September Retail Roundup: A Month of Contradictions

AI is doing the shopping, people are making the difference, and consumers are keeping everyone guessing.

The robots are shopping, the customers are watching their pennies, and somewhere in the middle, retail is still trying to work out what to do with its people.

Ah, September. That awkward month where summer is over all too quickly, it's starting to get dark too early, Christmas appears on the shelves and everyone's attention turns nervously towards the Golden Quarter.

Looking back at the retail news this month, there's a lot going on beneath the headlines. But what's particularly interesting isn't necessarily the individual stories. It's the contradictions.

Retailers are investing in technology to remove human effort from shopping, yet human interaction is driving luxury sales. Businesses are restructuring around automation, yet shop workers are facing increasing pressure. And consumers, apparently, are spending more despite being rather worried about how much money they have.

Perhaps that's what happens when an industry is in the middle of a transformation. Progress rarely moves in a straight line.

So. Let's look.

AI in retail has started doing the shopping. And it's getting rather crowded.

Oh yes, what a time to be alive.

On 9 September, Instacart launched Clementine, an AI assistant that can turn a recipe, conversation or shopping list into a ready-to-buy grocery basket.

Personally, I think this is great. The mental load of working out what to eat, finding the ingredients, comparing prices and then actually doing the online shop is quite frankly exhausting. I'm more than happy for AI to take on some of these mundane parts of life.

On the very same day, Target-owned Shipt launched Ask Shipt, offering customers similar capabilities.

Quite the coincidence. Or perhaps not.

DoorDash and others have been moving in the same direction, and by the time Groceryshop 2026 rolled around in Las Vegas, AI was firmly at the centre of the conversation.

Here in the UK, John Lewis revealed that the proportion of product searches originating from AI agents had grown from 0.3% to 2.5% in a year, as Reuters reported.

Still relatively small, yes. But the direction is interesting.

We've spent decades designing retail experiences around the assumption that customers will search, browse, compare and eventually buy. Now we're developing technology that can do much of that thinking for them.

And that changes considerably more than the shopping interface.

A retailer might spend millions building brand recognition, investing in visual merchandising and carefully planning promotions. But what happens when someone simply asks an AI assistant to organise five dinners for a family of four within a £60 budget?

Which products get recommended? Which brands get overlooked? And who ultimately decides what goes into the basket?

The commercial results are beginning to make this interesting too.

Albertsons reported that shoppers using its conversational AI tools had average order values around 10% higher, rising to 26% among those using more comprehensive assistants.

Encouraging figures, although that doesn't necessarily mean AI caused the increase. Nor does a bigger basket automatically mean greater profitability.

What I find more significant is the change in where influence sits.

The battle for the customer may increasingly be won before they ever reach a retailer's website.

And if that happens, good products and clever marketing won't be enough. Retailers will need accurate product data, dependable stock availability, sensible pricing and the operational ability to fulfil whatever an AI assistant has promised.

This is where I think Order Management Systems (OMS) are becoming particularly important. ERP has its place, of course, but managing complex, real-time availability and fulfilment decisions across multiple channels requires something more.

Because an intelligent recommendation for something you can't actually deliver isn't particularly intelligent, is it?

Retail jobs are disappearing. But is anyone asking what we're losing?

September also brought some rather uncomfortable news about retail employment.

Research from Q5, covering more than 160 UK retail and consumer businesses, found that demand for retail roles had fallen by 12% since 2023.

The biggest declines weren't necessarily where you might expect. Demand for centralised customer and digital roles was down 25%, people roles 23%, support roles 21% and finance roles 20%.

The research points towards rising employment costs, weak consumer confidence and AI as businesses rethink how they're organised.

Now, I understand the commercial argument. Retail operates on tight margins, and businesses are under enormous pressure to become more efficient.

And technology can absolutely help.

But there's something that bothers me about the way we discuss these changes.

We're very good at calculating the cost of employing people. I'm less convinced we're equally good at measuring the value they create.

I've written before about the importance of everyday human interactions in retail, particularly those conversations that happen in local shops. They're easy to dismiss as something nice to have until they disappear.

And this isn't just about the checkout assistant.

It's about experienced merchandisers who understand their customers, store managers who know their communities and customer service teams who recognise a problem before it escalates.

Perhaps automation will free employees to do more interesting, valuable work. I'd genuinely like to see that happen.

But removing work and creating better work are two different things. One doesn't automatically lead to the other.

Meanwhile, the people still working on the shop floor are dealing with a different set of challenges.

In September, the Retail Trust announced additional support for employees facing abuse, including guidance for workers being filmed without consent or followed outside stores. Nearly one in five retail workers were also reported to be struggling financially.

Read that again. People being followed home because they work in a shop.

