the ones that get the basics right first and stay clear on what still needs a human touch. It still comes back to people, process and systems, in that order.