We're talking about transforming the retail experience while some of the people delivering it are struggling to feel safe at work.

I think that deserves considerably more attention than another announcement about operational efficiency.

Consumers are spending more. Except they're not. Well, sort of.

US retail sales for September rose 0.8% month on month, beating expectations.

Excellent news, surely?

Well, that depends on what you look at.

Much of the increase came from petrol station sales, which rose 6.4%. Remove petrol stations and sales increased just 0.2%.

Meanwhile, UK retailers had their own difficulties.

BDO reported that discretionary retail sales fell 0.3% on a like-for-like basis, making it the weakest September for those categories in seven years, excluding the pandemic. Online sales were down 1.1%.

Different markets and different measures, of course. But there's a common thread.

Consumers haven't stopped spending. They still need to eat, travel, replace things and occasionally treat themselves.

What's changing is how they prioritise those purchases.

And I think we're sometimes too quick to describe this as consumers becoming more price-sensitive.

It's more complicated than that.

Someone might buy the cheapest supermarket essentials available and still spend £200 on a handbag. They might cancel one subscription while paying extra for convenience somewhere else. They might happily buy second-hand clothing but insist on premium beauty products.

That's not necessarily irrational behaviour. It's prioritisation.

Retailers are competing for a share of increasingly deliberate spending, and being cheaper isn't always the answer.

Understanding what customers genuinely value is becoming rather more important than simply assuming everyone wants the lowest price.

AI is removing people from shopping. TikTok is putting them back.

Now here's a story I particularly enjoyed.

In September, TikTok Shop revealed that 94% of its US luxury resale revenue so far that year had come through livestream shopping.

Luxury resale gross merchandise value on the platform was also up 400% year on year.

Pause on that for a moment.

At one end of retail, we're developing technology to remove interaction from the purchasing process. At the other, people are watching other people sell second-hand luxury handbags on live video.

And it's working. I don't think that's a contradiction at all. In fact, I think it tells us something important about why people buy.

For routine purchases, removing effort can be wonderful. If technology can organise the weekly grocery shop in a few minutes, brilliant. But buying a pre-owned designer handbag is quite different.

Customers want to see the product, understand its condition, ask questions and feel confident they're getting what they've paid for. Particularly when authenticity matters enormously.

A livestream offers something a static product page struggles to replicate. The opportunity to engage with another person, see their reactions and develop some sense of trust.

Of course, a friendly face isn't a substitute for proper authentication or consumer protection. But it can make the experience feel more reassuring.

And I think that's the point.

The future of retail isn't necessarily less human. It's knowing when human interaction adds value and when it simply adds friction.

That distinction is going to become increasingly important.

Retail leadership keeps changing. The difficult part comes next.

Finally, September saw Heidi O'Neill formally take over as CEO of Lululemon, following her appointment earlier in the year.

Another experienced female retail leader stepping into a business with significant expectations around growth, performance and brand relevance.

Regular readers will know I've been following women's leadership in retail closely.

And I'll keep following it, because I think we still spend far too much time celebrating appointments and not nearly enough examining the conditions in which leaders are expected to succeed.

At the same time, we're seeing new leadership responsibilities emerge around AI. Chief AI Officers, transformation leaders and increasingly complex technology governance structures.

Some of these roles will undoubtedly be valuable.

But I do wonder whether we're occasionally confusing creating accountability with creating another executive position.

Retailers don't simply need someone to own AI. They need leadership teams capable of understanding how technology changes decisions, operations, employee responsibilities and customer relationships.

A new job title doesn't resolve those questions. Neither does another restructure.

The real test isn't how impressive the leadership announcement sounds. It's whether the business becomes better at delivering what it promises.

And that takes rather more than a press release.

So what did September really tell us?

September was about the consequences of change.

AI is moving beyond experimentation and into everyday shopping. Consumers are making more deliberate decisions. Retailers are reconsidering how their businesses operate and who they need to employ. And some of the most interesting developments are pulling in opposite directions.

We're automating product discovery while rediscovering the power of human selling. We're investing in customer experience while employees face greater pressure. We're celebrating sales growth that doesn't necessarily reflect stronger purchasing power.

None of these contradictions means the industry is heading in the wrong direction.

But they do suggest we need to ask better questions.

Not simply whether AI can automate something, but whether automation makes the experience better.

Not just how much a retailer can save by restructuring, but what capabilities might disappear in the process.

And not whether consumers are still spending, but what their choices tell us about the things they genuinely value.

Because for all the investment, innovation and transformation happening across retail, I keep coming back to something rather simple.

Technology changes how retail works. People still determine whether it works well.

And I suspect that will remain true for considerably longer than the current AI hype cycle.